Generating action sequences by denoising handles multimodal demos far better than regression; now a standard IL baseline.以去噪方式生成动作序列，比回归更能处理多模态演示，已成为模仿学习标准基线。
@misc{chi2023diffusion,
title = {{Diffusion Policy: Visuomotor Policy Learning via Action Diffusion}},
author = {Cheng Chi and Zhenjia Xu and Siyuan Feng and Eric Cousineau and Yilun Du and Benjamin Burchfiel and Russ Tedrake and Shuran Song},
year = {2023},
eprint = {2303.04137},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2303.04137},
}
GPU-parallel motion generation for arms — many trajectory optimizations run at once, giving collision-free, smooth plans in milliseconds.面向机械臂的 GPU 并行运动生成：同时运行大量轨迹优化，在毫秒级给出无碰撞且平滑的轨迹。
@misc{sundaralingam2023curobo,
title = {{cuRobo: Parallelized Collision-Free Minimum-Jerk Robot Motion Generation}},
author = {Balakumar Sundaralingam and Siva Kumar Sastry Hari and Adam Fishman and Caelan Garrett and Karl Van Wyk and Valts Blukis and Alexander Millane and Helen Oleynikova and Ankur Handa and Fabio Ramos and Nathan Ratliff and Dieter Fox},
year = {2023},
eprint = {2310.17274},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2310.17274},
}
NEO-VLA drops the visual encoder: images and language evolve in one backbone and a flow-matching action expert reads several depths. Layer probing shows action signals often peak before the final visual layer.NEO-VLA 去掉独立视觉编码器：图像与语言在同一主干中共同演化，流匹配动作专家从多个深度读取特征。逐层探测显示，对动作关键的信号往往在最终视觉层之前达到峰值。
@misc{fan2026from,
title = {{From Pixels to Actions - Towards Native Vision-Language-Action Policies}},
author = {Weichen Fan and Haiwen Diao and Penghao Wu and Siyuan Hu and Ziqi Huang and Tianqi Liu and Mingze Sun and Ziyi Wang and Zhongang Cai and Hanming Deng and Quan Wang and Lei Yang and Lewei Lu and Fangzhou Hong and Zhaoxi Chen and Dahua Lin and Ziwei Liu},
year = {2026},
eprint = {2611.06918},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2611.06918},
}
Learned from 16,151 hours of handheld gripper demos with no robot data; reports 78% average success over ten tasks versus 27% for the best other VLA, and releases the dataset, weights and post-training recipe.从 16,151 小时手持夹爪演示中学习，全程不采机器人数据；报告十项任务平均成功率 78%，其他 VLA 最好为 27%，并开放数据集、权重和后训练方案。
@misc{wu2026airoa,
title = {{AIRoA-0: Long-Horizon, Precise, Cross-Embodiment Manipulation from Handheld Data}},
author = {Yueh-Hua Wu and Kei Ota and Hanlong Li and Hayata Yuasa and Jason Wang and Kengo Ikeuchi and Kensuke Chiyoma and Makoto Sugiura and Masafumi Takahashi and Masatoshi Tateno and Naoaki Kanazawa and Nicola Scianca and Petr Khrapchenkov and Satoshi Otsubo and Takehiko Ohkawa and Takuya Okubo and Tatsuya Matsushima and Yuki Noguchi and Yuma Shin and Zixuan Zhong},
year = {2026},
eprint = {2611.03815},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03815},
}
Hardware, data and learning designed together: a low-cost wheeled humanoid with compliant grippers, teleoperation and robot-free human data collection, one whole-body controller and a shared policy pipeline.把硬件、数据和学习放在一起设计：低成本轮式人形、柔顺夹爪，遥操作与免机器人的人类数据采集，统一的全身控制器，以及人类与机器人示范共用的策略学习流程。
@misc{tao2026pixel,
title = {{Pixel: A Co-Designed Full-Stack Robot System for Learning General-Purpose Mobile Manipulation}},
author = {Tony Tao and Jason Jingzhou Liu and Ruslan Salakhutdinov and Deepak Pathak},
year = {2026},
eprint = {2611.03648},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03648},
}
@misc{chi2024universal,
title = {{Universal Manipulation Interface: In-The-Wild Robot Teaching Without In-The-Wild Robots}},
author = {Cheng Chi and Zhenjia Xu and Chuer Pan and Eric Cousineau and Benjamin Burchfiel and Siyuan Feng and Russ Tedrake and Shuran Song},
year = {2024},
eprint = {2402.10329},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2402.10329},
}
iDP3: whole-upper-body teleop, a 25-DoF humanoid with 3D LiDAR, and an improved 3D Diffusion Policy for noisy human data, aiming at skills beyond a single scene.iDP3：上半身全身遥操作、带三维激光雷达的 25 自由度人形，加上改进的三维扩散策略，使技能不再局限于单一场景。
@misc{ze2024generalizable,
title = {{Generalizable Humanoid Manipulation with 3D Diffusion Policies}},
author = {Yanjie Ze and Zixuan Chen and Wenhao Wang and Tianyi Chen and Xialin He and Ying Yuan and Xue Bin Peng and Jiajun Wu},
year = {2024},
eprint = {2410.10803},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2410.10803},
}
HIL-SERL — real-world RL with human corrections, reported to reach near-perfect success on precise assembly and dynamic tasks after a few hours of training.HIL-SERL：结合人类纠正的真机强化学习，报告称在精密装配与动态任务上经数小时训练即可接近满分成功率。
@misc{luo2024precise,
title = {{Precise and Dexterous Robotic Manipulation via Human-in-the-Loop Reinforcement Learning}},
author = {Jianlan Luo and Charles Xu and Jeffrey Wu and Sergey Levine},
year = {2024},
eprint = {2410.21845},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2410.21845},
}
PyTorch library with datasets, pretrained policies (ACT, Diffusion, SmolVLA) and drivers for low-cost arms.PyTorch 机器人学习库，提供数据集、预训练策略（ACT、Diffusion、SmolVLA）与低成本机械臂驱动。
@misc{zhao2023learning,
title = {{Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware}},
author = {Tony Z. Zhao and Vikash Kumar and Sergey Levine and Chelsea Finn},
year = {2023},
eprint = {2304.13705},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2304.13705},
}
Drives robots with a frozen frontier LLM plus in-context 'knowledge packages' on 12 sim benchmarks + 8 real robots; reports beating prior bests on 9 of 12, but weak real-robot results (competence, not yet intelligence).用冻结的前沿大模型加上下文“知识包”直接控制机器人，覆盖 12 个仿真基准和 8 台真机；报告在 12 个基准中的 9 个超过此前最好结果，但真机表现较弱——结论是具备了具身“能力”，尚未达到具身“智能”。
@misc{sensenova2026has,
title = {{Has GPT-6-Astra Achieved Embodied Intelligence?}},
author = {SenseNova and Joint Research Team and Zhongang Cai and Ziang Cao and Tailai Chen and Han Chong Chua and Linjun Dai and Hanming Deng and Hokin Deng and Haiwen Diao and Chenjing Ding and Weichen Fan and Xiao Fu and Zerui Gong and Chenyang Gu and Zexin He and Zhi Hou and Bo Li and Qinyue Li and Quanzhou Li and Xiaowan Li and Zhichao Liao and Dahua Lin and Zilong Lin and Bo Liu and Chunxiao Liu and Pengfei Liu and Tianqi Liu and Ziwei Liu and Chen Change Loy and Lewei Lu and Jiapeng Luo and Yuexiao Ma and Liang Pan and Xingang Pan and Oscar Qian and Ziyi Shang and Yuyao Shao and Dawei Su and Dacheng Tao and Qingyi Tao and Quan Wang and Ruisi Wang and Xiaogang Wang and Ziyi Wang and Beichen Wen and Donghao Wu and Jiaxu Wu and Silei Wu and Xiaoming Wu and others},
year = {2026},
eprint = {2610.22471},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.22471},
}
A generalist AI model commands the arm directly through a reference harness (AMP1), with no demos or built-in grasp skills; reports 85.6% zero-shot on RoboLab-120 vs 49.2% for the best VLA, and 104 of 108 real trials.让通用 AI 模型通过参考框架 AMP1 直接指挥机械臂，无需示教，也不预置抓取技能；在 RoboLab-120 上零样本成功率 85.6%（最佳 VLA 为 49.2%），真机 108 次试验成功 104 次。
@misc{chen2026agentic,
title = {{On Agentic Manipulation Policies}},
author = {Chen Chen and Omid Bagherieh and Naveen Nikamanth Appiah Balaji and Jesse T. Gonzalez and Negin Heravi and Fahad Islam and Xueting Li and Jean-Baptiste Regli and Kejia Ren and Petr Švec and David J. Yoon and Matthias Müller and Vladlen Koltun},
year = {2026},
eprint = {2611.09892},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09892},
}
A coding agent drives a bimanual robot through one command line; after each failure it reads the logs and writes or fixes small reusable tool programs.编程智能体通过一条命令行驱动双臂机器人；每次失败后读取日志，编写或修复可复用的小工具程序。
@misc{gao2026hexashell,
title = {{HexaShell: A Universal Harness for Robot Coding Agents in Simulation and the Real World}},
author = {Hongcheng Gao and Alex Zhou and Jay Zhu and Zelin Zheng and Shijia Ge and Xiping Li and Di Wu and Yazhe Wang and Jianshu Zeng and Xuan Shangguan and Yongli Wang and Yexing Wan and Lingyu He and Zhiqi Jia and Sihang Wu and Xiao He},
year = {2026},
eprint = {2610.23479},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.23479},
}
DISCO runs progressive-widening MCTS in GPU-parallel simulation, with a learned policy proposing continuous actions, then distills plans into a policy-value model used without search. Reports wins over BC, RL and search.DISCO 在 GPU 并行物理仿真中用渐进展宽 MCTS 搜索长程重排计划，由学习的策略提出连续动作、价值函数打分，并周期性把搜索结果蒸馏回策略-价值模型；部署时无需仿真搜索。报告在仿真与真机重排任务上优于行为克隆、强化学习与搜索基线（CoRL 2026 Oral）。
@misc{saha2026long,
title = {{Long-Horizon Multi-Object Rearrangement via Distilled Tree Search over Continuous Actions}},
author = {Kallol Saha and Yining Hong and Amber Li and Yejin Choi and Maxim Likhachev and David Held},
year = {2026},
eprint = {2610.25785},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25785},
}
CoRL 2026 Spotlight. Addresses impedance gain selection for articulated tasks such as doors, bolts and drawers, where neural policies may violate kinematic constraints.CoRL 2026 Spotlight。针对开门、拧螺栓、拉抽屉等铰接任务解决阻抗增益选择问题，这类任务中神经策略可能违反运动学约束。
@misc{park2026constraint,
title = {{Constraint-Decomposed Impedance Control from Generative Policy for Articulated Object Manipulation}},
author = {Kyungseo Park and Byeongdo Lim and Jungbin Lim and Minseok Choi and Frank C. Park},
year = {2026},
eprint = {2610.16316},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.16316},
}
NISE targets the sharp failure of 3D policies under realistic depth noise, carrying position uncertainty into attention; tested on transparent and reflective objects with commodity RGB-D.NISE 针对 3D 策略在真实深度噪声下的急剧失效，把位置不确定性带入注意力计算；在透明和反光物体上用普通 RGB-D 相机测试。
@misc{yang2026noise,
title = {{Noise-Invariant Spatial Encoding for 3D Manipulation Policies under Depth Corruption}},
author = {Zhiyuan Yang and Jiaju He and Ruiteng Zhao and Anqi Cheng and Haiyue Zhu and Kezhi Mao},
year = {2026},
eprint = {2610.24674},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.24674},
}
Learns to predict future hand-object states from human interaction datasets and uses the predicted object states as a general RL reward, reducing reward engineering; deployed on different hands (AAAI oral).从人手交互数据学习预测未来手物状态，并把预测的物体状态作为通用强化学习奖励，减少逐任务奖励设计；已部署到不同灵巧手（AAAI 口头报告）。
Deployed robots pool their runs; failures retrieve matching successful segments from the fleet buffer to fine-tune on, with no new demos. Reports 93–99% on five real tasks; 96 robots lift phone packing 79% → 98%.部署中的机器人共享执行经验：用失败片段在机群经验库中检索相似的成功片段来微调策略，无需新增示教或人工纠正。论文报告五个真机任务成功率 93–99%，96 台机器人把手机装盒任务从 79% 提升到 98%。
@misc{yu2026learning,
title = {{Learning from Fleet Experience: Collective Experience Learning Across 96 Robots}},
author = {Chao Yu and Shu'ang Yu and Hao Lin and Hongzhi Zang and Yuxiang Yan and Chenxi Jiang and Zhen Guo and Chuankang Li and Yihao Chen and Xin Xu and Xu Fu and Jiakai Zhou and Liangzhi Shi and Hongyi Ren and Yiwei Liu and Jing Wu and Yuanqing Wang and Chen Wu and Bo Dai and Chunyu Yang and Hongliang Huang and Shuoyu Yue and Weiqi Yin and Yinuo Chen and Ruihua Chen and Chunyang Zhu and Han Xiao and Yanghao Li and Ruipu Wu and Zhikai Cai and Xiu Chen and Xu Tan and Si Liu and Yuchen Yan and Jialin Zhang and Wen Zeng and Yiming Zhou and Zixiao Huang and Xinlei Chen and Guanghui Ren and Quanlu Zhang and Boxun Li and Maoqing Yao and Yu Wang},
year = {2026},
eprint = {2610.25736},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25736},
}
CoRL 2026 Spotlight. Simulation benchmark for wires and cables in industrial assembly, aimed at the gap left by benchmarks tied to specific hardware or lacking modular, customizable task assets.CoRL 2026 Spotlight。面向工业装配中线缆等可变形线状物体（DLO）的仿真基准，针对现有基准绑定特定硬件、缺少模块化可定制任务资产等问题。
@misc{zhu2026wirecraft,
title = {{WireCraft: A Simulation Benchmark for Industrial DLO Manipulation}},
author = {Chongyu Zhu and Ramy ElMallah and Hyegang Kim and Zachary Tang and Jiachen Rao and Artem Arutyunov and Seungyeon Ha and Chi-Guhn Lee},
year = {2026},
eprint = {2606.18097},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.18097},
}
Simulation pipeline that writes and checks task code with multimodal LLMs to mass-produce bimanual demos under heavy randomization (clutter, lighting, backgrounds, instructions); also a 50-task dual-arm benchmark.用多模态大模型自动生成并校验任务代码，在强域随机化（杂物、光照、背景、指令）下批量产出双臂演示数据的仿真管线，同时提供 50 个双臂任务的基准。
@misc{chen2025robotwin,
title = {{RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation}},
author = {Tianxing Chen and Zanxin Chen and Baijun Chen and Zijian Cai and Yibin Liu and Zixuan Li and Qiwei Liang and Xianliang Lin and Yiheng Ge and Zhenyu Gu and Weiliang Deng and Yubin Guo and Tian Nian and Xuanbing Xie and Qiangyu Chen and Kailun Su and Tianling Xu and Guodong Liu and Mengkang Hu and Huan-ang Gao and Kaixuan Wang and Zhixuan Liang and Yusen Qin and Xiaokang Yang and Ping Luo and Yao Mu},
year = {2025},
eprint = {2506.18088},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2506.18088},
}
Puts ALOHA on a mobile base; co-training with static ALOHA data boosts mobile tasks like cooking and opening cabinets.将 ALOHA 装上移动底盘；与静态 ALOHA 数据联合训练显著提升做饭、开柜等移动任务表现。
@misc{fu2024mobile,
title = {{Mobile ALOHA: Learning Bimanual Mobile Manipulation with Low-Cost Whole-Body Teleoperation}},
author = {Zipeng Fu and Tony Z. Zhao and Chelsea Finn},
year = {2024},
eprint = {2401.02117},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2401.02117},
}
Conditions a diffusion policy on a compact point-cloud encoding; more data-efficient and robust to viewpoint than images.用紧凑点云编码作为扩散策略条件，比图像输入更省数据、对视角更鲁棒。
@misc{ze2024d,
title = {{3D Diffusion Policy: Generalizable Visuomotor Policy Learning via Simple 3D Representations}},
author = {Yanjie Ze and Gu Zhang and Kangning Zhang and Chenyuan Hu and Muhan Wang and Huazhe Xu},
year = {2024},
eprint = {2403.03954},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2403.03954},
}
Upgraded ALOHA bimanual rig (stronger grippers, gravity compensation, sturdier frame) with open hardware designs and a MuJoCo model for simulation.ALOHA 双臂平台的升级版（更强夹爪、重力补偿、更稳固的框架），开源硬件设计并提供 MuJoCo 仿真模型。
@misc{team2024aloha,
title = {{ALOHA 2: An Enhanced Low-Cost Hardware for Bimanual Teleoperation}},
author = {ALOHA 2 Team and Jorge Aldaco and Travis Armstrong and Robert Baruch and Jeff Bingham and Sanky Chan and Kenneth Draper and Debidatta Dwibedi and Chelsea Finn and Pete Florence and Spencer Goodrich and Wayne Gramlich and Torr Hage and Alexander Herzog and Jonathan Hoech and Thinh Nguyen and Ian Storz and Baruch Tabanpour and Leila Takayama and Jonathan Tompson and Ayzaan Wahid and Ted Wahrburg and Sichun Xu and Sergey Yaroshenko and Kevin Zakka and Tony Z. Zhao},
year = {2024},
eprint = {2405.02292},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2405.02292},
}
6D pose estimation and tracking of unseen objects from a CAD model or a few reference images, with no per-object training.只需 CAD 模型或少量参考图像即可对未见过的物体做 6D 位姿估计与跟踪，无需逐物体训练。
@misc{wen2023foundationpose,
title = {{FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects}},
author = {Bowen Wen and Wei Yang and Jan Kautz and Stan Birchfield},
year = {2023},
eprint = {2312.08344},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2312.08344},
}
Tracks the 6-DoF pose of a never-seen object in RGB-D video while building its neural 3D model online, with no CAD model needed; useful for manipulating novel objects.在 RGB-D 视频中跟踪从未见过物体的六自由度位姿，同时在线构建其神经三维模型，无需 CAD 模型；适用于操作新物体。
@misc{wen2023bundlesdf,
title = {{BundleSDF: Neural 6-DoF Tracking and 3D Reconstruction of Unknown Objects}},
author = {Bowen Wen and Jonathan Tremblay and Valts Blukis and Stephen Tyree and Thomas Muller and Alex Evans and Dieter Fox and Jan Kautz and Stan Birchfield},
year = {2023},
eprint = {2303.14158},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2303.14158},
}
AnyGrasp — dense 7-DoF parallel-gripper grasp detection trained on real sensor data with analytic labels and center-of-mass awareness, plus grasp tracking across frames for moving objects; released as an SDK.AnyGrasp：基于真实传感器数据与解析标注训练、考虑物体质心的稠密 7 自由度平行夹爪抓取检测，并能跨帧跟踪抓取以应对运动物体；以 SDK 形式发布。
@misc{fang2022anygrasp,
title = {{AnyGrasp: Robust and Efficient Grasp Perception in Spatial and Temporal Domains}},
author = {Hao-Shu Fang and Chenxi Wang and Hongjie Fang and Minghao Gou and Jirong Liu and Hengxu Yan and Wenhai Liu and Yichen Xie and Cewu Lu},
year = {2022},
eprint = {2212.08333},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2212.08333},
}
Ruckig — open-source C++/Python library for time-optimal, jerk-limited online trajectories to any target position, velocity and acceleration; a common drop-in for smooth arm motion on real controllers.Ruckig：开源 C++/Python 库，可实时生成到任意目标位置、速度和加速度的时间最优、限加加速度（jerk）轨迹；常被直接用于真实控制器上的机械臂平滑运动。
robomimic — a controlled study of imitation and offline RL on human-collected manipulation demos (demo quality, observation choices, history-aware BC), released with the framework and datasets.robomimic：在人类采集的操作示教数据上系统比较模仿学习与离线强化学习（示教质量、观测选择、带历史的行为克隆），并开源框架与数据集。
@misc{mandlekar2021what,
title = {{What Matters in Learning from Offline Human Demonstrations for Robot Manipulation}},
author = {Ajay Mandlekar and Danfei Xu and Josiah Wong and Soroush Nasiriany and Chen Wang and Rohun Kulkarni and Li Fei-Fei and Silvio Savarese and Yuke Zhu and Roberto Martín-Martín},
year = {2021},
eprint = {2108.03298},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2108.03298},
}
Wraps a frozen VLA in two loops — one keeps user preferences and scene memory for planning, one blocks hazardous actions and calls recovery skills; reports violations down from 52.6% to 6.1%, recovery up from 13% to 80%.给冻结的 VLA 套上双环框架：规划环保存用户偏好与场景记忆以做个性化任务分解，执行环拦截危险接触并调用可复用的恢复技能；报告安全违规率从 52.6% 降到 6.1%，失败恢复成功率从 13% 提升到 80%。
@misc{wang2026carryon,
title = {{CarryOn: A Dual-Loop Harness for Personalized and Reliable Robot Assistance}},
author = {Zixuan Wang and Yuqi Liu and Yuxin Chen and Sengqiao Yang and Jinhui Ye and Pengguang Chen and Changsheng Lu and Shu Liu and Bei Yu and Jiaya Jia},
year = {2026},
eprint = {2610.21453},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.21453},
}
Six frontier general-purpose models drive a LIBERO-10 arm through tool calls (end-effector poses, gripper), no robot training; best reported success is 44%. More reasoning effort usually helps but costs more.让六个前沿通用大模型通过工具调用（指定末端位姿与夹爪开合）直接控制 LIBERO-10 机械臂，无任何机器人专门训练；报告的最高成功率为 44%。提高推理强度通常有帮助，但推理成本更高。
Harness where a frontier LLM reviews a frozen π0.5 policy less often as experience builds up and focuses on recovery after errors; reports ~57% fewer reviewer tokens and higher success on four sim tasks.让前沿大模型监督冻结的 π0.5 策略的执行框架：随着经验积累逐步减少审查频率，并把监督集中在出错后的恢复上；报告在四个仿真任务上审查 token 平均减少约 57%，成功率也更高。
@misc{lin2026fastharness,
title = {{FastHarness: Experience-Driven Review Scheduling for Vision-Language-Action Policies}},
author = {Tao Lin and Yuxin Du and Haoxiao Wang and Zile Zhong and Bo Zhao},
year = {2026},
eprint = {2610.31427},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.31427},
}
Between direct action prediction and full video prediction: VLDA learns a compact latent 'Delta' of task-relevant state change and uses it to seed action denoising. Built on Baidu's open LoongForge.介于直接预测动作与完整预测未来画面之间：VLDA 学习一个紧凑的隐变量 Delta 表示与任务相关的状态变化，并用它引导动作去噪；实现于百度开源训练框架 LoongForge。
@misc{zhang2026vision,
title = {{Vision-Language-Delta-Action: Learning Compact Deltas as Actionable Transition Interfaces for Robotic Manipulation}},
author = {Lingsen Zhang and Wei Li and Jie He and Henghua Zhang and Kai Mo and Shiyong Li and Yanpeng Wang and Rui Shao},
year = {2026},
eprint = {2610.25874},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25874},
}
Deployment knob for flow-matching VLAs: fewer solver steps hurt under fast replanning but help with long open-loop chunks. Tuning both together, training-free, reports higher success and lower latency for π0/π0.5/GR00T.流匹配 VLA 的部署调参发现：求解步数少在频繁重规划时会降低精度，但在长时开环执行动作块时反而有利。报告在 π0、π0.5、GR00T 上联合调整两者，无需训练即可提高成功率并降低推理延迟。
@misc{yang2026coupling,
title = {{Coupling Execution Horizon and Sampling Depth in Flow-Matching VLA Policies}},
author = {Jiabing Yang and Yuan Xu and Yixiang Chen and Qisen Ma and Peiyan Li and Kai Wang and Chenhang Cui and Tao Yu and Zichen Wen and Wen Ye and Bowen Fang and Ziheng He and Xiangnan Wu and Yingda Li and Yan Huang and Liang Wang},
year = {2026},
eprint = {2610.32102},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.32102},
}
Report: GPT-6-Astra drives a robot on EBench via waypoints or by calling a VLA/WAM policy. Direct control leads on short tasks, trails on long-horizon and contact-rich ones; calling a policy cuts token cost.评测报告：在 EBench 上让 GPT-6-Astra 直接输出路径点控制机器人，或通过执行框架调用 VLA/WAM 策略。直接控制在短时任务上领先，但在长时程与接触密集任务上落后；调用策略可显著减少 token 开销。
@misc{yao2026profiling,
title = {{Profiling a Frontier Agent with EBench: From Direct Control to Harness with Policy}},
author = {Yifei Yao and Yude Zou and Guanbo Jiao and Xing Gao and Ning Gao and Yukai Wang and Jiantong Chen and Zihou Zhu and Xinyu Li and Ningyu Yan and Shuai Wang and Jianming Ma and Haoyu Guo and Weinan Zhang and Chunhua Shen},
year = {2026},
eprint = {2610.35832},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.35832},
}
Mixed-precision quantization for VLAs that sets each weight matrix's bit width by how much it affects actions; reports 92.7% LIBERO success at 3 bits (above AWQ and GPTQ) and 70.5% at 2.5 bits.面向 VLA 的混合精度量化：根据每个权重矩阵对输出动作的影响分配比特数；报告在 3 比特下 LIBERO 成功率 92.7%（高于 AWQ 与 GPTQ），2.5 比特下仍有 70.5%。
@misc{qiu2026agq,
title = {{AGQ: Action-Guided Ultra-Low-Bit Quantization for Vision-Language-Action Models}},
author = {Xuhong Qiu and Zhuoyu Wu and Qiawei Zheng and Wenhui Ou and Wenqi Fang and Raphaël C. -W. Phan},
year = {2026},
eprint = {2610.33074},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2610.33074},
}
Pretrains a 4B VLA on 3D-grounded latent actions mined from 30K hours of human, robot and sim video, then adapts to real robot actions. Reports 98.8% on LIBERO and 88.6% on RoboTwin 2.0.从 3 万小时人类、机器人与仿真视频中学习带 3D 约束的连续潜在动作（RynnLAM），用于预训练 4B 参数 VLA，再适配到具体机器人动作空间。报告 LIBERO 98.8%、RoboTwin 2.0 平均 88.6%。
@misc{zhao2026rynnvla,
title = {{RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions}},
author = {Haoyu Zhao and Xingyue Zhao and Zhian Su and Bohan Hou and Yinzhou Tang and Fengyi Wang and Kehan Li and Yuming Jiang and Siteng Huang and Xin Li and Zhenyu Zeng and Mingxiu Chen and Deli Zhao and Zhongyu Li},
year = {2026},
eprint = {2610.34356},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.34356},
}
Teaches a VLA to know when it is done or failing: stop and failure signals are two extra action dimensions, denoised faster than the actions so they can guide them; no frame-level progress labels. Real-robot tests.让 VLA 自己判断任务完成或失败：把停止与失败信号作为动作块的两个额外维度，与动作一起去噪且去噪更快，从而引导后续动作生成；无需逐帧进度标注，在真机操作实验中验证。
@misc{liu2026echo,
title = {{Echo: Coupled Action and Task Status Signal Denoising for Vision Language Action Models}},
author = {Yufeng Liu and Bohao Fan and Juntu Zhao and Xinyi Wang and Hang Yu and Wenzhuan Luo and Yingzhi Wang and Qinghai Li and Yingdong Hu and Junliang Guo and Richard Wang and Dequan Wang and Yang Gao and Cheng Ma},
year = {2026},
eprint = {2610.31569},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.31569},
}
Lets a VLA pick, at each step, between fine and temporally coarser action decoders: redundant motion runs faster while precision phases stay fine-grained. Tested on 3 sim benchmarks and 6 real tasks.让 VLA 在每一步自行选择细粒度或时间上更粗的动作解码器：冗余动作阶段加速执行，精细接触阶段保持细粒度控制；在 3 个仿真基准和 6 个真实任务上验证效率提升且成功率不降。
@misc{jeon2026pacevla,
title = {{PaceVLA: Adaptive Temporal Action Quantization in Vision-Language-Action Models}},
author = {Hojin Jeon and Jimin Lee and Taekwan Kim and Nayoung Oh and Taeyoung Kim and Myungkyu Koo and Jinmo Koo and Joonwoo Ahn and Jinwoo Shin},
year = {2026},
eprint = {2610.25961},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25961},
}
Test-time add-on for frozen VLAs: a GNN retrieves future motion primitives from a graph of training demos and gates them into the actions. Reports up to +12% in sim and +40% real (GR00T N1.7, π0.5, StableVLA).面向冻结 VLA 的测试时插件：把训练示范组织成运动基元图，用 GNN 定位当前执行并检索后续基元，再按维度门控融合到预测动作中；报告在 GR00T N1.7、π0.5 等上仿真最高 +12%、真机 +40%。
@misc{quang2026retrieve,
title = {{Retrieve to Act: Motion Primitive Graph Retrieval for Robust VLA Execution}},
author = {Pham Tri Quang and Bao-Ngoc Dao and Minh Duc Nguyen and Duc-Duy Nguyen and Hung T. Ho and Linh Dang Le and Trong-Bao Ho and Thien-Loc Ha and Tan Q. Nguyen and Tuan Dam and Vu N. Duong and Daniel Sonntag and Ngan Le and Khoa D. Doan and Jan Peters and An Thai Le and Duy M. H. Nguyen and Vien Anh Ngo},
year = {2026},
eprint = {2610.37215},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.37215},
}
Finds VLAs over-rely on the wrist camera; predicts future interaction intent from past actions to route between global and wrist views per stage. Reports +10.1 pp over π0.5 on RoboTwin 2.0 and 71.7% → 85.0% real.发现现有 VLA 过度依赖腕部相机；根据历史动作预测未来交互意图，按任务阶段在全局与腕部视角间动态路由；报告在 RoboTwin 2.0 上比 π0.5 高 10.1 个百分点，真机成功率由 71.7% 提升到 85.0%。
PAL has a high-level VLM write future-relevant subtasks into an executable memory while watching, then hand the current one to a low-level VLA. Reports 56.05% vs 44.51% on RoboMME and 60.83% on 4 real tasks.PAL 让高层 VLM 在观察过程中把未来需要的子任务提前写入可执行记忆（subtask_list），执行时再挑出当前子任务交给底层 VLA；报告在 RoboMME 上成功率 56.05%（最强基线 44.51%），4 个真实任务总体 60.83%。
@misc{liu2026plan,
title = {{Plan First, Act Later: Streaming Reasoning for Executable Memory in Robot Control}},
author = {Qinghang Liu and Conglin Wang and Zongsheng Li and Boyong He and Han Gao and Pengcheng Wang and Tao Sun and Lige Liu and Chang Liu},
year = {2026},
eprint = {2610.29329},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29329},
}
Fourth-place MetaFine Challenge entry: a task router picks between adapted DINOv2-DiT and π0.5 policies behind one interface, a practical recipe for adapting pretrained policies to precise tasks.MetaFine 挑战赛第四名方案：任务路由在统一接口下选择微调后的 DINOv2-DiT 或 π0.5 策略，是把预训练策略适配到精细任务的实用做法。
@misc{zhang2026finecraft,
title = {{FineCraft: Enhancing VLA Models for Fine-Grained Manipulation}},
author = {Shuaike Zhang and Ziheng Quan and Junhao Ran and Muhe Ding and Shaokun Wang and Jianlong Wu},
year = {2026},
eprint = {2611.03163},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03163},
}
Coding agents keep real-robot experience as memory and reusable code skills, refining them through physical trials; reports 91.1% success on eight real tasks and 84.4% on 42 RoboDojo tasks.让编码智能体把真机交互经验存为记忆和可复用的代码技能，并在实物试验中持续改进；报告八个真实任务成功率 91.1%，RoboDojo 的 42 个任务成功率 84.4%。
@misc{wang2026pyrua,
title = {{PyRUA-REAL: Robot Exploration and Adaptive Learning through RSI in the Real World}},
author = {Qiang Wang and Jianxin Bi and Shulong Jiang and Ruiyang Si and Wenbo Huang and Kaiqi Chen and Jiaxin Li and Liyun Li and Xuanyi Li and Daquan Zhou},
year = {2026},
eprint = {2611.06768},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.06768},
}
Memory for VLAs that keeps only sparse keyframes the policy itself predicts will matter later. Adds the RoboTwin-MeM benchmark; reports +40% average success over memory-augmented VLAs across 17 sim and 4 real tasks.为 VLA 加入稀疏记忆：由策略自己预测哪些关键帧之后会有用并只保存这些帧。同时提出 RoboTwin-MeM 基准；报告在 17 个仿真和 4 个真机任务上平均成功率比带记忆的 VLA 高 40%。
@misc{yang2026eventvla,
title = {{EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies}},
author = {Ganlin Yang and Zhangzheng Tu and Yuqiang Yang and Sitong Mao and Junyi Dong and Tianxing Chen and Jiaqi Peng and Jing Xiong and Jiafei Cao and Jifeng Dai and Wengang Zhou and Yao Mu and Tai Wang},
year = {2026},
eprint = {2606.20092},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2606.20092},
}
PAINT keeps action chunks consistent under async execution by picking the initial noise (backward Euler inversion), not by guiding the flow. No training; reports RTC-level task success with better prefix consistency.PAINT 通过选择初始噪声（后向欧拉反演）而非引导流场，让异步执行时的动作块首尾一致；无需训练，报告任务成功率与 RTC 相当、前缀一致性更好。
@misc{ho2026start,
title = {{Start Right, Arrive Right: Asynchronous Execution via Initial Noise Selection}},
author = {Trong-Bao Ho and Quang-Tan Nguyen and Thien-Loc Ha and Gia-Binh Nguyen and Viet-Thanh Nguyen and Long Dinh and Minh N. Vu and Duy M. H. Nguyen and An Thai Le and Ngo Anh Vien},
year = {2026},
eprint = {2606.19774},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.19774},
}
A frozen frontier VLM, prompted with the whole episode, outputs waypoints directly and reports 92.1% on RoboMME memory tasks with no task training.冻结的前沿 VLM 以整段回合为上下文直接输出路径点，在 RoboMME 记忆类任务上报告 92.1% 成功率，且无需任务训练。
@misc{jedlicki2026can,
title = {{Can a Frontier VLM Serve as a Robot Policy for Memory-Dependent Manipulation?}},
author = {Jack B. Jedlicki and Amish Sethi and Heng Yang},
year = {2026},
eprint = {2610.17350},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17350},
}
Asks whether joint readings alone suffice for millimetre-precise insertion — trains sim-to-real policies with no camera, force or tactile input.探究仅靠关节读数能否完成毫米级精密插装：训练不用相机、力或触觉输入的 sim-to-real 策略。
WACL learns from human demos to push and hold with the whole arm, choosing where on the arm to make contact while keeping the gripper's constraints.WACL 从人类示范中学习用整条手臂推、抵物体，自主选择手臂上的接触位置，同时满足末端执行器的约束。
@misc{huang2026beyond,
title = {{Beyond the End-Effector: Learning Whole-Arm Contact Manipulation from Human Demonstrations}},
author = {Bo-Ruei Huang and Daniel Seita and Erdem Bıyık},
year = {2026},
eprint = {2610.17802},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17802},
}
Grasps several objects at once with a dexterous hand by learning one-step residual corrections to a reference trajectory, with a strategy that adapts to how the objects are laid out.让灵巧手一次抓取多个物体：学习对参考轨迹的单步残差修正，并根据物体的空间布局调整抓取策略。
CTR retimes the two arms in bimanual demos at training time while keeping task-required constraints, so the policy stops using the other arm's progress as a shortcut cue.CTR 在训练时对双臂示教中两只手臂的动作时序重新组合，同时保留任务必需的约束，避免策略把另一只手臂的进度当作捷径线索。
@misc{hu2026learn,
title = {{Learn the Task, Not the Schedule: Reducing Spurious Cross-Arm Coupling in Bimanual Imitation Learning}},
author = {Shuyuan Hu and Deng-Ke Wu and Yifan Jin and Qianhang Luo and Mingchen Ju and Kai Ye and Bo Zhao},
year = {2026},
eprint = {2610.26507},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.26507},
}
One two-hand policy (PPO + BC on its own verified successes) tracks retargeted human hand-object motions; ships BiHOI (1,575 sequences, 229 objects). Reports 86.8% on unseen trajectories, 57.3% on new objects.用一个双手跟踪策略（PPO 结合对自身成功执行的行为克隆）复现重定向后的人手-物体交互动作，并发布 BiHOI 数据集（1575 段双手序列、229 个物体）。论文报告在未见轨迹上成功率 86.8%、新物体上 57.3%。
@misc{sun2026donic,
title = {{DONIC: Generalist Dexterous Bimanual Tracking Control from Human Demonstrations}},
author = {Mingze Sun and Ziang Cao and Yuan Fei and Qiankuan Tao and Zihan Wang and Tianqi Liu and Junhao Chen and Haiwen Diao and Fangzhou Hong and Zhaoxi Chen and Ziwei Liu},
year = {2026},
eprint = {2610.29997},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29997},
}
SERA turns the high-level VLM into an agent with an editable harness that tracks task progress for a fixed VLA executor, and revises the harness from rollouts; reports 51.4% vs 44.4% overall success on RoboMME.SERA 把高层 VLM 规划器做成带可编辑执行框架的智能体，为固定的底层 VLA 记录任务进度与执行反馈，并利用环境回放自动改进框架；报告在 RoboMME 上总体成功率 51.4%，未自我演化时为 44.4%。
@misc{du2026from,
title = {{From VLM Planners to Robotic Agents: Self-Evolving Harnesses for History-Dependent Manipulation}},
author = {Tingwen Du and Haoyu Guo and Hanxue Zhang and Xuanxu Wang and Junting Dong and Chao Gao and Mingfei Han and Tao Lu and Mulin Yu and Chunhua Shen and Xiaojun Chang and Zhihui Li},
year = {2026},
eprint = {2610.29560},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29560},
}
A coding agent that reads wrist force/torque as well as images, writes probing and force-feedback programs, and revises them from force logs; no demos or training. Beats its vision-only variant on 5 real tasks.同时读取腕部力/力矩与图像的编程智能体：编写接触探测与力反馈控制程序，并在每次执行后依据力记录诊断失败、修改代码；无需示教或训练。在 5 个真实任务上优于仅视觉版本。
@misc{tan2026forca,
title = {{FORCA: A Force-Aware Coding Agent for Contact-Rich Robot Manipulation}},
author = {Xiao Tan and Yixuan Wang and Zhexi Wang and Wenzhen Yuan},
year = {2026},
eprint = {2610.31841},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.31841},
}
36 contact-rich tasks built around eight everyday skills, with 3D-printable objects that each have a matching sim model. Reports that tested visuo-tactile policies did not clearly beat vision-only ones.围绕八种日常接触技能设计的 36 个接触密集型任务，40 多个物体可用消费级 3D 打印机复现，且每个都有对应仿真模型。报告称所测的视触觉策略并未明显优于纯视觉策略。
Mobile dexterous grasping learned in sim: a whole-body controller does reaching and active vision, local RL experts the object, all distilled into one point-cloud policy; reports zero-shot real transfer.在仿真中学习移动灵巧抓取：全身控制器负责底盘-手臂协同到达与主动视觉，局部强化学习专家负责物体交互，二者蒸馏成一个基于点云的视觉策略；报告可零样本迁移到真实世界的数百个物体与场景。
@misc{yang2026cowboy,
title = {{COWBOY: A Scalable Sim-to-Real Framework for Learning Contextual Whole-Body Manipulation}},
author = {Jiahui Yang and Jason Jingzhou Liu and Ray Muxin Liu and Andrew E. Wang and Kenneth Shaw and Deepak Pathak and Ruslan Salakhutdinov},
year = {2026},
eprint = {2610.28948},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.28948},
}
ARPG pretrains an action encoder on human manipulation data, then trains a 130M generative policy in its space. Reports bimanual dexterous success up from 10.1% (π0.5) to 44.7% on DexJoCo.ARPG 把动作表征学习与动作生成解耦：先在大规模人类操作数据上预训练动作编码器，再在其表征空间训练 1.3 亿参数的生成策略。报告在 DexJoCo 双手灵巧任务上平均成功率从 π0.5 的 10.1% 提升到 44.7%。
@misc{xu2026scaling,
title = {{Scaling Dexterous Manipulation via Action Representation Pretraining}},
author = {Yanjinghao Xu and Jiexin Zhang and Huan Zhao and Jianjia Mo and Zhenghao Yin and Dingkang Liang and Xiang Bai and Han Ding},
year = {2026},
eprint = {2610.34453},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.34453},
}
Language-to-dexterous-manipulation as modulated tracking: a distilled generalist tracker plus an RL-trained modulator that keeps adjusting the reference, object latent and position target; zero-shot sim-to-real.把语言条件灵巧操作变成“可调制的跟踪”：蒸馏得到的通用跟踪器加 RL 训练的调制器，在执行中持续修正参考轨迹、物体隐表征与位置目标，抑制误差累积；零样本迁移到真机。
@misc{adalibieke2026adadextrack,
title = {{AdaDexTrack: Dynamic Modulation for Adaptive and Generalizable Dexterous Manipulation Tracking}},
author = {Jianibieke Adalibieke and Qianwei Han and Xueyi Liu and Yuzhe Qin and Li Yi},
year = {2026},
eprint = {2610.33852},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.33852},
}
Foresight without predicting future frames: a training-only head forecasts the next action chunk and backpropagates through the policy's latent. No inference cost; reports up to +14 pts on mm-precision real tasks.不预测未来图像也能获得“预见性”：仅在训练时用级联头预测下一个动作块，并把损失回传到在线策略的隐状态，推理零额外开销；报告在耳机装配、穿鞋带等毫米级真机任务上最高提升 14 个百分点。
@misc{yang2026adapolicy,
title = {{AdaPolicy: Action-Driven Anticipatory Robot Policies via Next-Chunk Prediction}},
author = {Fan Yang and Yuting Su and Chen Zhao and Fugui Fan and Peiguang Jing},
year = {2026},
eprint = {2610.34280},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.34280},
}
@misc{strong2026real,
title = {{Real2Sem2Real: 3D Semantic Augmentations for Robot Learning from a Single Human RGB Demonstration}},
author = {Matthew Strong and Ankush K. Dhawan and Julien Poffet and Giuse Pham and Jinho So and Monroe Kennedy},
year = {2026},
eprint = {2610.28880},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.28880},
}
Sim-only RL recipe for in-hand object pose reaching on a LEAP hand, zero-shot to hardware; reports 95.2% of 4,800 unseen goals reached and a tunable residual safety filter that cuts drops by 75%.在仿真中无需人类数据训练 LEAP 灵巧手的手内物体位姿到达策略，零样本迁移到真机；报告在 4800 个未见目标上 95.2% 到达，可调节的残差安全滤波器使掉落减少 75%。
@misc{chuang2026dexterous,
title = {{Dexterous In-Hand Object Pose Reaching with Learned Safety Filters}},
author = {Ian Chuang and Hiya Shah and Vamsi Chilamkur and Iman Soltani and Claire Tomlin},
year = {2026},
eprint = {2610.37652},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.37652},
}
Retargets human hand demos (video, mocap, tactile gloves) by keeping contact-force targets, not just kinematics; a tactile tracking controller replays them. Reports contact kept on 43–78% of frames vs <10%.把人手示教（视频、动捕、触觉手套）重定向到灵巧手时显式保留接触力目标而不只是运动学，并用触觉跟踪控制器回放；报告 43–78% 的接触帧保持接触，纯运动学方法不到 10%。
@misc{dai2026card,
title = {{CARD: Contact-Anchored Retargeting for Dexterous Manipulation}},
author = {Cunxi Dai and Botao He and Francois R Hogan and Wei Pan and Rao Fu and Wen-Han Hsieh and Ruoshi Liu and Jitendra Malik and Guanya Shi and Akash Sharma},
year = {2026},
eprint = {2610.36539},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36539},
}
Learns history-aware progress rewards for long-horizon manipulation so missing prerequisites are not rewarded, then refines policies from BC to online RL. Reports 31.8% on 18 held-out RLBench tasks (+9.2 pts).为长时程操作学习依赖执行历史的进度奖励，避免跳过前置步骤也得高分，再从行为克隆过渡到在线强化学习微调策略；报告在 18 个未参与预训练的 RLBench 任务上平均 31.8%（高出最强基线 9.2 个百分点），真机 57.5%。
@misc{chen2026robolongr,
title = {{RoboLongR: Causally Consistent Rewards Teach Robots to Progressively Master Long-Horizon Manipulation}},
author = {Zixuan Chen and Lin Cao and Fenggang Zhang and Haotong Zhang and Yangtao Chen and Jieqi Shi and Jing Huo and Yang Gao},
year = {2026},
eprint = {2610.33863},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.33863},
}
Builds pseudo-bimanual demos from single-arm data by sagittal-plane mirroring plus LLM-written coordination constraints. On AgileX ALOHA it reports 2.11× cups collected vs. a single-arm baseline.将单臂示范沿矢状面镜像（观测、本体感觉与动作），再由 LLM 生成时空协调约束，构造伪双臂示范；在 AgileX ALOHA（RoboTwin 仿真与真机）上报告收集杯子数达到单臂基线的 2.11 倍。
@misc{wu2026robostd,
title = {{RoboSTD: Zero-Shot Single-to-Dual Transfer via Sagittal Mirroring for Bimanual Learning}},
author = {Hongyu Wu and Chong Liu and Shaoxuan Xie and Shaqi Luo and Hanyu Feng and Shihan Wu and Guocai Yao},
year = {2026},
eprint = {2610.36685},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36685},
}
@misc{huang2026cuarms,
title = {{cuArms: GPU-Accelerated Conflict-Based Planning for Multi-Arm Manipulation}},
author = {Philip Huang and Fabio Ramos and Jiaoyang Li and Caelan Garrett and Lars Johannsmeier},
year = {2026},
eprint = {2610.36250},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36250},
}
SpatialDex pre-trains a VLM-based dexterous policy on scene geometry and hand motion reconstructed from 1M RGB-only human clips. Reports +38.3 pts over prior methods on 6 real bimanual tasks.SpatialDex 从 100 万段纯 RGB 人类视频中重建场景几何与手部运动，并以几何作为输入和预测目标预训练基于 VLM 的灵巧策略；报告在 9 个仿真任务上比现有方法高 16.5 分、6 个真实双臂灵巧任务上高 38.3 分。
@misc{he2026reconstruction,
title = {{Reconstruction-Augmented Visuomotor Pre-Training of 3D-Grounded Bimanual Dexterity from RGB-Only Human Videos}},
author = {Xingyi He and Zhiwen Qiu and Adhitya Polavaram and Yunhao Cao and Cory Fan and Julius Samwer and Saleh Kalantari and Kuan Fang},
year = {2026},
eprint = {2610.29716},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29716},
}
MimicGen-style demo synthesis that respects multi-arm constraints: plans and schedules skills across arms. Reports 79.6% generation success on 28 sim variants and 50-80% real success via zero-shot sim-to-real.面向双臂/多臂机器人的示范自动生成：在少量源示范基础上联合做规划与技能调度，显式处理多臂间约束，避免现有生成方法产生的碰撞与失败。28 个仿真任务变体上数据生成成功率 79.6%，训练出的策略经零样本 sim-to-real 在 8 个真实变体上成功率 50–80%。
@misc{jawale2026schedulegen,
title = {{ScheduleGen: Demonstration Generation via Skill Scheduling for Multi-Arm Imitation Learning}},
author = {Neel Jawale and Stephan Pleines and Soha Pouya and Yuke Zhu and Ajay Mandlekar and Caelan Garrett},
year = {2026},
eprint = {2610.39429},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.39429},
}
Robot-free dexterous demos with shared contact geometry, fingertip RGB and contact-first retargeting. Reports at least 50 points higher final-stage success than baselines on LEAP and Wuji hands; hardware is open.免机器人灵巧演示框架：共享接触几何、指尖 RGB 视角、接触对齐优先的重定向，加上预测接触面轨迹的层次策略。报告在 LEAP 和 Wuji 手的六组手-任务上，最终阶段成功率比基线至少高 50 个百分点；硬件与软件公开。
@misc{xu2026contact,
title = {{Contact Imitation Interface: Transferring Dexterous Skills by Showing How Contact Unfolds}},
author = {Zhixuan Xu and Yichen Li and Bingchen Wu and Rui Wang and Yiwen Hou and Zixuan Liu and Tianyu Qiu and Lin Shao},
year = {2026},
eprint = {2610.41028},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.41028},
}
Multi-view policy compressing camera views into a compact token for a flow-matching action head. Reports 52.26% on RoboTwin 2.0 (50 tasks; π0 46.4%) and 32.9 Hz inference on an RDK S600 edge device in INT16.把多视角视觉融合压缩为紧凑 token，再条件化流匹配动作生成的轻量策略。报告 RoboTwin 2.0 五十任务平均 52.26%（π0 为 46.4%），真实双臂平台 72.0%，在 RDK S600 边缘设备上以 INT16 推理达 32.9 Hz。
@misc{hou2026rtpolicy,
title = {{RTPolicy: Real-Time (32.9 Hz) Multi-View Visuomotor Policy for Robotic Manipulation on Edge Devices}},
author = {Xianbao Hou and Zehao Ni and Yonghao He and Lingfeng Qian and Xin Shi and Hu Su and Wei Sui and Cong Yang},
year = {2026},
eprint = {2610.41625},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.41625},
}
Tutorial on data-driven, Jacobian-based shape control of deformable objects: problem setup, shape representation, online Jacobian estimation, a local stability condition, and why some shapes are unreachable.数据驱动、基于雅可比的柔性物体形状控制教程：问题建模、形状表示、形变雅可比在线估计、局部稳定条件，以及为何部分目标形状不可达。附代码。
@misc{cuiralzueco2026tutorial,
title = {{A Tutorial on Shape Servoing of Deformable Objects-Part I: Fundamentals}},
author = {Ignacio Cuiral-Zueco and Alessio Caporali and Miguel Aranda and Gianluca Palli and Gonzalo López-Nicolás and Youcef Mezouar},
year = {2026},
eprint = {2610.40559},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.40559},
}
CoRL 2026 Oral. Trains a VLA on over 20,854 hours of action-labeled egocentric human video to test whether large-scale human data can support fine, high-DoF dexterous manipulation.CoRL 2026 Oral。用超过 20,854 小时带动作标注的第一视角人类视频训练 VLA，检验大规模人类数据能否支撑高自由度精细灵巧操作。
@misc{zheng2026egoscale,
title = {{EgoScale: Scaling Dexterous Manipulation with Diverse Egocentric Human Data}},
author = {Ruijie Zheng and Dantong Niu and Yuqi Xie and Jing Wang and Mengda Xu and Yunfan Jiang and Fernando Castañeda and Fengyuan Hu and You Liang Tan and Letian Fu and Trevor Darrell and Furong Huang and Yuke Zhu and Danfei Xu and Linxi Fan},
year = {2026},
eprint = {2602.16710},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.16710},
}
CoRL 2026 Oral. Generates executable dexterous grasps from free-form language, inferring the function a task implies from multi-view images and geometry, supported by the OpenDexVerse data.CoRL 2026 Oral。根据自由语言生成可执行的灵巧抓取：结合多视角图像与几何推断任务所需的功能，由 OpenDexVerse 数据支撑。
@misc{zhang2026opendexgrasp,
title = {{OpenDexGrasp: Open-vocabulary Task-Oriented Dexterous Grasping}},
author = {Jiyao Zhang and Junhan Wang and Tianyu Wang and Zeyuan Chen and Anthony Bolton and Yitong Peng and Hao Dong},
year = {2026},
eprint = {2609.18117},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.18117},
}
CoRL 2026 Oral. Asks what dexterous play pretraining actually needs before precise assembly, a contact-rich, sparse-reward task that is hard for both imitation and direct RL.CoRL 2026 Oral。探究灵巧“玩耍”预训练究竟需要什么，才能胜任精密装配这类富接触、稀疏奖励、模仿与直接 RL 都难以解决的任务。
@misc{lum2026play,
title = {{Play2Perfect: What Matters in Dexterous Play Pretraining for Precise Assembly?}},
author = {Tyler Ga Wei Lum and Kushal Kedia and C. Karen Liu and Jeannette Bohg},
year = {2026},
eprint = {2606.26428},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.26428},
}
CoRL 2026 Spotlight. A large-scale bimanual dexterous grasp dataset, built with a two-stage synthesis pipeline, together with a learning-based grasp generation framework.CoRL 2026 Spotlight。大规模双手灵巧抓取数据集（由两阶段合成流程生成），并配套基于学习的抓取生成框架。
@misc{lin2026bidexgrasp,
title = {{BiDexGrasp: Coordinated Bimanual Dexterous Grasps across Object Geometries and Sizes}},
author = {Mu Lin and Yi-Lin Wei and Jiaxuan Chen and Yuhao Lin and Shuoyu Chen and Jiangran Lyu and Jiayi Chen and Yansong Tang and He Wang and Wei-Shi Zheng},
year = {2026},
eprint = {2604.06589},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.06589},
}
CoRL 2026 Spotlight. Augments contact-rich teleoperation with corrective behaviors learned in simulation, using a simple kNN surrogate of the human operator.CoRL 2026 Spotlight。为富接触遥操作叠加在仿真中学到的纠正行为，并用简单的 kNN 模型替代人类操作员进行训练。
@misc{sha2026efficient,
title = {{Efficient and Reliable Teleoperation through Real-to-Sim-to-Real Shared Autonomy}},
author = {Shuo Sha and Yixuan Wang and Binghao Huang and Antonio Loquercio and Yunzhu Li},
year = {2026},
eprint = {2603.17016},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2603.17016},
}
CoRL 2026 Spotlight. Conditions a mobile manipulation policy on a spatiotemporal map of neural points that represents environment and robot body in one latent space, updated online.CoRL 2026 Spotlight。用时空神经点地图作为移动操作策略的条件：环境与机器人本体处于同一隐空间，并在线更新。
@misc{kim2026serf,
title = {{SERF: Spatiotemporal Environment and Robot Feature Map for Long-Horizon Mobile Manipulation}},
author = {Sunghwan Kim and Byeonghyun Pak and Kehan Long and Yulun Tian and Nikolay Atanasov},
year = {2026},
eprint = {2606.12956},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.12956},
}
Sensorized human demos to whole-body policies: scene-aware retargeting, masking of mismatched views in training, then DAgger corrections. Reports 87.0% task progress on five tasks.把传感化的人类示范变成全身策略：结合重建的场景几何做重定向，训练时用重定向误差遮蔽不匹配的视角，再用 DAgger 人工纠正补足接触。作者报告五个任务平均进度 87.0%。
@misc{liu2026pixel,
title = {{Pixel-UMI: Learning Environment-Aware Whole-Body Manipulation from Sensorized Human Demonstrations}},
author = {Jason Jingzhou Liu and Tony Tao and Deepak Pathak and Ruslan Salakhutdinov},
year = {2026},
eprint = {2611.03647},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03647},
}
A teacher sees future force/torque and optical flow; its features are distilled into a student using only current observations, so the policy anticipates contact. Tested on five real contact-rich tasks.教师网络看到未来的力/力矩与光流，再把动力学特征蒸馏给只用当前观测的学生网络，使策略在部署时能预判接触。在五个真实接触丰富任务上测试。
@misc{tang2026forceldm,
title = {{ForceLDM: A Force-aware Latent Dynamics Model for Contact-Rich Manipulation}},
author = {Peiyuan Tang and Yao Li and Xiaodong Zhang and Wuyang Zhang and Qin Xia and Yanyong Zhang and Zijiang Yang},
year = {2026},
eprint = {2611.03294},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03294},
}
An LLM agent writes partial task plans over perception, planning and control tools; classical search completes them and geometric constraints are bound through the tools. Run in closed loop on Rainbow Y1 and Unitree G1.大模型智能体基于感知、规划、控制工具生成部分任务规划，再由经典搜索补全，并通过工具满足几何约束；在 Rainbow Y1 与 Unitree G1 上闭环运行，交替观察与操作。
@misc{mishra2026tampagent,
title = {{TAMPAgent: Agentic Belief-Space Planning for Mobile Manipulation}},
author = {Utkarsh A. Mishra and Alperen Degirmenci and Yongxin Chen and Danfei Xu and Tucker Hermans and Caelan Garrett},
year = {2026},
eprint = {2611.03584},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03584},
}
Airflow extension for Isaac Lab with experimentally identified models, plus a light blower beside the gripper. Reports zero-shot real push-and-blow at 80% over 20 trials and 2.6° heading error with air alone.为 Isaac Lab 增加基于实测模型的气流扩展，并在夹爪旁加装轻型鼓风机。报告零调试真机部署：推吹混合策略 20 次成功率 80%，纯气流转向平均误差 2.6°。
@misc{yang2026beyondcontact,
title = {{BeyondContact: A Reinforcement Learning Framework for Airflow-Based Non-Contact Manipulation}},
author = {Peiyu Yang and Mohamad S. Moudallal and Antonio Langella and Giulio Turrisi and Victor Barasuol and Claudio Semini},
year = {2026},
eprint = {2611.03059},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03059},
}
Coding-agent search over robot programs that adds physical understanding at object, scene and interaction levels: a knowledge base, scene grounding, and self-designed experiments that test competing hypotheses.让编码智能体搜索机器人程序时具备物体、场景和交互三层物理理解：知识库、场景落地，以及用自行设计的实验检验相互竞争的假设。
@misc{liu2026roboauto,
title = {{RoboAuto: Autonomous Robot Program Search through Multi-Level Physical Understanding}},
author = {Buzhao Liu and Jianing Ye and Xiu Yuan and Yangtong Feng and Chengsong Huang and Tyler Dinh and Yevgeniy Vorobeychik and Ning Zhang and Chongjie Zhang},
year = {2026},
eprint = {2610.36533},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36533},
}
Fuses per-object VLM plausibility and geometric occupancy into one 3D voxel belief that decides where to look, go or open next; on a bimanual mobile humanoid it beats geometry-only search and LLM planning.把逐物体的 VLM 合理性评分与几何占据证据融合为统一的三维体素信念，决定下一步看哪、去哪、开哪个柜子；在双臂移动人形上优于纯几何搜索和 LLM 规划。
@misc{agarwal2026knowing,
title = {{Knowing Where to Look: Interactive Object Search via Spatial-Semantic Fusion}},
author = {Aditya Agarwal and Lucy Cai and Kelly Lu and Travis Manderson and Nicholas Roy and Tomás Lozano-Pérez and Leslie Pack Kaelbling},
year = {2026},
eprint = {2611.00499},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.00499},
}
Treats human-to-robot transfer as virtual try-on: keeps the motion and swaps the person for a target robot in video; reports novel-task success rising from 2% to 60% with no real demos of those tasks.把人到机器人的迁移看作虚拟试穿：保留动作，把视频中的人换成目标机器人；报告在没有这些任务真机演示时，新任务成功率从 2% 升到 60%。
@misc{lee2026synthesizing,
title = {{Synthesizing Dexterous Robot Demonstrations from Human Videos via Virtual Try-On}},
author = {Seungyong Lee and Suhyeok Jang and Minyoung Kim and Jinwoo Shin},
year = {2026},
eprint = {2611.05163},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05163},
}
A language-model agent resolves instructions against a posed RGB-D scan at test time instead of a prebuilt representation; 35 of 50 real Spot trials succeed, and most failures come from physical skills.语言模型智能体在测试时对带位姿的 RGB-D 扫描做指令落地，不依赖预建表示；Spot 真机 50 次试验成功 35 次，失败大多来自物理技能而非推理。
@misc{agarwal2026feijoa,
title = {{FEIJOA: Task-Driven Grounding and Execution for Mobile Manipulation}},
author = {Aditya Agarwal and William Shen and Sahit Chintalapudi and Lucy Cai and Nishanth Kumar and Aidan Curtis and Travis Manderson and Tomás Lozano-Pérez and Leslie Pack Kaelbling},
year = {2026},
eprint = {2611.04080},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.04080},
}
Sets dexterous goals as joint hand-object configurations and learns one transfer primitive from synthetic pairs, with no human motion; reports 69.4% on unseen objects in simulation and 61.5% over 200 real transfers.把灵巧操作目标定义为手与物体的联合构型，从合成构型对学习统一的转移基元，无需人类动作；报告仿真未见物体成功率 69.4%，200 次真机转移成功率 61.5%。
@misc{gui2026dexhot,
title = {{DexHOT: Learning Hand-Object Configuration Transfer for Dexterous Manipulation}},
author = {Zhewei Gui and Anthony Bolton and Jiyao Zhang and Zeyuan Chen and Qiyang Yan and Tianyu Wang and Hao Dong},
year = {2026},
eprint = {2611.05119},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05119},
}
A battery-powered handheld UMI whose soft fingers match the robot and show force by deforming, plus a 99.7-hour dataset; reports a top camera adds 13 points of success with only about half the data including it.电池供电的手持 UMI，软指与机器人夹爪匹配，并以形变反映受力，附 99.7 小时数据集；报告即使仅约一半数据含顶部相机，成功率也提高 13 个百分点。
@misc{kelemen2026robot,
title = {{Robot-Matched Soft Fingers for UMI Data Collection: Lessons from 100 Hours of Pick and Place}},
author = {Kobi Kelemen and Tomasz Niedziałkowski and Marcin Paszczak and Victoriano Montesinos},
year = {2026},
eprint = {2611.03779},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03779},
}
Second part of a tutorial on data-driven, Jacobian-based shape control: which shape changes are reachable, how feature choice shapes behaviour, and the three ways the control loop fails.基于数据与雅可比的形状控制教程第二部分：哪些形变可达、特征选择如何影响行为，以及控制回路失效的三种方式。
@misc{cuiralzueco2026tutorial,
title = {{A Tutorial on Shape Servoing of Deformable Objects-Part II: Advanced Concepts}},
author = {Ignacio Cuiral-Zueco and Alessio Caporali and Miguel Aranda and Gianluca Palli and Gonzalo López-Nicolás and Youcef Mezouar},
year = {2026},
eprint = {2611.03825},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03825},
}
CORD-SLS pairs a GPU-parallel differentiable simulator with robust MPC and conformal bounds for rope and cloth, reporting millisecond planning with safety margins on hardware.CORD-SLS 将 GPU 并行可微仿真与鲁棒 MPC、共形预测误差界结合，用于绳索和布料操作，报告在实机上以毫秒级规划并保持安全裕度。
@misc{li2026robustness,
title = {{Robustness without Wrinkles: Parallel Simulation and Robust MPC for Certified Deformable Manipulation}},
author = {Wei-Chen Li and Jeffrey Fang and Sasanka Polisetti and Yuexi Song and Glen Chou},
year = {2026},
eprint = {2606.14188},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.14188},
}
Refines cloth actions online from one RGB image: a real-to-sim module estimates the cloth state, then simulator rollouts with prior-guided MPPI correct the policy's trajectory.仅凭单张 RGB 在线修正布料动作：先由 real-to-sim 模块估计布料状态，再用仿真器并行推演和先验引导的 MPPI 修正策略轨迹。
@misc{liu2026enabling,
title = {{Enabling Robust Cloth Manipulation via Inference-Time Simulator-in-the-Loop Refinement}},
author = {Xin Liu and Yulin Li and Ziming Li and Pengyu Jing and Zhenhao Huang and Bingyang Zhou and Ziqiu Zeng and Siyuan Luo and Chenkun Qi and Fan Shi},
year = {2026},
eprint = {2606.24552},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.24552},
}
Generates dexterous grasps as 3D hand keypoints with a plain Transformer flow model, with no contact losses or refinement; reports 76.3% success on Dexonomy and 0.032 s per grasp.以 3D 手部关键点表示灵巧抓取，用普通 Transformer 流模型生成，无需接触损失或后处理优化；报告 Dexonomy 成功率 76.3%，每次抓取 0.032 秒。
@misc{huang2026kpgrasp,
title = {{KPGrasp: Scalable Keypoint Flow Matching for Dexterous Grasp Generation}},
author = {Yuansen Huang and Jiayi Chen and Haoran Liu and Yubin Ke and Bing Han and Jiangran Lyu and Mi Yan and Li Yi and He Wang},
year = {2026},
eprint = {2606.09314},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.09314},
}
Helps RL explore contact-rich tasks by resetting episodes with a contact-aware constrained state sampler, covering diverse contact modes; trains universal in-hand and whole-body policies.用感知接触的约束状态采样器重置回合，覆盖多样接触模式，帮助 RL 探索接触丰富任务；可训练通用的手内和全身操作策略。
@misc{toussaint2026combining,
title = {{Combining Constrained Sampling and Reinforcement Learning for Manipulation}},
author = {Marc Toussaint and Cornelius V. Braun and Armand Jordana and Sayantan Auddy and Eckart Cobo-Briesewitz and Denis Shcherba and Tilman Burghoff and Justin Carpentier},
year = {2026},
eprint = {2602.08557},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.08557},
}
SR-DP splits a trained diffusion denoiser by SVD into rank-one experts with task routing, so joint training stops hurting single tasks; reports beating per-task baselines on six RoboTwin tasks.SR-DP 用 SVD 把训练好的扩散去噪器拆成秩一专家并按任务路由，缓解多任务联合训练的相互干扰；报告在六个 RoboTwin 任务上超过逐任务基线。
Retargets human demos under contact without a differentiable simulator: gradients come from archived sampling rollouts. Reports 92.8% success on 1,352 bimanual OakInk2 demos with no task-specific contact rewards.无需可微仿真器即可在接触动力学下重定向人类示教：从存档的采样轨迹中恢复梯度。报告在 1352 条双手 OakInk2 示教上成功率 92.8%，且不用任务专属接触奖励。
@misc{liu2026grasp,
title = {{GRASP: Gradient-Guided Sampling for Contact-Rich Physics-Based Motion Retargeting}},
author = {Yang Liu and Tong Wang and Yiming Bao and Heng Yang},
year = {2026},
eprint = {2611.05754},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05754},
}
PHIL shows an operator the policy's predicted motion in XR so they can step in before an error, then trains on rejected vs corrected chunks; reports 70.3% vs 44.0% for HG-DAgger with a diffusion policy.PHIL 在 XR 中预览策略预测的动作，让操作者在出错前介入，再用被否决与修正后的动作片段做偏好训练；报告相比 HG-DAgger，扩散策略成功率从 44.0% 升至 70.3%。
@misc{jiang2026correct,
title = {{Correct Robots Before They Make Mistakes: Proactive Human-in-the-Loop Intervention via Preference Learning}},
author = {Xinkai Jiang and Jing Jiang and Hongyi Zhou and Pankhuri Vanjani and Ahmad Baki and Jakub Suliga and Ge Li and Zhuoyue Li and Gerhard Neumann and Rudolf Lioutikov},
year = {2026},
eprint = {2611.05667},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05667},
}
Two-stage post-training for VQ-style skill policies: align the skill prior with successful online rollouts, then optimise per skill with the encoder frozen; reports LIBERO success rising from 80.1% to 97.5%.面向 VQ 类技能策略的两阶段后训练：先让技能先验对齐线上成功轨迹，再冻结编码器按技能优化；报告 LIBERO 成功率从 80.1% 升至 97.5%。
@misc{xiang2026cso,
title = {{CSO: Refining Robotic Policies via Skill Distribution Alignment and Skill-Grained Optimization}},
author = {Zhiyuan Xiang and Xiang Deng and Wanjie Tao and Xu Chen and Zhao Jielun and Weili Guan},
year = {2026},
eprint = {2610.38357},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.38357},
}
Updates a latent estimate of hidden object properties during execution to condition an imitation policy; helps with center-of-mass changes, though it reports only about 30% success on unseen locations in sim.执行中在线更新物体隐藏物性的潜变量，并以此调节模仿策略；对质心变化有帮助，但报告仿真中未见过的质心位置成功率仅约 30%。
Adapts a frozen diffusion or flow policy at test time: a VLM writes differentiable rewards that steer denoising when obstacles or clutter shift the scene. Reports +31% on CALVIN and +13% on LIBERO-PRO; no retraining.在测试时适配冻结的扩散或流匹配策略：当障碍或杂物改变场景时，由 VLM 生成可微奖励来引导去噪。报告 CALVIN 提升 31%、LIBERO-PRO 提升 13%，无需重新训练。
Dexterous visuotactile VLA on a mixture-of-transformers: contact-aware gating decides when touch matters; the model jointly imagines future visual and tactile latents. Reports 71% average success on contact-rich tasks.基于混合 Transformer 的灵巧视触觉 VLA：接触感知门控决定何时引入触觉，并联合想象未来的视觉与触觉潜变量。报告在富接触任务上平均成功率 71%。
@misc{fu2026decal,
title = {{DeCAL: Towards Physically-Grounded Dexterous Vision-Language-Action Models via Contact-Aware Latent Co-Imagination}},
author = {Yankai Fu and Ning Chen and Junkai Zhao and Heng Zhang and Guocai Yao and Pengwei Wang and Zhongyuan Wang and Shanghang Zhang},
year = {2026},
eprint = {2609.09119},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.09119},
}
Mobile-manipulation agent with two loops: a VLM plans and composes flow-matching skills at run time, and an outer loop mines rollouts to find and fine-tune skills. Reports +22.5 pts over π0.5 on BEHAVIOR-1K.双循环移动操作智能体：内循环由 VLM 规划并组合流匹配技能，外循环从部署数据中自动发现并微调技能。报告在 BEHAVIOR-1K 上比 π0.5 高 22.5 个百分点。
@misc{liu2026mobiagent,
title = {{MobiAgent: Dual-Loop Recursive Policy Self-Improvement for Long-Horizon Mobile Manipulation}},
author = {Chenzhi Liu and Yue Zhang and Jiehong Lin and Jianan Wang and Bo Wang and Zhongrui Wang and Xiaojuan Qi},
year = {2026},
eprint = {2610.03476},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.03476},
}
Seeker learns where to look from action supervision alone, giving progress-aware crops on frozen DINOv3 features. Reports real-robot success 48.3% to 76.7%, and 20% to 60% under lighting and background shifts.Seeker 仅凭动作监督学习“看哪里”，在冻结的 DINOv3 特征上给出随任务进展变化的裁剪区域。报告真机成功率从 48.3% 升到 76.7%，光照与背景变化下从 20% 升到 60%。
@misc{zhuang2026attention,
title = {{Attention from Action, for Action: Emergent Visual Bottlenecks for Policy Learning}},
author = {Zheyu Zhuang and Ruiyu Wang and Nick Heppert and Johannes Fabian Hahn and Abhinav Valada and Florian T. Pokorny and Danica Kragic},
year = {2026},
eprint = {2608.13422},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2608.13422},
}
5B world action model pretrained on 605 h of RGB-D UMI data that predicts RGB, metric depth and actions. Adds RoboTwin-UMI; reports 85% average real success, +24.2 points over the best baseline.在 605 小时 RGB-D UMI 数据上预训练的 5B 世界动作模型，同时预测 RGB、度量深度和动作。并提出 RoboTwin-UMI 基准；报告真机平均成功率 85%，比最强基线高 24.2 个百分点。
@misc{yuan2026spatial,
title = {{Spatial-UMI-Dream: Scaling World Action Models with Metric Geometry for UMI Manipulation}},
author = {Weiduo Yuan and Ze Yuan and Sihan Liu and Jionghao Wang and Liang Heng and Zhenyu Zhao and Yue Wang},
year = {2026},
eprint = {2611.07229},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.07229},
}
Augments mobile-manipulation demos with new base poses, re-rendering the moving object too, via Gaussian splatting and action retargeting. Reports more than doubling the best baseline's success on unseen base poses.用高斯泼溅与动作重定向为移动操作示教合成新的底盘位姿视角，连被操作的运动物体也一起重渲染。报告在未见底盘位姿下成功率是最强基线的两倍以上。
@misc{kumar2026momasplat,
title = {{MOMASPLAT: Dynamic Novel-View Demonstration Augmentation for Mobile Manipulation}},
author = {Sateesh Kumar and Arnav Balaji and Arpit Bahety and Skand Peri and Georgios Pavlakos and Roberto Martín-Martín},
year = {2026},
eprint = {2611.08372},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08372},
}
Distills sim RL rollouts into a frozen flow-matching action decoder, then learns only a visual interface to its latent from a few demos. Reports beating fine-tuned π0.5 and GR00T N1.7 with 5-10x fewer demos.把仿真中强化学习的经验蒸馏进冻结的流匹配动作解码器，之后只需用少量示教学习从图像到其潜变量的视觉接口。报告用少 5–10 倍的示教超过微调后的 π0.5 与 GR00T N1.7。
@misc{wang2026latent,
title = {{Latent RL Priors for Data-Efficient Dexterous Manipulation}},
author = {Yixiao Wang and Peiyi Chen and Mustafa Mukadam and Shreekant Gayaka and Masayoshi Tomizuka and Son Tran},
year = {2026},
eprint = {2611.08458},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08458},
}
Improves a frozen one-step generative policy offline by learning a residual on its input noise, trained with advantage-weighted reconstruction instead of critic gradients, so the base behavior does not drift.离线改进冻结的一步生成策略：学习对输入噪声的残差修正，用优势加权的动作重建来训练而不是用评论家梯度，从而避免偏离原有行为。
@misc{tang2026awls,
title = {{AWLS: Advantage-Weighted Latent Steering for One-Step Generative Robotic Manipulation Policies}},
author = {Jiahong Tang and Jiaju He and Zhiyuan Yang and Ruyang Han and Haiyue Zhu},
year = {2026},
eprint = {2610.27736},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.27736},
}
Hierarchical real-world RL for functional dexterous tasks: a hand skill handles contact, an arm-hand policy coordinates and triggers it. Reports 98.3% average success on six tasks within one hour of online training.面向功能性灵巧操作的分层真机强化学习：手部技能负责精确接触，臂-手协调策略负责手臂运动与技能触发。报告在六个任务上一小时在线训练即达 98.3% 平均成功率。
@misc{huang2026hifun,
title = {{HiFun: a Hierarchical Framework for Efficient Functional Dexterous Manipulation Learning}},
author = {Linyi Huang and Guowei Huai and Weibin Liu and Shulong Jiang and Ping Tan and Weixuan Zhang and Hui Zhang and Jie Song},
year = {2026},
eprint = {2610.19518},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.19518},
}
Adds a learned, phase-conditioned affordance field to SE(3)-equivariant policies, anchoring observations and actions on the active region instead of the gripper frame. Improves MimicGen and multi-object real tasks.为 SE(3) 等变策略加入随阶段变化的可学习可供性场，以当前作用区域而非夹爪坐标系作为观测与动作的锚点。在 MimicGen 与多物体真机任务上都有提升。
@misc{chen2026phase,
title = {{Phase-Conditioned Affordance Fields for SE(3)-Equivariant Visuomotor Policies}},
author = {Xinzhe Chen and Liqi Huang and Sihua Ren and Haowen Sun and Mingyang Li and Zeyang Liu and Xingyu Chen and Xuguang Lan},
year = {2026},
eprint = {2610.17281},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17281},
}
Meta-learned policy and forward model share a belief latent; at deployment the latent is updated from action-outcome mismatches, so the policy adapts to unknown dynamics on the fly.元学习的策略与前向动力学模型共享信念潜变量；部署时根据动作结果的偏差更新该潜变量，使策略能即时适应未知动力学。
@misc{li2026scouting,
title = {{Scouting the Dynamics Gap: Test-Time Policy Adaptation via Action-Outcome Feedback}},
author = {Yishu Li and Liyuan Geng and Xinyi Mao and Amber Li and David Held},
year = {2026},
eprint = {2609.36107},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.36107},
}
Slow system plans sparse long-horizon intent, fast system refines dense motion, linked by a latent Schrödinger bridge. Reports 45.8% lower latency than Cosmos-Policy and 70.8% success on six bimanual tasks.慢系统生成稀疏的长时程意图，快系统细化高频动作，二者以潜空间薛定谔桥耦合。报告延迟比 Cosmos-Policy 低 45.8%，六个双臂灵巧任务成功率 70.8%。
@misc{chen2026lima,
title = {{LiMA: Bridging Long-term Imagination to Real-time Dexterous Manipulation via Asynchronous Diffusion}},
author = {Ning Chen and Yankai Fu and Junkai Zhao and Qianpu Sun and Guocai Yao and Pengwei Wang and Zhongyuan Wang and Shanghang Zhang},
year = {2026},
eprint = {2609.28431},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.28431},
}
For tasks people cannot teleoperate well, an LLM writes state-based guardrails that filter teleop and policy commands. Reports data-collection success rising from 0-10% to 70-90%.针对人难以遥操作完成的任务，由 LLM 编写基于状态的护栏来过滤遥操作与策略指令。报告数据采集成功率从 0–10% 提升到 70–90%。
@misc{song2026learning,
title = {{Learning Beyond What Humans Can Demonstrate}},
author = {Yuchen Song and Aditya Mittal and Unnat Jain},
year = {2026},
eprint = {2609.24996},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.24996},
}
Uses hand-object proximity from reconstructed point clouds as a hardware-agnostic contact signal, for VR teleop feedback and a policy with learned proximity dynamics. More robust under occlusion.以重建点云得到的手-物距离作为与硬件无关的接触信号，用于 VR 遥操作反馈，并训练带接近度动力学的策略；在遮挡下更稳健。
@misc{bai2026proxidex,
title = {{ProxiDex: Learning Dynamics-Guided Proximity Policy for Dexterous Manipulation}},
author = {Yushan Bai and Boyu Zheng and Zhiyang Mao and Hongzheng Sun and Yuchuang Tong and En Li and Zhengtao Zhang},
year = {2026},
eprint = {2609.16586},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.16586},
}
RL for mating two parts within one dexterous hand (capping a marker, a syringe, a bottle) with no second arm. Trained in sim, transfers zero-shot with a single camera.用强化学习在单只灵巧手中完成两个零件的装配（如给马克笔盖帽、注射器、瓶子），不用第二只手臂；仿真训练，单相机零样本迁移到真机。
@misc{pei2026assembling,
title = {{Assembling Two Parts in One Hand}},
author = {Liuao Pei and Tianyue Wu and Hui Zhang and Ping Luo and Jie Song},
year = {2026},
eprint = {2609.10137},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.10137},
}
Represents skills as short interaction segments of relative SE(3) motion between task objects, independent of scene layout and robot. Needs much less data than prior policies on RLBench and COLOSSEUM.把技能表示为任务物体之间相对 SE(3) 运动的短交互片段，与场景布局和机器人无关。在 RLBench 与 COLOSSEUM 上所需数据明显少于以往方法。
@misc{fu2026foci,
title = {{FOCI Policy: Focus on Object-Centric Interactions for Relational Manipulation Policies}},
author = {Ze Fu and Pinhao Song and Yutong Hu and Renaud Detry},
year = {2026},
eprint = {2609.08743},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.08743},
}
Plug-in adapter that projects the robot state onto the image to build a grounded state token and FiLM-modulate nearby features. Reports +8.7% for Diffusion Policy in sim and 10.6% average real-world gain.即插即用适配器：把机器人状态投影到图像平面，采样局部视觉特征构成“落地”的状态 token，并用 FiLM 调制对应特征。报告仿真中 Diffusion Policy 提升 8.7%，真机平均提升 10.6%。
@misc{zhao2026geoprop,
title = {{GeoProp: Grounding Robot State in Vision for Generalist Manipulation}},
author = {Guoyang Zhao and Quanhao Qian and Gongjie Zhang and Wenhao Li and Jiuniu Wang and Xiaowei Lu and Deli Zhao and Ran Xu},
year = {2026},
eprint = {2607.07101},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.07101},
}
Knife peeling as a test case for tasks with subjective quality: force-aware imitation first, then preference-based fine-tuning with a learned reward. Reports 100% success on cucumbers, apples and potatoes.以用刀削皮为例研究“质量难以量化”的精细操作：先做力感知模仿学习，再用学习到的奖励模型按人类偏好微调。报告在黄瓜、苹果、土豆上成功率 100%。
@misc{lin2026how,
title = {{How to Peel with a Knife: Aligning Fine-Grained Manipulation with Human Preference}},
author = {Toru Lin and Shuying Deng and Zhao-Heng Yin and Pieter Abbeel and Jitendra Malik},
year = {2026},
eprint = {2603.03280},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2603.03280},
}
Lets text-only LLMs do few-shot bimanual manipulation without fine-tuning by treating the arms as leader and follower agents. Reports 70.5% average success on 13 TWIN tasks.把双臂视为“主-从”两个智能体，让纯文本 LLM 无需微调即可通过上下文学习完成少样本双臂操作。报告在 TWIN 的 13 个任务上平均成功率 70.5%。
@misc{palma2026bimanual,
title = {{Bimanual Robot Manipulation via Multi-Agent In-Context Learning}},
author = {Alessio Palma and Indro Spinelli and Vignesh Prasad and Luca Scofano and Yufeng Jin and Georgia Chalvatzaki and Fabio Galasso},
year = {2026},
eprint = {2604.20348},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.20348},
}
Co-trains a bimanual policy on sim data (for robot-valid actions) and human videos (for real visuals), then fine-tunes on 80 real episodes per task. Reports 62.5% on held-out scenes, 53.7 points above real-only.用仿真数据（提供机器人可执行动作）和人类视频（提供真实视觉）联合预训练双臂策略，再在每任务 80 条真机数据上微调。报告在未见场景上成功率 62.5%，比纯真机数据高 53.7 个百分点。
@misc{fang2026sim,
title = {{Sim-and-Human Co-training for Data-Efficient and Scene-Generalizable Bimanual Manipulation}},
author = {Kaipeng Fang and Weiqing Liang and Yuyang Li and Ji Zhang and Pengpeng Zeng and Heng Tao Shen and Jingkuan Song and Lianli Gao},
year = {2026},
eprint = {2601.19406},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2601.19406},
}
Lets a policy trained only on static objects handle moving ones: demos are augmented by moving the object to a counterfactual future pose and warping the action chunk; a pose predictor supplies the target at test time.让只用静止物体演示训练的策略也能抓取运动物体：把物体平移到反事实的未来位姿并相应变形动作块来扩增数据，部署时由位姿预测器给出目标位置。
@misc{shin2026static,
title = {{Static In, Dynamic Out: Counterfactual Action Augmentation for Moving Object Manipulation}},
author = {Woo Chul Shin and Zhenyang Chen and Alfred Cueva and Nadun Ranawaka Arachchige and Yingyan Celine Lin and Benjamin Joffe and Shreyas Kousik and Danfei Xu},
year = {2026},
eprint = {2607.27890},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.27890},
}
Composes simple behaviours (semantic constraints from foundation models, keypoints from imagined video, tactile/force corrections) into one SE(3) pose command per step. 8 real tasks, incl. seating a GPU.把语义约束（基础模型）、想象视频中的关键点预测、触觉/力实时修正等简单行为在 SE(3) 上组合成每一步的位姿指令。8 个真机任务，包括把显卡插入 PCIe 槽。
@misc{chen2026costream,
title = {{CoStream: Composing Simple Behaviors for Generalizable Complex Manipulation}},
author = {Haonan Chen and Yuxiang Ma and Stephen Tian and Xiaoshen Han and Wenlong Huang and Feiyang Wu and Yunzhu Li and Jiajun Wu and Edward H. Adelson and Yilun Du},
year = {2026},
eprint = {2606.26423},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.26423},
}
New manipulation tasks where the target must be found by sound, plus an imitation framework that adds acoustic spatial and spectral features to ACT, Diffusion Policy, VQ-BeT and π0. Sim + real.提出需要靠声音确定操作目标的新任务，以及把声源空间信息与频谱特征接入 ACT、Diffusion Policy、VQ-BeT 和 π0 的多模态模仿学习框架。仿真与真机验证。
@misc{hiratsuka2026s,
title = {{S2A2: Audio-Visual Imitation Learning for Manipulation Tasks Using Acoustic Spatial Information}},
author = {Kaneyoshi Hiratsuka and Benjamin Yen and Ryosuke Kojima},
year = {2026},
eprint = {2607.26047},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.26047},
}
Learns an object-conditioned human grasp prior from 1.8K human grasps to pick contact modes and wrist starts for force-closure optimisation; synthesises 3.2M grasps from pinches to bimanual, 2–30 cm objects.从 1800 次人类抓取中学习物体条件的人类先验，用来选择接触模式和手腕初值以指导力封闭优化；合成 320 万个抓取，覆盖两指捏取到双手抓取、2–30 cm 物体。
@misc{yu2026hugs,
title = {{HUGS: Guiding Unified Dexterous Grasp Synthesis Across Modes and Scales via Learned Human Priors}},
author = {Mingrui Yu and Yongpeng Jiang and Yongyi Jia and Kangchen Lv and Li Huang and Yi Ren and Xiang Li},
year = {2026},
eprint = {2607.04554},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.04554},
}
Puts miniature cameras in each fingertip of a multi-fingered hand and trains a whole-body diffusion policy on third-view plus fingertip views. Reports 56.6% relative gain over external and wrist cameras.在多指手的每个指尖装入微型摄像头，用第三视角加多指尖视角训练全身扩散策略。报告比外部+腕部相机方案平均成功率相对提升 56.6%。
@misc{zhang2026fingervip,
title = {{FingerViP: Learning Real-World Dexterous Manipulation with Fingertip Visual Perception}},
author = {Zhen Zhang and Weinan Wang and Hejia Sun and Qingpeng Ding and Xiangyu Chu and Guoxin Fang and K. W. Samuel Au},
year = {2026},
eprint = {2604.21331},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.21331},
}
Pretrains a low-level dexterous controller on 355K trajectories built from grasp datasets, then fine-tunes for articulated tool use. Introduces the DexCraft benchmark; reports +33.3 points over DP3 on a real robot.用抓取数据集构造的 35.5 万条轨迹预训练灵巧手底层控制器，再微调到需要操作工具活动部件的任务；提出 DexCraft 基准。报告真机完整任务成功率比 DP3 高 33.3 个百分点。
@misc{yuan2026from,
title = {{From Grasps to Dexterity: Large-Scale Grasp Pretraining for Dexterous Manipulation}},
author = {Ying Yuan and Xinyu Liu and Sriram Krishna and David Held},
year = {2026},
eprint = {2606.30749},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.30749},
}
In-context imitation for folding: from a single human demo, a flow-matching transformer infers and executes the shown folding mode without gradient updates. Trained only in sim, transfers zero-shot to real.面向折叠的上下文模仿学习：只看一条人类演示，流匹配 Transformer 即可推断并执行其中的折叠方式，无需梯度更新。完全在仿真训练，零样本迁移到真机。
6-DoF grasping that keeps learning on the job through memory in an embedding space: outcomes rescore future grasps and user demos are recalled in new scenes. Reports over 90% on hard categories after 50 tries.通过嵌入空间中的记忆在部署中持续学习的 6-DoF 抓取：抓取结果更新后续评分，用户示范可在新场景中被召回。报告在困难类别上仅 50 次尝试后成功率超过 90%。
@misc{schiavi2026continual,
title = {{Continual Learning for 6-DoF Grasp Synthesis via Experience and Demonstrations}},
author = {Giulio Schiavi and Andrei Cramariuc and Michael Pantic and Roland Siegwart},
year = {2026},
eprint = {2610.01301},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.01301},
}
Diffusion grasp generator conditioned on a graph of the hand's kinematics, so grasps transfer zero-shot to unseen hand designs; comes with a cross-morphology grasp benchmark.以手部运动学结构图为条件的扩散抓取生成模型，可零样本迁移到未见过的机器人手型；同时提出跨手型抓取评测基准。
@misc{wu2026diffmorphgrasp,
title = {{DiffMorphGrasp: A Diffusion Framework Conditioned on Hand Morphology for Cross-Embodiment Grasp Generation}},
author = {Zhiyuan Wu and Xiangyu Zhang and Zhuo Chen and Jiankang Deng and Rolandos Alexandros Potamias and Shan Luo},
year = {2026},
eprint = {2602.00915},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.00915},
}
A short safe wiggle reveals rope parameters that condition a trajectory optimiser for zero-shot dynamic throws. Reports 3.55 cm average error on real 3D target striking vs 15.29 cm for uninformed baselines.先做一次短暂安全的“抖绳”动作来辨识绳子参数，再以此驱动轨迹优化，零样本完成动态甩绳。报告真机 3D 目标击打平均误差 3.55 cm，无辨识基线为 15.29 cm。
@misc{jakobsson2026wiggle,
title = {{Wiggle and Go! System Identification for Zero-Shot Dynamic Rope Manipulation}},
author = {Arthur Jakobsson and Abhinav Mahajan and Karthik Pullalarevu and Krishna Suresh and Yunchao Yao and Yuemin Mao and Bardienus Duisterhof and Shahram Najam Syed and Jeffrey Ichnowski},
year = {2026},
eprint = {2604.22102},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.22102},
}
Uses motor current as a touch signal on low-cost dexterous hands: a model predicts a compliant PD target position whose error yields the right grasp force, with no tactile or force sensors.把电机电流当作触觉：在低成本灵巧手上预测一个“柔顺参考位置”，让 PD 控制器的位置误差产生合适的抓握力，无需触觉或力传感器。
@misc{ma2026current,
title = {{Current as Touch: Proprioceptive Contact Feedback for Compliant Dexterous Manipulation}},
author = {Chenyang Ma and Yunchao Yao and Zhenyu Wei and Ruogu Li and Daniel Szafir and Mingyu Ding},
year = {2026},
eprint = {2607.03529},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.03529},
}
Models a dexterous skill as a latent linear (Koopman) system in which visual flow and robot motion co-evolve; it replans when observed and predicted flow diverge. Matches strong baselines with much faster inference.把灵巧技能建模为潜在线性（Koopman）系统，视觉特征流与机器人动作流共同演化；当观测与预测的视觉流偏离时自动重规划。性能与强基线相当，推理快得多。
@misc{han2026going,
title = {{Going with the Flow: Koopman Behavioral Models as Pseudo Planners for Visuo-Motor Dexterity}},
author = {Yunhai Han and Jiaqi Fu and Linhao Bai and Ziyu Xiao and Zhaodong Yang and Yogita Choudhary and Krishna Jha and Chuizheng Kong and Shreyas Kousik and Harish Ravichandar},
year = {2026},
eprint = {2602.07413},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.07413},
}
Generative augmentation from a single box-annotated image, plus a region-contrastive loss that steers attention to task regions. Beats augmentation baselines on UR5e, AgileX and Tian Gong 2.0 under scene shifts.只需一张带框标注的图片即可做生成式语义增强，并用区域对比损失把注意力引向任务相关区域。在 UR5e、AgileX 与天工 2.0 上，面对背景、干扰物和光照变化时优于其他增强方法。
@misc{wang2026roboaug,
title = {{RoboAug: One Annotation to Hundreds of Scenes via Region-Contrastive Data Augmentation for Robotic Manipulation}},
author = {Xinhua Wang and Kun Wu and Zhen Zhao and Hu Cao and Yinuo Zhao and Zhiyuan Xu and Meng Li and Shichao Fan and Di Wu and Yixue Zhang and Ning Liu and Zhengping Che and Jian Tang},
year = {2026},
eprint = {2602.14032},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.14032},
}
From one human demo: retarget hand-object motion to a robot reference, train residual RL in sim to track object keypoints, transfer zero-shot with careful system ID. Scissors and screwdriver use on two hands.从一条人类演示出发：把手-物运动重定向为机器人参考，在仿真中用残差强化学习跟踪物体关键点，并通过细致的系统辨识零样本迁移。在两种多指手上完成用剪刀、拧螺丝刀等任务。
@misc{feng2026minimalist,
title = {{A Minimalist Retargeting-Guided Reinforcement Learning Recipe for Dexterous Manipulation}},
author = {Yunhai Feng and Natalie Leung and Jiaxuan Wang and Lujie Yang and Haozhi Qi and Preston Culbertson},
year = {2026},
eprint = {2607.11874},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.11874},
}
Teleoperation controller that turns operator intent into learned hand-object contact execution, trained with single-stage RL on human motions. Reports 75% average success on seven in-hand and tool-use tasks.遥操作控制器把操作者意图转化为学习到的手-物接触执行，基于人类动作用单阶段强化学习训练，零样本迁移真机。报告七个手内操作与工具使用任务平均成功率 75%。
@misc{li2026towards,
title = {{Towards Human-level Dexterous Teleoperation}},
author = {Puhao Li and Zeyuan Chen and Yingying Wu and Pengkun Wei and Yuyang Li and Tianyu Wang and Jiaxiao Shi and Mingrui Yu and Baoxiong Jia and Song-chun Zhu and Tengyu Liu and Siyuan Huang},
year = {2026},
eprint = {2607.11481},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.11481},
}
Pretrains a dexterous RL policy on generic object reposing, then post-trains downstream tasks with BC distillation, critic warm-up and conservative updates. Zero-shot sim-to-real on 23-DoF and 29-DoF hand-arm systems.先在通用物体重定位任务上用强化学习预训练灵巧策略，再用行为克隆蒸馏、critic 预热和保守更新后训练下游任务。在 23 与 29 自由度手-臂系统上零样本仿真到真机。
@misc{lee2026adept,
title = {{ADEPT: Accelerating Dexterity via Pre-Training and Post-Training using Reinforcement Learning}},
author = {Jayjun Lee and Jessica Yin and Asif Rana and Nicholas Blauch and Octi Zhang and Sam Mady and Mohak Bhardwaj and Nima Fazeli and Nathan Ratliff and Karl Van Wyk and Ankur Handa},
year = {2026},
eprint = {2608.19182},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2608.19182},
}
Learns object-object interaction priors from human videos and composes them with robot feasibility reasoning and motion planning, so one prior serves a Franka and an RBY-1 mobile humanoid without robot data.从人类视频学习以物体间交互运动表示的先验，再与机器人可行性推理和运动规划组合，同一先验无需机器人数据即可用于 Franka 与 RBY-1 移动人形机器人。
@misc{cheng2026human,
title = {{Human2Any: Human-to-Robot Transfer via Constraint-Aware Compositional Planning}},
author = {Shuo Cheng and Chuye Zhang and Alfred Cueva and Caelan Garrett and Ajay Mandlekar and Danfei Xu},
year = {2026},
eprint = {2606.28813},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.28813},
}
Task-agnostic 'any pose to any pose' dexterous policy trained in sim on thousands of objects and conditioned on 3D point tracks; prompted with tracks from generated videos, it runs real tasks zero-shot.在仿真中用数千种物体训练与任务无关的“任意位姿到任意位姿”灵巧策略，以 3D 点轨迹为条件；用生成视频提取的点轨迹作提示，即可零样本完成真实任务。
Multimodal LLM that reads assembly manuals, point clouds and text to predict 6D assembly poses, plus AssemBench with 900K+ samples. Reports state-of-the-art 6D pose reasoning and real multi-step assembly.读取装配说明书、点云与文字指令来预测 6D 装配位姿的多模态大模型，并发布含 90 万余样本的 AssemBench。报告 6D 位姿推理达到最优，并完成真实多步装配。
@misc{jing2026assemlm,
title = {{AssemLM: A Spatial Reasoning Multimodal Large Language Model for Robotic Assembly}},
author = {Zhi Jing and Jinbin Qiao and Ouyang Lu and Jicong Ao and Shuang Qiu and Huazhe Xu and Yu-Gang Jiang and Chenjia Bai},
year = {2026},
eprint = {2604.08983},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.08983},
}
Diffusion policy that denoises one action in several coordinate frames at once with frame-specialised denoisers and fuses the results. Beats oracle single-frame choice on nine bimanual tasks; two real mobile tasks.扩散策略在多个坐标系中同步去噪同一动作：各坐标系有专门的去噪器，再把噪声预测融合回规范坐标系。在九个双臂任务上优于“最优单坐标系”选择，并在两个真实移动操作任务上验证。
@misc{wang2026mixture,
title = {{Mixture of Frames Policy: Multi-Frame Action Denoising for Bimanual Mobile Manipulation}},
author = {Dian Wang and Jisang Park and Xiaomeng Xu and Han Zhang and Shuran Song and Jeannette Bohg},
year = {2026},
eprint = {2607.11884},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.11884},
}
Task-aware robot-to-human handovers: a VLM generates images of how a person would hold the object for the task, matched live to the person's hand. 83.3% of 12 users felt it understood the task better.面向任务的机器人递物：用 VLM 生成人在该任务下如何握持物体的图像假设，并与实时观测的人手姿态匹配。12 名参与者中 83.3% 认为它比已有方法更懂任务。
A VLM first localizes the task region, then real-world human-in-the-loop residual RL with force feedback learns the contact part. Object-centric background augmentation adds robustness; six real tasks.分两阶段：先由视觉语言模型定位任务相关区域，再在真实世界用带力反馈、人在回路的残差强化学习学习接触细节。以物体为中心的背景增强使其在六个真机任务上保持稳健。
@misc{qiao2026semantics,
title = {{Semantics-to-Contact: A Stagewise Framework for Robust Contact-Rich Manipulation}},
author = {Guanren Qiao and Ruixiang Ouyang and Sheng Xu and Yueci Deng and Yunxin Tai and Kui Jia and Guiliang Liu},
year = {2026},
eprint = {2611.07532},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.07532},
}
Splits bimanual policy features into a swap-symmetric part (coordination that holds when arms trade roles) and a swap-antisymmetric part (role-specific behaviour), recomposed as residuals on each arm's features.把双臂策略特征分解为交换对称分量（两臂互换角色时不变的协调结构）与交换反对称分量（随角色变化的行为差异），再以残差形式加回各臂的表示。
@misc{shen2026biroad,
title = {{BiRoAD: Learning Shared and Role-Adaptive Representations for Bimanual Manipulation}},
author = {Yan Shen and Yuchen Liu and Feng Jiang and Hangtian Hu and Xiaoqi Li and Shu Chen and Ruihai Wu and Hao Dong},
year = {2026},
eprint = {2609.23445},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.23445},
}
Creative tool use (cut a cake with a plate): a VLM picks the object to use as a tool and the relevant parts, then vision foundation models ground those parts as 3D regions from one RGB-D image. No end-to-end training.创造性用工具（没有刀就用盘子切蛋糕）：VLM 选出可充当工具的物体及其相关部位，再由视觉基础模型从单张 RGB-D 图像把这些部位定位为 3D 区域。无需端到端训练。
@misc{deng2026grow,
title = {{GROW$^2$: Grounding Which and Where for Robot Tool Use}},
author = {Yuhong Deng and Yuyao Liu and David Hsu},
year = {2026},
eprint = {2606.30632},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.30632},
}
Diffusion policy for whole-body mobile manipulation: a Fourier encoder turns per-pixel 3D points into tokens gated in by depth validity, and a hybrid decoder gives arm and base separate action subspaces.面向全身移动操作的扩散策略：傅里叶编码器把逐像素 3D 坐标变成几何 token，并按深度可靠性门控注入视觉基础模型特征；混合解码器为机械臂与底盘分配独立的动作子空间。
@misc{zhu2026geohat,
title = {{GeoHAT: Geometry-Adaptive Hybrid Action Transformer for Mobile Manipulation}},
author = {Xiangyu Zhu and Renjun Wu and Luzhou Ge and Jinyan Liu and Xuesong Li},
year = {2026},
eprint = {2606.13394},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.13394},
}
Learns a relational GNN heuristic over a simple symbolic model, with a motion planner revealing hidden geometric preconditions during training. Its top-ranked action is feasible in 94% of states (22% random).在简单符号模型上学习关系图神经网络启发式，训练时由运动规划器揭示隐藏的几何前提。其排名第一的动作在 94% 的状态下几何可行（随机选择为 22%）。
Predicts object flow to synthesize future observations, aligned to real future frames by a contrastive loss, so an asynchronous policy can act ahead of its own latency on moving objects.利用预测的物体流合成未来观测，并以对比学习目标使其特征与真实未来帧对齐，让异步策略能提前补偿自身延迟，处理运动中的物体。
@misc{wei2026f,
title = {{F2F-AP: Flow-to-Future Asynchronous Policy for Real-time Dynamic Manipulation}},
author = {Haoyu Wei and Xiuwei Xu and Ziyang Cheng and Hang Yin and Angyuan Ma and Jie Zhou and Jiwen Lu and Bingyao Yu},
year = {2026},
eprint = {2604.02408},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.02408},
}
Fluent expert demos rush through alignment and recovery. Slowing down there and resampling those segments helps; STAIR adds a compact short-horizon motion feature between the VLM and the action expert.熟练的示教会把对准与纠错一带而过，导致策略在关键处缺少监督。在这些片段放慢、重采样有帮助；STAIR 进一步在 VLM 与动作专家之间加入紧凑的短时运动特征。
@misc{liu2026perfect,
title = {{Perfect Demo Makes Poor Teacher: Learning Robust Alignment from Critical Motion Segments}},
author = {Mingyu Liu and Zeju Li and Jiuhe Shu and Hanqing Wang and Yuhao Chao and Hao Chen and Chunhua Shen},
year = {2026},
eprint = {2606.15587},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.15587},
}
Replaces the sampling noise in diffusion's backward process with an optimized correction, so reachability, collision or trackability constraints hold at sampling time without retraining.把扩散反向过程中的采样扰动替换为经优化得到的修正，从而在采样时满足可达性、避碰或可跟踪性约束，无需重新训练扩散模型。
@misc{bodmer2026grounding,
title = {{Grounding Generative Policies in Physics: Optimization-Guided Diffusion for Robot Control}},
author = {Sabrina Bodmer and René Zurbrügg and Tifanny Portela and Hao Ma and Alexandre Didier and Marco Hutter and Colin Jones and Melanie Zeilinger},
year = {2026},
eprint = {2606.24208},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.24208},
}
ReStruct splits a policy into a state-machine skeleton and residual controller; a VLM turns a language preference into a temporal-logic automaton that reshapes the skeleton. 92.5% preference success on a real robot.ReStruct 把视觉运动策略分解为状态机骨架和残差控制器；VLM 把语言偏好编译成时序逻辑自动机来重构骨架。真机上偏好遵循成功率 92.5%。
@misc{pan2026inference,
title = {{Inference-Time Robot Behavior Steering through Physically-Aware Reconfiguration of Task-Structure}},
author = {Yiyuan Pan and Hanjiang Hu and Shangtao Li and Xusheng Luo and Changliu Liu},
year = {2026},
eprint = {2606.26588},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.26588},
}
Distributed depth sensors on the robot form an obstacle-aware point cloud; a flow-matching policy trained on 100K+ simulated safety-critical trajectories plans collision-free motion and transfers zero-shot to hardware.机器人身上的分布式深度传感器汇成障碍感知点云；在 10 万多条仿真安全关键轨迹上训练的流匹配策略据此生成无碰撞运动，零样本迁移到真机。
@misc{shen2026escape,
title = {{ESCAPE: Egocentric Sensing for Collision Avoidance and Policy Execution}},
author = {Weiqing Shen and Wenhao Huang and Baokang Zhang and Shuo Shang and Yanmiao Hu and Haojun Li and Yifan Han and Wenzhao Lian},
year = {2026},
eprint = {2610.23943},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.23943},
}
End-to-end in-hand reorientation from RGB and proprioception, with no object pose estimation: distilled from privileged-state teachers, and the target is given as an image. One policy transfers to five real objects.仅凭 RGB 与本体感觉、无需物体位姿估计的端到端手内重定向：由特权状态教师蒸馏而来，目标姿态直接用图像给出。单一策略直接迁移到五个真实物体。
@misc{bhardwaj2026goaldex,
title = {{GoalDex: Learning Image-Goal-Conditioned Policies for Dexterous In-Hand Manipulation}},
author = {Arjun Bhardwaj and Mayank Mittal and Maximum Wilder-Smith and Vaishakh Patil and Marco Hutter},
year = {2026},
eprint = {2611.02171},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.02171},
}
Category-level in-hand manipulation of articulated objects: procedural objects and functional grasps, then two-stage training with articulation physics randomization, reward curriculum and latent distillation.类别级的铰接物体手内操作：程序化生成物体与功能性抓取，再用两阶段训练（铰接物理随机化、奖励课程、潜表示蒸馏）学习策略，可泛化到未见实例。
@misc{yang2026artmanip,
title = {{ArtManip: Category-Level Articulated In-Hand Manipulation}},
author = {Yang Yang and Tengyu Liu and Puhao Li and Zeyuan Chen and Yuyang Li and Xingwan Wang and Yingying Wu and Zhaopeng Cui and Siyuan Huang},
year = {2026},
eprint = {2609.12498},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.12498},
}
Dual-arm grasps for large objects from partial point clouds: a convolutional occupancy network reasons about local thickness and clearance, producing force-closure grasp pairs refined by sampling-based optimization.从局部点云为大型物体生成双臂抓取：卷积占据网络推断局部厚度与夹爪间隙等几何，生成满足力封闭的抓取对，再用基于采样的优化修正几何不完整带来的歧义。
@misc{kaura2026partialbigrasp,
title = {{PartialBiGrasp: Inferring Hidden Local Geometry for Bimanual Grasping from Partial Views}},
author = {Ayush Kaura and Vignesh Vembar and Md Faizal Karim and Keshab Patra and K Madhava Krishna},
year = {2026},
eprint = {2608.19188},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2608.19188},
}
A simulation benchmark of about 0.5B grasp labels over eight dexterous hands and ~50K objects, plus a diffusion model for cross-embodiment grasps in clutter: 88% average success, 0.3 s inference.构建约 5 亿条抓取标注、覆盖 8 种灵巧手和约 5 万个物体的仿真基准，并提出跨本体的杂乱场景灵巧抓取扩散模型：平均成功率 88%，推理 0.3 秒。
@misc{fei2026t,
title = {{T(R,O) Grasp++: Cross-Embodiment Dexterous Grasping in Cluttered Environments at Scale}},
author = {Xin Fei and Zhixuan Xu and Huaicong Fang and Fangjun Chen and Zhaoyu Mu and Rui Yang and Zhexi Luo and Lin Shao},
year = {2026},
eprint = {2610.25410},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25410},
}
Models robot flow as dense velocity fields in a flow-matching formulation rather than sparse keypoint displacements, about 24x faster than standard flow matching; tested in 13 real tasks with 260 trials per method.在流匹配框架中把机器人流建模为稠密速度场，而非稀疏关键点位移，生成速度比标准流匹配快约 24 倍；在 13 个真实任务上每种方法测试 260 次。
@misc{seno2026flow,
title = {{Flow as Flow: Modeling Robot Velocity Fields as Probability Velocity Fields for Flow-Based Object Manipulation}},
author = {Koki Seno and Daichi Yashima and Yusuke Takagi and Kento Tokura and Komei Sugiura},
year = {2026},
eprint = {2606.23090},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.23090},
}
Diffusion policy with an obstacle-aware encoder that splits the scene into target, obstacle and background. 61 real greenhouse trials per method: 75.41% success and 8.20% collision rate.带障碍感知编码器的扩散策略，把场景分解为目标、障碍和背景。每种方法在温室真机测试 61 次：成功率 75.41%，碰撞率 8.20%。
@misc{wang2026obstadiff,
title = {{ObstaDiff: Generalizable Diffusion Policy Learning via Obstacle-aware Representations}},
author = {Jiawen Wang and Kevin Yao and Khalid Jawed},
year = {2026},
eprint = {2609.10918},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.10918},
}
Single-camera diffusion policy: points along intermediate end-effector trajectories in the denoising loop sample multi-scale image features, acting as attention anchors. Comparable to multi-camera baselines in sim.单相机扩散策略：在去噪循环中沿中间末端轨迹的点采样多尺度图像特征，作为视觉注意力锚点。仿真中表现与多相机基线相当。
@misc{kim2026spatially,
title = {{Spatially Conditioned Diffusion Policy: Learning Precise and Robust Manipulation with a Single RGB Camera}},
author = {Seoyoon Kim and Kanghyun Kim and Dongwoo Ko and Yeong Jin Heo and Min Jun Kim},
year = {2026},
eprint = {2606.14535},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.14535},
}
Dexterous food cutting from a handful of real demos: reconstruct objects, randomize physical parameters within category limits and generate augmented demos in sim, then transfer back to the real hand.只用少量真实示教学习灵巧切割食物：重建物体实例，在类别约束下随机化物理参数，在仿真中生成扩增的灵巧示教，再迁移回真实灵巧手。
@misc{lou2026dexninja,
title = {{DexNinja: Learning Robust Dexterous Cutting Policy with a Real-to-Sim-to-Real Data Engine}},
author = {Haozhe Lou and Runyi Yang and Wanqi Zhong and Caiyun Liu and Yicheng Liu and Wenhao Liu and Wenlong Ma and Jiawei Xia and Kanzhi Wu and Danda Pani Paudel and Luc Van Gool and Hang Zhao and Yiming Li},
year = {2026},
eprint = {2610.17208},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17208},
}
Turns monocular egocentric human videos into robot training data: tracks hand and object, retargets to a gripper, and renders photorealistic wrist-camera views with Gaussian Splatting.把单目第一人称人类视频转为机器人训练数据：跟踪手与物体，重定向到机械夹爪，并用高斯泼溅渲染逼真的腕部相机视角。
@misc{freeman2026warped,
title = {{WARPED: Wrist-Aligned Rendering for Robot Policy Learning from Egocentric Human Demonstrations}},
author = {Harry Freeman and Chung Hee Kim and Jenny Wang and George Kantor},
year = {2026},
eprint = {2604.10809},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.10809},
}
CHORD guides dexterous RL by comparing human and robot motions in object-centric contact wrench space. Also a 4,739-task bimanual benchmark built from MoCap and videos; 82.12% average on 1,831 tasks.CHORD 在以物体为中心的接触力旋量空间中比较人与机器人的动作来引导灵巧操作强化学习，并构建来自动捕与视频的 4739 个双手任务基准；在其中 1831 个任务上平均成功率 82.12%。
@misc{zhu2026learning,
title = {{Learning Dexterous Manipulation Using Contact Wrench Guidance From Human Demonstration}},
author = {Xinghao Zhu and Zixi Liu and Shalin Jain and Chenran Li and Milad Noori and Michael Andres Lin and Huihua Zhao and John Welsh and Mrinal Verghese and Wei Liu and Tingwu Wang and Xingye Da and Zhengyi Luo and Vishal Kulkarni and Naema Bhatti and Yuke Zhu and Linxi Fan and Bowen Wen and Danfei Xu and Soha Pouya and Yan Chang},
year = {2026},
eprint = {2607.00033},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.00033},
}
Before contact, a latent world model rolls the policy's planned action chunk forward in multi-view embedding space, and a small probe aborts if the predicted post-contact state looks like failure.在接触发生前，用潜空间世界模型在多视角嵌入空间中推演策略计划的动作块，若预测的接触后状态呈现失败迹象，轻量探针即中止执行。
For catching and intercepting moving objects, forecasts the object's future SE(3) trajectory from short pose histories (15 ms) and picks a reachable contact state; reports real success rising from 15.3% to 67.3%.面向抓取运动物体的主动式动态抓取：由短时位姿历史预测物体未来 SE(3) 轨迹（15 毫秒），再选择可达的接触状态；报告真实环境成功率从 15.3% 提升到 67.3%。
@misc{zhou2026physics,
title = {{Physics-Aware Object-Centric SE(3) Forecasting for Real-Time Proactive Dynamic Grasping}},
author = {Yirui Zhou and Hongli Xu and Ying Zheng and Yi Wang and Lap-Pui Chau},
year = {2026},
eprint = {2610.17843},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17843},
}
Robot-DIFT distills a noisy diffusion feature extractor into a deterministic single-pass backbone with a feature pyramid, so policies see fine correspondence cues for the last millimeters of contact-rich tasks.Robot-DIFT 将依赖噪声的扩散特征提取器蒸馏为确定性的单次前向骨干网络并配合特征金字塔，使策略获得精细的对应关系线索，改善接触密集任务最后几毫米的精度。
@misc{deng2026robot,
title = {{Robot-DIFT: Correspondence-Sensitive Diffusion Features for Contact-Rich Robot Manipulation}},
author = {Yu Deng and Yufeng Jin and Xiaogang Jia and Jiahong Xue and Gerhard Neumann and Georgia Chalvatzaki},
year = {2026},
eprint = {2602.11934},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.11934},
}
DexFLEX treats fingertip motion commands from teleop or a policy as intent: it proposes short motion chunks, predicts their contact outcome and picks the best; real teleop success rises from 29.2% to 78.3%.DexFLEX 把来自遥操作或策略的指尖运动指令视为意图：生成若干短时动作块、预测其接触结果并择优执行；真实遥操作成功率从 29.2% 提升到 78.3%。
@misc{wan2026dexflex,
title = {{DexFLEX: Contact-Aware Foundation Controller for Command-Guided Dexterity}},
author = {Weikang Wan and Zeyu Lai and Jianyu Hou and Yuchen Zhou and Jiawei Fu and Henrik I Christensen and Hao Su},
year = {2026},
eprint = {2610.14372},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.14372},
}
Low-calibration hand retargeting for teleop: self-supervised fingertip correspondence plus a few human-guided anchors, and a contact classifier to refine pinch poses; tested on several dexterous hands.面向遥操作的灵巧手动作重定向，几乎无需标定：自监督学习指尖对应关系，再用少量人工引导的锚点约束映射，并以接触分类器细化捏取姿态；在多种灵巧手上验证。
@misc{wang2026anydexrt,
title = {{AnyDexRT: Calibration-Efficient Dexterous Hand Retargeting with Few-Shot Human Guidance}},
author = {Chenxi Wang and Ying Feng and Hongjie Fang and Shangning Xia and Lixin Yang and Chuan Wen and Cewu Lu},
year = {2026},
eprint = {2607.08341},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.08341},
}
Chooses the initial grasp for large-object rearrangement by distilling a multi-grasp PPO critic into a small utility network that ranks grasps before contact; reports 89.8% vs 79.9% for the best fixed-grasp policy.为大物体重排选择初始抓取：把多抓取 PPO 策略的价值函数蒸馏成轻量效用网络，在接触前为候选抓取排序；成功率 89.8%，最佳固定抓取策略为 79.9%。
@misc{tu2026grasp,
title = {{Grasp-TAGS: Task-Aware Grasp Selection via Critic Distillation for Prehensile Loco-Manipulation}},
author = {Xun Tu and Minghao Zou and Karthik Desingh},
year = {2026},
eprint = {2611.09864},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09864},
}
Drops an obstacle into a fixed grasp's approach path and lets a planner relabel it: eight feasibility predictors fall from 0.92–0.996 AUROC to chance, so they ignore collision geometry. Labels and pipeline released.固定抓取位姿、在接近路径中插入障碍物并由规划器重新标注：八种可行性预测器的 AUROC 从 0.92–0.996 跌至随机水平，说明它们并未利用碰撞几何。发布标签与流程。
One shared object-pose reward, varied training scenes: RL experts discover regrasping, in-hand adjustment and handover on their own, then distil into one policy. Reports 61 of 80 real trials with unseen objects.只用统一的物体位姿奖励并变换训练环境，强化学习专家自行学会重抓、手内调整与双手交接，再蒸馏成单一策略。报告在未见物体上真实实验 80 次成功 61 次。
@misc{yan2026simduoreal,
title = {{SimDuoReal: Object-Centric Bimanual Policy for Emergent Dexterity}},
author = {Qiyang Yan and Yulin Liu and Akash Sharma and Haozhi Qi and Guanya Shi and C. Karen Liu},
year = {2026},
eprint = {2611.10410},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10410},
}
Trains a world model of 3D hand-point and scene motion, then aligns policy latents to it; at test time only the RGB policy runs, with no depth or future frames. Reports 49.4% → 60.5% on 11 DexJoCo tasks.先训练预测 3D 手部点与场景变化的世界模型，再把策略潜表示对齐到它；部署时只保留 RGB 策略，无需深度或未来预测。在 11 个 DexJoCo 任务上成功率从 49.4% 提升到 60.5%。
@misc{chen2026dexforesight,
title = {{DexForesight: Grounding Dexterous Policies with Implicit Spatial Foresight}},
author = {Zhongxi Chen and Yi Yang and Xinyi Chen and Mingzhe Ye and Yaping Li and Zhihong Liu and Congsheng Xu and Haonan Wang and Zhijie Deng},
year = {2026},
eprint = {2611.11166},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11166},
}
Review (Annual Review of Control, Robotics, and Autonomous Systems) tracing tactile manipulation from contact-mechanics models to learned policies: sensing basics, state estimation, IL, RL, VLAs, predictive models.综述（Annual Review of Control, Robotics, and Autonomous Systems 录用稿）：从接触力学建模到数据驱动学习，梳理触觉操作的发展，涵盖传感基础、状态表征、模仿学习、强化学习、VLA 与预测模型。
A compliant fingertip under $50 gives a dense 12×32 normal and shear map at 60 Hz. Visuo-tactile policies beat vision-only baselines by 46.7 points on average across scooping, pouring and test-tube tasks.每个成本低于 50 美元的柔顺指尖，以 60 Hz 输出 12×32 的法向与剪切稠密响应。在舀取、倾倒和试管任务上，视触觉策略平均比纯视觉基线高 46.7 个百分点。
@misc{binglin2026panotac,
title = {{PanoTac: A Dense Three-Axis Tactile Sensor for Robust Touch-Centric Manipulation}},
author = {Binglin and Wang and Xiaole and Jiang and Zicheng Teng and Xinqi Liu and Yunzhu Li and Lerrel Pinto and Zichen Jeff Cui and Binghao Huang},
year = {2026},
eprint = {2611.03520},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.03520},
}
Plugs several pretrained tactile foundation models into Diffusion Policy and compares them in one setup, including fine-tuning and mid-training strategies, with over 5,000 real-world rollouts.把多种预训练触觉基础模型接入 Diffusion Policy，在统一设置下比较，并考察编码器微调与任务专用中训练策略，含 5000 多次真实机器人回合。
@misc{murooka2026evaluating,
title = {{Evaluating Tactile Foundation Models for End-to-End Imitation Learning in Contact-Rich Manipulation}},
author = {Masaki Murooka and Ryoichi Nakajo and Tomohiro Motoda and Koshi Makihara and Hanbit Oh and Keisuke Shirai and Yukiyasu Domae},
year = {2026},
eprint = {2611.00957},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.00957},
}
Keeps a vision-based flow-matching policy and learns to steer its velocity field from tactile forecasts and live feedback, so touch can correct actions. Reports gains over baselines in six sim and six real tasks.保留视觉流匹配策略，用触觉预测与实时反馈学习对其速度场的残差引导，让触觉能修正动作。报告在六个仿真和六个真实任务上优于基线。
@misc{chen2026tf,
title = {{$TF^2$-Steering: Action Refinement with Tactile Forecasts and Feedback via Velocity-Field Steering}},
author = {Yiming Chen and Xiaofei Wei and Chun Gu and Wei Xiao and Peidong Liu and Li Zhang},
year = {2026},
eprint = {2611.01483},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.01483},
}
VISTA: equivariant visuotactile diffusion policy that fuses touch into spherical visual tokens and rotates them with the gripper pose, for more data-efficient contact-rich imitation.VISTA：等变视触觉扩散策略，把触觉融合进球面视觉 token 并随末端姿态旋转，提升富接触模仿学习的数据效率。
@misc{wong2026equivariant,
title = {{Equivariant Visual-Tactile Diffusion Policy for Contact-Rich Manipulation}},
author = {Lik Hang Kenny Wong and Yiyao Ma and Xiu-Shen Wei and Zelong Tan and Zhuheng Song and Dongsheng Xie and Kai Chen and Qi Dou},
year = {2026},
eprint = {2610.03333},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.03333},
}
Tactile simulator for the BrainCo Revo3 hand modelling 285 piezoresistive taxels and five visuotactile fingertips (pressure, depth, RGB, marker motion); up to 4,096 parallel envs on one RTX 5090.为 BrainCo Revo3 灵巧手构建的触觉仿真器，同时模拟 285 个压阻触点和五个视触觉指尖（压力、深度、RGB、标记点位移）；单张 RTX 5090 最多 4096 个并行环境。
@misc{zhang2026revosim,
title = {{RevoSim: Scalable Multimodal Tactile Simulation for Dexterous Manipulation}},
author = {Ke Zhang and Lixin Xu and Ziyi Wang and Xiyue Dong and Jie Tan and Lianghu Shao and Xianpeng Wang and Chuanyu Li and Renjing Xu},
year = {2026},
eprint = {2611.08939},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08939},
}
Learns visuo-tactile latents end-to-end inside a diffusion world-action model not frozen encoders; multi-task cross-modal losses avoid feature collapse. Reports up to 50% gains on real bolt screwing and nut mating.在扩散式世界-动作模型中端到端学习视触觉潜表示，而非使用冻结的预训练编码器；借助跨模态多任务学习避免特征坍缩。在真实拧螺栓、螺母配合等任务上报告最高 50% 的提升。
HumanVerse, 500 h of egocentric human video with body and hand motion, aligned to the Unitree G1 for mid-training a humanoid model; reports 90% on SIMPLE and 62.5% on real loco-manipulation tasks.构建 500 小时的第一视角人类数据集 HumanVerse（含同步的身体与手部动作），对齐到 Unitree G1 后用于人形基础模型 λ0 的中期训练；论文报告 SIMPLE 上成功率 90%、真实移动操作任务 62.5%，人类数据越多下游表现越好。
@misc{team2026scaling,
title = {{$λ_0$: Scaling Open-World Egocentric Human Data for Humanoid Loco-Manipulation}},
author = {Lambda-humanoid Team and : and Chongyang Xu and Jin Chen and Jinhui Ye and Yilun Chen and Yiming Jiang and Zhao Wu and Hu Wei and Lin Qu and Shifeng Zhang and Yisheng He and Yuming Jiang and Zenan Meng and Zheng Wu and Mu Xu and Steven Hoi},
year = {2026},
eprint = {2610.18950},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.18950},
}
A quadruped solves a giant Rubik's Cube with its feet: onboard RGB-D estimates cube pose and colours, a planner sequences skills, and RL whole-body skills turn layers. Reports three full autonomous solves.四足机器人只用脚完成巨型魔方：机载 RGB-D 估计魔方位姿与颜色状态，规划器把解法转成技能序列，强化学习训练的全身技能负责翻转魔方、一脚固定一脚推动转层。报告完成三次完全自主求解。
@misc{amiet2026solving,
title = {{Solving a Giant Rubik's Cube with a Quadruped Robot}},
author = {Lukas Amiet and Gabriel Sánchez-Blanco Repgen and Mayank Mittal and Victor Klemm and Andrei Cramariuc and Marco Hutter and Cesar Cadena},
year = {2026},
eprint = {2610.24610},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.24610},
}
FLoRe: robot-free demos from force-sensing UMI-style grippers + body trackers train a diffusion policy that outputs per-hand force commands for a compliant whole-body controller; reports 80% vs. 26.7% for position-only.FLoRe：用带力传感的 UMI 式手持夹爪与人体追踪器采集无机器人示教，训练扩散策略输出每只手的力指令，由柔顺全身控制器执行；报告在三个接触密集任务上成功率 80%，仅位置控制为 26.7%。
@misc{li2026learning,
title = {{Learning Force-Reactive Humanoid Loco-Manipulation}},
author = {Sihang Li and Xiaofeng Guo and Jinkun Cao and Charlie Cheng and Ruihan Yang and Chen Feng and Guanya Shi and Haozhi Qi},
year = {2026},
eprint = {2610.35862},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.35862},
}
Real-to-sim-to-real: smartphone scans become physics-ready sim scenes, whole-body motions are corrected by rollouts, giving 4,800 demos; a sim-only GR00T N1.7 on Unitree G1 reports 66.7% vs. 46.7% trained on real data.真实到仿真再到真实：用手机拍摄重建可物理仿真的场景，并根据仿真回放修正全身动作，生成 4800 条示教；仅用仿真数据训练的 GR00T N1.7 在宇树 G1 上报告平均成功率 66.7%，真实数据训练为 46.7%。
@misc{yang2026zero,
title = {{Zero-HLM: Generalizable Humanoid Loco-Manipulation without Real Robot Data}},
author = {Dianyi Yang and Mengya Liu and Zhumei Wang and Yu Liu and Xiaochen Zhang and Yixuan Li and Le Ma and Yuhang Yang and Weiqi Huang and Wei Liang and Baoxiong Jia and Siyuan Huang},
year = {2026},
eprint = {2610.35075},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.35075},
}
Whole-body humanoid teleoperation from a compact target (root height and velocity + 4 keypoint poses, current frame only): ~20 ms headset-to-policy latency; a Unitree G1 catches a tossed ball.只用紧凑参考目标（根高度、根速度与 4 个稀疏关键点的当前帧位姿）实现人形机器人全身遥操作：头显到策略输出延迟约 20 ms，Unitree G1 能接住抛来的球并完成移动操作。
@misc{cao2026humit,
title = {{HuMiT: Low-Latency Whole-Body Humanoid Teleoperation via Compact Reference Tracking}},
author = {Liu Cao and Botian Xu and Ruoqu Chen and Jiajun Liu and Mingzhi Pei and Zihao Wang and Zhongyu Li and Mengdi Xu},
year = {2026},
eprint = {2610.29903},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29903},
}
FIBER plans in the latent space of a frozen behavioral foundation model: skills are projected into it, one perceptive flow planner is trained offline. Reports 42.0% mean sim success (latent replay: 11.6%); real G1.FIBER 以冻结的行为基础模型（BFM）潜在空间作为统一规划表示，把各任务专用策略投影进该空间，再离线训练单一感知流匹配规划器；仿真中搬箱、开门、抓取平均成功率 42.0%（潜在回放基线 11.6%），并在真实 Unitree G1 上验证。
@misc{lin2026action,
title = {{Action-Aligned Latent Planning for Versatile Humanoid Loco-Manipulation}},
author = {Yutang Lin and Guanya Shi and C. Karen Liu},
year = {2026},
eprint = {2610.24675},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.24675},
}
When a quadruped manipulator keeps failing, a coding agent writes an RL problem to learn the missing whole-body skill, adds it to a library and retries; transfers to a real robot (18/20 rack retrieval).四足机械臂反复失败时，由编码智能体判断缺少哪种全身控制能力，自动编写强化学习问题（奖励、课程、终止条件）训练新控制器并加入技能库再重试；习得扶墙抬升等技能，仿真五项超出初始能力的任务成功率 98.4–100%，无需真机训练即迁移到实体机器人（货架取物 18/20）。
@misc{liu2026elevate,
title = {{ELEVATE: Agentic Controller Acquisition from Task Failure for Quadrupedal Manipulation}},
author = {Shiqi Liu and Radu Corcodel and Ding Zhao},
year = {2026},
eprint = {2610.38388},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.38388},
}
Trained in simulation with no real demos. Controller-in-the-loop filtering rejects 73% of kinematically valid grasps that would destabilize the robot; reports 87.5% real-world success on a multi-level shelf task.完全在仿真中训练，无真实示范。把控制器放进回路筛选数据，剔除 73% 运动学可行但会使机器人失稳的抓取；作者报告多层货架任务真实成功率 87.5%。
@misc{peng2026hug,
title = {{HuG: Learning Humanoid Whole-Body Affordance-Aware Grasping with Active Perception}},
author = {Xuanbin Peng and Lai Wei and Yuze Zhai and Qingyuan Yu and Keyi Wang and Jianglong Ye and Xiaolong Wang},
year = {2026},
eprint = {2611.01659},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.01659},
}
Turns diverse human videos into deployable humanoid loco-manipulation skills without task-specific rewards, despite occlusion and retargeting errors in the priors.把多样的人类视频转化为可部署的人形移动操作技能，无需任务专用奖励，能应对先验中的遮挡与重定向误差。
@misc{wu2026sugar,
title = {{SUGAR: A Scalable Human-Video-Driven Generalizable Humanoid Loco-Manipulation Learning Framework}},
author = {Tianshu Wu and Xiangqi Kong and Yue Chen and Qize Yu and Hang Ye and Jia Li and Yizhou Wang and Hao Dong},
year = {2026},
eprint = {2605.20373},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2605.20373},
}
Trains the low-level humanoid controller by motion imitation with the object physically in the scene, then a high-level policy on top; on a Unitree G1 it carries varied boxes and recovers from drops.让人形底层控制器在场景中真实带着物体做动作模仿训练，再在其上训练高层策略；在宇树 G1 上可搬运不同箱子，并能从掉落中恢复。
@misc{sleiman2026hlo,
title = {{HLO-AMP: Hierarchical Loco-Manipulation from Object-Aware Motion Priors}},
author = {Jean-Pierre Sleiman and Bilal Hammoud and He Li and Jonathan Foster and Sylvain Bertrand and Jessica K. Hodgins and Farbod Farshidian},
year = {2026},
eprint = {2611.05113},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05113},
}
Adds an explicit 3-D force command per contact body to motion tracking, set online by a planner. On hardware it pushes and kicks boxes up to 20 kg where a force-blind baseline fails, and adapts to added loads.在动作跟踪中为每个接触部位加入显式三维力指令，由高层规划器在线给出。实机可推、踢最重 20 kg 的箱子（不感知力的基线会失败），并能适应任务中途增加的负载。
@misc{zhao2026ctrl,
title = {{Ctrl-F: Force as an Interface for Humanoid Loco-Manipulation}},
author = {Mengjie Zhao and Yuxin Chen and Yiyang Shao and Chuye Hong and Jiakang Chen and Koushil Sreenath and Masayoshi Tomizuka},
year = {2026},
eprint = {2611.05507},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05507},
}
PRISM turns a few real videos into hundreds of counterfactual human-object clips via video generation, rebuilds them in sim, and trains one humanoid carrying policy that handles unseen objects with no real fine-tuning.PRISM 用视频生成把少量真实视频扩增为数百段反事实人-物交互视频，在仿真中重建并训练单一人形搬运策略，无需真机微调即可泛化到未见物体。
@misc{wang2026counterfactual,
title = {{Counterfactual Video Generation Enables Scalable Humanoid Loco-Manipulation}},
author = {Zihan Wang and Zhen Wu and Pieter Abbeel and Rocky Duan and Jitendra Malik and Carmelo Sferrazza and C. Karen Liu and Guanya Shi and Angjoo Kanazawa},
year = {2026},
eprint = {2609.38172},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.38172},
}
Modular humanoid grasping: large vision models handle the scene, a sim-trained end-effector tracker with IK and a learned forward model handles the arm. Reports 2.44 cm tracking error, 5.5x better than prior work.模块化人形抓取：大视觉模型负责场景理解，仿真训练的末端跟踪策略结合 IK 与学习的正运动学模型负责执行。报告末端跟踪误差 2.44 厘米，比先前最好方法低 5.5 倍。
@misc{dong2026hero,
title = {{HERO: Learning Humanoid End-Effector Control for Visual Whole-Body Open-Vocabulary Object Grasping}},
author = {Runpei Dong and Ziyan Li and Arjun Gupta and Xialin He and Saurabh Gupta},
year = {2026},
eprint = {2602.16705},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.16705},
}
Follow-up to SUGAR: synthesizes robot and object reference motions from reusable task templates, then trains visual policies with one mimic reward and teacher distillation; 91.0% mean sim success on five tasks.SUGAR 的后续：用可复用的任务模板合成机器人与物体参考动作，再以统一的模仿奖励和教师蒸馏训练视觉策略；五项任务仿真平均成功率 91.0%，并完成真机迁移。
@misc{wu2026sugarv,
title = {{SUGARv2: Learning Diverse Visual Humanoid Loco-Manipulation Skills from Synthesized Interaction References}},
author = {Tianshu Wu and Tianxiang Gui and Yue Chen and Tingxu Liu and Ziye Peng and Guangjie Yang and Xiangqi Kong and Ye Shi and Jingya Wang and Hao Dong},
year = {2026},
eprint = {2611.09266},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09266},
}
World action model that jointly predicts video, motion and contact wrench, outputting time-aligned motion–wrench chunks as control references (hybrid motion-force style) for force-sensitive manipulation.一个同时预测未来视频、运动与接触力旋量的世界动作模型，借鉴力位混合控制思路，输出时间对齐的运动-力旋量片段作为控制参考，面向仅靠视觉无法感知接触与负载的力敏感操作任务。
@misc{ding2026cw,
title = {{CW-WAM: A Structured Contact-Wrench World Action Model for Force-Sensitive Manipulation}},
author = {Houzhu Ding and Xinda Qi and Luqi Wang and Ming Yang},
year = {2026},
eprint = {2610.30117},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.30117},
}
AW-0 jointly generates code, future video and actions in one mixture-of-transformers; code structures planning and becomes reusable skills. Reports gains over VLA and WAM baselines on RoboTwin 2.0 and real two-arm tasks.AW-0 在同一个混合 Transformer 中联合生成代码、未来视频和机器人动作三路输出：代码既组织多步预测与控制，又沉淀为可复用技能库；报告在 RoboTwin 2.0 和真机双臂任务上优于 VLA 与 WAM 基线。
@misc{liu2026agentic,
title = {{Agentic World Action Models}},
author = {Zhuoyang Liu and Dantong Niu and Botong Zhang and Boning Shao and Jing Wang and Ruijie Zheng and Yuvan Sharma and Runyu Lu and Jiaming Liu and Renrui Zhang and Roei Herzig and Yutong Bai and Ken Goldberg and Danfei Xu and Yuke Zhu and Guanzhi Wang and Linxi "Jim" Fan and Trevor Darrell},
year = {2026},
eprint = {2610.30136},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.30136},
}
Dexterous-manipulation world model that first predicts object motion and deformation in metric 3D from a hand trajectory, then renders video. Reports 27–40% lower moved-point error than PointWorld.灵巧操作世界模型：先根据手部轨迹在度量 3D 空间中显式预测物体运动与形变，再据此渲染视频。报告在 HOT3D、OakInk2、PhysTwin 上运动点误差比 PointWorld 低 27–40%，两步因果生成大幅提速。
@misc{lee2026dexplicit,
title = {{DeXplicit: Making Physical State Explicit in Dexterous Manipulation World Models}},
author = {Adam Lee and Mutian Xu and Yukang Cao and Shulin Tian and Ziwei Liu},
year = {2026},
eprint = {2610.30781},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2610.30781},
}
World action model for moving targets: forecasts the object's future masks, steers a video model with a Trajectory ControlNet, and distills the pipeline to one-step video-action inference to cut latency.面向动态目标的世界动作模型：先预测目标物体的未来掩码轨迹，用 Trajectory ControlNet 引导视频模型生成与轨迹一致的未来，再通过解耦蒸馏实现一步视频-动作推理以降低延迟。
@misc{geng2026trajwam,
title = {{TrajWAM: An Efficient Trajectory-Guided World Action Model for Dynamic Robotic Manipulation}},
author = {Haohan Geng and Sha Zhang and Weishuo Sun and Guanting Ye and Wenhao Yu and Yuang Xie and Qicheng Zeng and Peng Ye and Jiajun Deng and Jianmin Ji},
year = {2026},
eprint = {2610.25763},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25763},
}
Compact (<100M) world-action model: predicts frozen-DINOv3 tokens at two future horizons and decodes action chunks, with no robot or video pretraining. Reports beating billion-parameter WAMs on RoboTwin 2.0.参数不到 1 亿的紧凑世界动作模型：在冻结 DINOv3 特征空间中预测两个未来时刻的 token 并直接回归动作块，无需机器人或视频预训练；报告在 RoboTwin 2.0 上超过十亿参数级 WAM。
@misc{liu2026gawm,
title = {{GAWM: A Compact Latent World-Action Model for Multi-Task Robot Manipulation}},
author = {Yang Liu and Xiaoli Liu and Liuzhao Li and Yuduo Song and Muyuan Liu and Xiang Gao},
year = {2026},
eprint = {2610.37091},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.37091},
}
Turns a pretrained video diffusion transformer into a policy by encoding each action step as one patch in the video latent space; at test time it outputs an action chunk in one DDIM step (84 ms), no video generation.把预训练视频扩散 Transformer 改造成策略：每个时间步的动作编码为视频潜空间中的一个 patch，联合训练视频、视频+动作与纯动作三种扩散目标；测试时一步 DDIM 生成动作块（84 毫秒），无需生成视频。
@misc{collins2026tapestry,
title = {{TAPESTRY: A Single Backbone for Video Generation and Robot Control}},
author = {Jeremy A. Collins and Seungjae Lee and Rachanon Wachakorn and Krishnan Srinivasan and Animesh Garg and Vitor Campagnolo Guizilini and Paarth Shah},
year = {2026},
eprint = {2610.36823},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36823},
}
Adds future 3D point-motion prediction as an auxiliary task to a world action model (video/motion/action MoT); inference stays one visual pass. Reports gains over Fast-WAM with few demos and under shift.在世界动作模型中加入未来三维点运动预测作为辅助目标（视频/运动/动作三专家 MoT），推理仍只需一次视觉前向；报告在少量示范和分布偏移下明显优于 Fast-WAM（RoboCasa-GR1、LIBERO、LIBERO-Plus）。
@misc{hu2026trace,
title = {{Trace-WAM: Learning to Act via 3D Motion Prediction for World Action Models}},
author = {Xu Hu and Changjie Wu and Linzhe Shi and Hang Zhang and Chen Yang and Lingjun Zhang and Guowen Zhang and Mu Xu and Junran Peng and Qing Li and Zhaoxiang Zhang},
year = {2026},
eprint = {2610.36840},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36840},
}
@misc{kim2026adapting,
title = {{Adapting Video Models for Robot Control via Motion-guided Asymmetric Prediction}},
author = {Taeyoung Kim and Junyoung Sung and Heeseung Kwon and Minseop Kim and Jinwoo Shin},
year = {2026},
eprint = {2610.36624},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2610.36624},
}
Do world action models need video generation? ImageWAM conditions a flow-matching action expert on image-editing KV caches instead. Reports beating VLA baselines at 1/6 the FLOPs and 1/4 the latency of video WAMs.追问世界动作模型是否真需要视频生成：ImageWAM 改用图像编辑模型去噪时的 KV 缓存来条件化流匹配动作专家；报告优于 VLA 基线，计算量为视频 WAM 的 1/6、延迟为 1/4。
@misc{zhang2026imagewam,
title = {{ImageWAM: Do World Action Models Really Need Video Generation, or Just Image Editing?}},
author = {Yuyang Zhang and Wenyao Zhang and Zekun Qi and He Zhang and Haitao Lin and Jingbo Zhang and Yao Mu and Xiaokang Yang and Wenjun Zeng and Xin Jin},
year = {2026},
eprint = {2606.19531},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2606.19531},
}
Lightweight world action model: compact video backbone with future supervision in a downsampled latent, and an action expert that pools several layers to predict chunks in one pass. Reports 0.44B trainable params, 72 ms.轻量世界动作模型：紧凑视频主干在降采样潜空间中做未来监督，动作专家融合多层特征一次前向预测动作块；报告可训练参数仅 0.44B，推理延迟 72 毫秒。
@misc{li2026light,
title = {{Light-WAM: Efficient World Action Models with State-Fusion Action Decoding}},
author = {Ziang Li and Dongzhou Cheng and Yibin Wang and Shiyue Wang and Xiaoyang Xu and Lingxuan Weng and Juan Wang and Jiaqi Wang},
year = {2026},
eprint = {2606.08242},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2606.08242},
}
World model that predicts 3D traces of key points (objects, tools, hands) instead of pixels or actions, pretrained from video without action labels. Trace-conditioned policies are reported competitive with π0-style VLAs.不预测像素也不预测动作，而是预测物体、工具、手等关键点的 3D 轨迹，可用无动作标签的视频预训练。报告以轨迹为条件的策略可与 π0 类 VLA 相当。
@misc{lee2026scalable,
title = {{$μ_0$: A Scalable 3D Interaction-Trace World Model}},
author = {Seungjae Lee and Yoonkyo Jung and Jusuk Lee and Jonghun Shin and Amir Hossein Shahidzadeh and Yao-Chih Lee and H. Jin Kim and Jia-Bin Huang and Furong Huang},
year = {2026},
eprint = {2606.13769},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.13769},
}
Generative world model of 3D particle and rigid-body motion covering rigid, deformable, cloth and granular materials. Reports 31% lower error than prior learned dynamics; planning searches its latent noise.生成式 3D 粒子与刚体运动世界模型，统一覆盖刚体、可变形体、布料与颗粒材料。报告预测误差比此前最好的学习动力学低 31%，并通过在潜噪声空间搜索来做规划。
@misc{ren2026gpwm,
title = {{GPWM: Modeling the World as 3D Particle Motion}},
author = {Robert Ren and Matthew Bronars and He Zhu and Yikai Tang and Jianing Zhang and Sergey Zakharov and Pavel Tokmakov and Katerina Fragkiadaki},
year = {2026},
eprint = {2611.08338},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08338},
}
World action model that predicts oriented bounding-box trajectories of task objects instead of future video frames: much cheaper, and reported to predict actions better.世界动作模型不再预测未来视频帧，而是预测任务相关物体的有向包围盒轨迹：计算量大幅下降，并报告动作预测更好。
@misc{yu2026box,
title = {{Box-WAM: Box Trajectories as Compact State Representations for World Action Models}},
author = {Qinxi Yu and Wenzhou Ding and Gloria X. Wang and Albert J. Zhai and Shenlong Wang},
year = {2026},
eprint = {2610.17526},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17526},
}
Represents object motion as dense 3D trajectories over the whole surface, occluded parts included, for rigid and deformable objects. Adds a 1,210-hour flow dataset and a world model predicting it from image + task.把物体运动表示为覆盖整个表面（含遮挡部分）的稠密 3D 轨迹，适用于刚体与可形变物体。发布 1210 小时的全物体流数据集，并训练从图像和任务描述预测该流的世界模型。
@misc{chen2026objectworld,
title = {{ObjectWorld: Scaling Object-Centric World Models with A Large-Scale 3D Full-Object Flow Dataset}},
author = {Haonan Chen and Sicheng Isabella Lu and Haitian Li and Jiajin Wu and Hexiang Wei and Jiawei Zhang and Xin Fei and Yiwen Hou and Chenrui Tie and Chongkai Gao and Sheng Wang and Lin Shao},
year = {2026},
eprint = {2611.11085},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11085},
}
Finds a VLA at 80% on atomic skills drops to 19% when composed into mobile manipulation, diagnoses why, and builds an agentic system around it.发现 VLA 在原子技能上 80% 成功率、组合成移动操作后仅 19%，分析失败原因并据此构建智能体式系统。
@misc{wang2026cracking,
title = {{Cracking Mobile Manipulation: From Failure Diagnosis to Agentic Robot System}},
author = {Zixuan Wang and Jia-Ning Li and Yinzhou Tang and Yanzhao Yu and Biao Wu and Meng Fang and Xin Li and Yiwu Zhong},
year = {2026},
eprint = {2610.19600},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.19600},
}
HiRe-MoMa fine-tunes a frozen mobile-manipulation policy in the real world by learning residual corrections, with only occasional human guidance.HiRe-MoMa 在真实环境中通过学习残差修正来微调冻结的移动操作策略，只需偶尔的人类引导。
@misc{gong2026hierarchical,
title = {{Hierarchical Residual Policy Learning for Real-World Mobile Manipulation with Sparse Human Guidance}},
author = {Zhefei Gong and Hefeifei Jiang and Zehao Wang and Jiachen Li},
year = {2026},
eprint = {2610.20607},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.20607},
}
Navigates cluttered spaces by deciding, with VLM affordance reasoning, whether to swipe, pick up or avoid each blocking object. Zero-shot; reports +27 points over the best baseline on a real mobile manipulator.在杂乱环境中导航：用 VLM 推理每个挡路物体该推开、拿起还是绕开。零样本，报告在真实移动操作机器人上比最强基线高 27 个百分点。
@misc{llontop2026messynav,
title = {{MessyNav: Zero-Shot Navigation via Manipulation in Messy Environments}},
author = {Kenneth Llontop and Cherie Ho and Dylan Zhou and Anuva Banwasi and Priya Sundaresan and Carlota Parés-Morlans and Jeannette Bohg},
year = {2026},
eprint = {2611.07417},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.07417},
}
Persistent memory for mobile manipulators: a 3D scene graph updated with what the robot learns by acting (locked cabinets, object locations). Reports 80% progress over a 25-task, 3-hour run.移动操作机器人的持久记忆：在三维场景图中记录通过交互学到的知识（如柜子上锁、物体位置）。报告在连续 25 个任务、3 小时的仿真中任务进度达 80%。
@misc{banwasi2026messymem,
title = {{MessyMem: Learning-from-Doing Memory for Mobile Manipulation}},
author = {Anuva Banwasi and William Muckelroy and Priya Sundaresan and Linfeng Zhao and Jeannette Bohg and Cherie Ho},
year = {2026},
eprint = {2609.15976},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.15976},
}
Fine-tunes Qwen3.5-9B on ~1.5M grounding samples auto-built from robot recordings plus other visual data; reports a 75.41 average over seven grounding benchmarks vs 73.68 for GPT-6 Astra.用自动标注的机器人操作录像加多来源视觉数据构建约 150 万条可验证的定位监督，微调 Qwen3.5-9B；报告在七个公开视觉定位基准上平均 75.41 分，GPT-6 Astra 为 73.68。
@misc{wang2026ace,
title = {{ACE-Brain-0.6: Scaling Spatial Grounding for Physical Foundation Models}},
author = {Yinhe Wang and Tao Tao and Zexin He and Shuang Luo and Wei Mao and Yao Mu and Feng Pan and Hanming Deng and Lewei Lu and Zhi Hou and Xingang Pan and Dahua Lin and Dacheng Tao and Xiaogang Wang},
year = {2026},
eprint = {2610.27040},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.27040},
}
MoMa-SG builds 3D scene graphs that encode how objects articulate, from RGB-D sequences, and adds the Arti4D-Semantic dataset; demonstrated on a quadruped and a mobile manipulator.MoMa-SG 从 RGB-D 序列构建包含物体铰接方式的 3D 场景图，并发布 Arti4D-Semantic 数据集；在四足和移动操作机器人上演示。
@misc{buchner2026articulated,
title = {{Articulated 3D Scene Graphs for Open-World Mobile Manipulation}},
author = {Martin Büchner and Adrian Röfer and Tim Engelbracht and Tim Welschehold and Zuria Bauer and Hermann Blum and Marc Pollefeys and Abhinav Valada},
year = {2026},
eprint = {2602.16356},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.16356},
}
Makes dense VLM features spatially consistent by rebuilding each patch from language-labelled visual prototypes; improves open-vocabulary retrieval and 3D mapping. Also ships a compact 25M dense VLM for robots.通过由带语言描述的视觉原型重建每个图像块，让稠密 VLM 特征在空间上更一致，提升开放词汇检索与三维建图；并提供 25M 参数的小型稠密 VLM。
@misc{schwaiger2026resireg,
title = {{ReSiReg: Towards Spatially Consistent Semantics in Language-Conditioned Robotic Tasks}},
author = {Simon Schwaiger and David Seyser and Alessandro Scherl and Wilfried Wöber and Gerald Steinbauer-Wagner},
year = {2026},
eprint = {2606.19088},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.19088},
}
An open-source ROS 2 multimode controller that switches control modes and triggers safety reflexes in under a millisecond, without stopping the task; shown on a Franka arm.开源的 ROS 2 多模式控制器，可在亚毫秒内切换控制模式并触发安全反射而无需中断任务；在 Franka 机械臂上验证。
Signed friction distance as a task-adaptive slip margin for bimanual tilting about a pivot; a sequence of SOCPs gives optimal contact forces. Real tests show adaptive margins beat fixed or no margins.以带符号摩擦距离度量防滑裕度，并在双臂绕支点倾转物体时按任务需求自适应调整；一系列二阶锥规划给出全局最优接触力。真机实验显示自适应裕度优于固定裕度和无裕度。
@misc{lee2026keeping,
title = {{Keeping Contact: Task-Adaptive Friction Margins for Bimanual Non-Prehensile Manipulation}},
author = {Qian Ying Lee and Lin Yang and Domenico Campolo},
year = {2026},
eprint = {2610.36961},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36961},
}
Continual off-policy RL for VLAs — small actor/critic networks carry over between tasks, distilled on a few kept episodes to avoid forgetting; reports 31–68.5% less real-robot time on new assembly tasks.面向 VLA 的持续离策略强化学习：轻量 actor/critic 在任务间延续，用少量保留的旧任务轨迹做蒸馏以防遗忘。论文报告新的接触式装配任务所需真机训练时间减少 31–68.5%，旧任务几乎不遗忘。
@misc{kim2026accrue,
title = {{ACCRUE: Continual Off-Policy Reinforcement Learning for Robot Foundation Models}},
author = {Changyeon Kim and Kimin Lee and Yashraj Narang and Yijie Guo},
year = {2026},
eprint = {2610.18596},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.18596},
}
TIDE updates a pretrained policy online from human interventions and its own rollouts, with no critic or reward model; reports gains on four real bimanual tasks, each session finishing in under an hour.TIDE 在部署过程中利用人类干预和策略自身的执行数据在线更新预训练策略，无需价值函数或奖励模型；报告在四个真实双臂任务上优于初始策略，每次在线后训练不到一小时完成。
@misc{gu2026training,
title = {{Training in Deployment: Human-in-the-loop Online Post-Training for Robot Manipulation}},
author = {Chun Gu and Xiaofei Wei and Yiming Chen and Bozhou Zhang and Juntong Zhang and Xiatian Zhu and Li Zhang},
year = {2026},
eprint = {2610.29264},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29264},
}
Uses RL as a demo generator, not a controller: a latent-conditioned policy finds several distinct successful solutions, exported as an imitation dataset; reports 78% real-robot success vs. 39% for the best baseline.把强化学习当作示教数据生成器而非部署控制器：以隐变量为条件的策略为同一任务找到多种不同的成功解法，导出为模仿学习数据集；报告真机下游成功率从最强基线的 39% 提升到 78%。
@misc{park2026discodemo,
title = {{DiscoDemo: Discovering Efficient and Diverse Robot Demonstrations for Imitation Learning}},
author = {Minho Park and Kinam Kim and Donghu Kim and Byungkun Lee and Dongyoon Hwang and Yongjae Shin and Junha Hyung and Hojoon Lee and Jaegul Choo},
year = {2026},
eprint = {2610.32093},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.32093},
}
Replay-free lifelong imitation learning: hierarchical dual-routing attention retrieves experts by instruction, then refines by context, plus key regularization against drift. Reports beating replay baselines on LIBERO.无回放的终身模仿学习：分层双路由注意力先按指令检索专家、再按上下文细化，并用键级几何正则抑制表征漂移带来的路由干扰；报告在多个 LIBERO 任务序列上优于无回放和有回放基线。
Fine-tunes diffusion and flow policies with automatic preference labels: rollouts are ranked by progress measured against a few expert demos, no human labels or reward design. Reports beating SFT on long-horizon tasks.用自动生成的偏好标签微调扩散与流匹配策略：以少量专家示教为参照衡量自主回放的顺序进度并据此排序，无需人工标注或手工奖励；报告在长时程操作任务上优于监督微调基线。
@misc{luu2026progress,
title = {{Progress-Guided Preference Generation for Aligning Generative Control Policies}},
author = {Tung M. Luu and Tri Ton and Duc M. Nguyen and Xuan Thanh Nguyen and Nguyen Gia Binh and Quang Tan Nguyen and Vu Duong and Daniel Sonntag and Duy M. H. Nguyen and Vien Anh Ngo and Chang D. Yoo},
year = {2026},
eprint = {2610.25692},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/2610.25692},
}
FLEX lets language corrections edit a robot's task features via add, generalize and mask (not add-only). Reports higher normalized reward than other editing methods in 2D nav, Franka tabletop and a real user study.FLEX 用自然语言纠正来迭代修改机器人的任务特征集合，支持添加、泛化、屏蔽三种操作（而非只能添加）；在二维导航、Franka 桌面操作和真实用户研究中报告比其他特征编辑方法获得更高的归一化奖励。
@misc{lee2026flexible,
title = {{Flexible Language-Guided Representation Editing for Adaptive Robot Behavior}},
author = {Audrey Lee and Nathaniel Dennler and Ilia Sucholutsky and Andreea Bobu},
year = {2026},
eprint = {2610.28893},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.28893},
}
Simulation study varying robot and object mass: RL copes, but behavior cloning does better when the robot is lighter than the object, and more data narrows the gap without closing it. Lower-inertia demos help.在仿真中改变机器人与物体质量：强化学习影响不大，行为克隆则在机器人比物体轻时更可靠，增加数据只能缩小而无法消除差距。采集低惯量示教可部分缓解。
Jointly trains a latent action model, latent policy and action decoder, so plentiful action-free video plus a few labelled robot demos can train a policy. Reports beating prior latent-action baselines.联合训练潜在动作模型、潜在策略和动作解码器，用大量无动作标注的视频加少量带标注示教即可训练策略。报告在仿真和真实任务上优于已有潜在动作方法。
Inverse RL from observation only (videos, no actions or rewards) made practical on a real robot via online world modeling. Reports pick-and-place learned from scratch to 82% in under 40 minutes.仅从观测（视频，无动作和奖励）做逆强化学习，借助在线世界模型在真机上实现。报告不到 40 分钟从零学会抓放，成功率 82%。
@misc{han2026online,
title = {{Online World Modeling Enables Real-World Inverse Reinforcement Learning from Observation}},
author = {Tyler Han and Bat Nemekhbold and Siyang Shen and Rohan Baijal and Richard Ebock and Harine Ravichandiran and Sanghun Jung and Kevin Huang and Byron Boots},
year = {2026},
eprint = {2602.24121},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2602.24121},
}
Composed diffusion policies often learn identical components. A drop-in score-decorrelation loss makes each factor specialize, with no loss in multitask performance.组合式扩散策略的各个分量常常学成相同的副本。即插即用的得分去相关损失让各分量各自专门化，且不损失多任务性能。
@misc{laprevotte2026learning,
title = {{Learning to Be Different: Component Specialization for Compositional Diffusion Policies}},
author = {Alec Laprevotte and Haonan Chen and Xiaoshen Han and Yilun Du},
year = {2026},
eprint = {2610.27194},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.27194},
}
Scores how much a small action error at each demo state would hurt the outcome, using an adversarial disturbance policy. The scores reweight behavior cloning and select smaller, higher-value demo subsets.用对抗扰动策略衡量每个示教状态上的小动作误差会对结果造成多大损害。该分数可用于给行为克隆加权，并挑选更小但价值更高的示教子集。
@misc{ciftci2026where,
title = {{Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning}},
author = {Yusuf Umut Ciftci and Joonwon Kang and Ajay Suresha Sathya and Somil Bansal},
year = {2026},
eprint = {2610.17709},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/2610.17709},
}
Learns how a known symmetry acts on pixel observations and actions from offline data, then uses it for frame-averaged behavior cloning. Reports average success 80.5% to 85.1% on four Robosuite tasks.从离线交互数据中学习已知对称群如何同时作用于像素观测和动作，再用于帧平均行为克隆。报告在四个 Robosuite 任务上平均成功率从 80.5% 升到 85.1%。
@misc{liu2026mirror,
title = {{MIRROR: Reflecting Hidden Symmetries from Offline Robot Interaction Data}},
author = {Shiqi Liu and Yihang Yao and Peide Huang and Mengdi Xu and Ding Zhao},
year = {2026},
eprint = {2610.17422},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.17422},
}
@misc{houssaini2026dexsuite,
title = {{DexSuite: An All-in-One Framework for Dexterous Manipulation Across Arms, Hands, and Tasks}},
author = {Anas Houssaini and Hugo He and Junming Shi and Amin Abyaneh and Ricardo Chahine and Shuo Wen and Mohamad H. Danesh and Charles-Étienne Fréchette and Mariana Sosa Guzman and Mathias Desrochers and Rayan Houssaini and Adam Landa and Theophile Soulie and Charlotte Morissette and Annie Tims and Martino Russi and Jonathan Lussier and Gregory Dudek and Doina Precup and Julen Urain and Hsiu-Chin Lin and David Meger},
year = {2026},
eprint = {2610.37799},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.37799},
}
PGRD simulates deformables with an optimizable spring-mass backbone plus learned residuals, reporting better accuracy than either alone; used for MPC and action-conditioned video prediction.PGRD 以可优化的弹簧质点模型为骨架，加上学习得到的残差来仿真可变形物体，报告精度优于单用任一方法；可用于 MPC 和动作条件视频预测。
Turns diagnosed policy failures into new simulated scenes and shaped training signals, so generator and policy improve together; on LIBERO-Pro, reports success rising from 25% to 84% under position perturbations.把诊断出的策略失败转成新的仿真场景和塑形训练信号，让生成器与策略共同进化；在 LIBERO-Pro 上报告位置扰动下成功率从 25% 升至 84%。
@misc{yu2026coda,
title = {{CoDA: Co-Evolution of Data Engines and Action Policies}},
author = {Tianjiao Yu and Zhilin Ou and Yifan Shen and Xinzhuo Li and Yuanzhe Liu and Onkar Kishor Susladkar and Fan Wang and Ismini Lourentzou},
year = {2026},
eprint = {2610.22362},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.22362},
}
100 h of wearable-captured human loco-manipulation (mostly professional work) aligning egocentric video, whole-body and finger motion, touch and step-level text; reports 71.9% lower body error than baselines.用可穿戴设备采集的 100 小时人类移动操作数据，侧重职业场景，对齐第一视角视频、全身与手指动作、触觉和步骤级文字描述；其补全全身动作的方法报告身体关键点误差比最强基线低 71.9%。
@misc{zhou2026egoexplorer,
title = {{EgoExplorer: A Vision-Motion-Touch Observatory for Human Loco-Manipulation In Situ}},
author = {Yunsong Zhou and Xuekun Jiang and Haiqi Shen and Hangxu Liu and Chaojian Shi and Yujie Liu and Jingyuan Li and Haoran Yang and Jiafei Cao and Dongzhan Zhou and Dong Wang and Yao Mu and Chunhua Shen and Weinan Zhang},
year = {2026},
eprint = {2610.18872},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.18872},
}
Isaac Sim benchmark with 5 arms, 6 dexterous hands, ~200 assets and 18k human demos over 66 tasks; reports the best trained policy (a WAM) at only 29.1% on Core-50, near 0% on precision-fit and deformable tasks.基于 Isaac Sim 的灵巧操作基准：5 种机械臂、6 种灵巧手、约 200 个资产、1.8 万条人类演示、66 个任务；报告最好的训练策略（一种世界动作模型）在 Core-50 上仅 29.1%，精密装配和柔性物体任务接近 0%。
@misc{luo2026handson,
title = {{HandsOn: A Large-Scale Simulation Benchmark for Generalizable Dexterous Manipulation}},
author = {Tianwei Luo and Yiming He and Hongjin Wu and Ting Pan and Debin Ren and Haorui Guo and Junlin Yang and Yuxin Zuo and Zhehao Kou and Kaisen Yang and Zhuoqun Lai and Yi Ding and Jiale Yu and Yuchen Zhang and Zijie Qiu and Haozhan Li and Ning Ding},
year = {2026},
eprint = {2610.26765},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.26765},
}
Tests robot value (progress) models on 15 sim and 20 real tasks across task state, progress, failure/recovery and consistency; reports that good forward-progress scores hide weak history and recovery reasoning.在 15 个仿真和 20 个真实任务上评测机器人价值（进度）模型，覆盖任务状态、进度监测、失败与恢复、一致性四个维度；报告称前向进度得分高的模型在依赖历史和恢复判断上仍然薄弱。
@misc{liu2026robovalue,
title = {{RoboValue: A Fine-Grained Sim-and-Real Benchmark for Unified Evaluation of Robotic Value Models}},
author = {Shengbang Liu and Zhengye Du and Zhilong Wan and Honghao Su and Chenxiang Xia and Chang Ge and Jinyang Xiao and Nan Wang and Zhipeng Hu and Huining Yuan and Shu'ang Yu and Yue Chen and Tianxing Chen and Chuankang Li and Maoqing Yao and Wenbo Ding and Yu Wang and Guanghui Ren and Chao Yu},
year = {2026},
eprint = {2610.26663},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.26663},
}
216 LIBERO-based tasks with industrial constraints — conveyor loading/unloading/transfer, speed, time windows, tight tolerances; reports that VLAs and world action models degrade as constraints are combined.基于 LIBERO 的 216 个工业约束任务：传送带上料、下料与转运，可调节物体形状、带速、交互时间窗与公差；报告 VLA 和世界动作模型在多重约束叠加时性能逐步下降。
@misc{sung2026libero,
title = {{LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints}},
author = {Ming Shin Cindy Sung and Duan Leijiahao and Haoxiang Wang and Boning Tan and Liangna Fu and Chao Tan and Yafu Li and Yulong Chen and Chaoqun Wan and Zhen Huang and Shuxian Liang and Xian-Sheng Hua},
year = {2026},
eprint = {2610.16782},
archivePrefix = {arXiv},
primaryClass = {cs.AI},
url = {https://arxiv.org/abs/2610.16782},
}
106 sim tasks for two dexterous hands along five axes (generalization, instructions, few demos, long horizon, memory), each with an out-of-domain twin; 8,080 teleoperated demos; seven VLAs benchmarked.双灵巧手仿真基准：106 个任务按泛化、指令遵循、少示教、长时程、记忆五个维度组织，每个任务配有域外变体；提供 8,080 条遥操作示教，并在统一协议下评测了七个 VLA 模型。
@misc{wang2026bidex,
title = {{BiDex-100: A Comprehensive Benchmark for Bimanual Dexterous Manipulation Across 100+ Simulation Tasks}},
author = {Yong Wang and Kecheng Zheng and Xing Zhu and Chao Xu and Yujun Shen and Shengze Cai},
year = {2026},
eprint = {2610.25459},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25459},
}
Benchmark for catching drifting, tumbling objects inside a spacecraft with a fixed-base arm, plus storage and servicing tasks. Reports a frozen generalist policy captured only 2/200 static targets.太空机器人操作基准：用固定基座机械臂抓取舱内漂浮、旋转的物体，另含样本存储与在轨维护任务。报告称所测的冻结通用策略在 200 个静止目标中只抓住 2 个。
@misc{wang2026astrobench,
title = {{AstroBench: Benchmarking Free-Floating Capture and Space Robotic Manipulation}},
author = {Zizhe Wang and Jia Deng and Xin Lu and Zehao Jia and Yi Huang and Yao Zhu},
year = {2026},
eprint = {2610.31064},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.31064},
}
Isaac Lab digital-twin benchmark for language-conditioned spatial tasks on the low-cost SO-101 arm (138 objects, 916 episodes); sim ranks policies exactly as the real arm. Best VLAs ~61%, an LLM-agent harness 83.2%.基于 Isaac Lab 数字孪生的低成本 SO-101 机械臂语言条件空间操作基准（138 个物体、916 个评测回合）；仿真中的策略排名与真机完全一致。最好的 VLA 约 61%，由 LLM 写代码调用抓取策略的方案达 83.2%。
Benchmark of 22.5K verified egocentric interactions with pixel-precise contact points: current models often miss where to touch. As a memory bank for an affordance policy it lifts real grasp success from 16% to 58%.第一视角接触点预测基准：2.25 万个人工核验的交互事件，每个含接触前画面、语义意图与像素级接触位置。现有方法常常定位不准甚至找不到目标物体；把 EgoContact 用作已有可供性迁移策略的记忆库，真机抓取成功率从 16% 提升到 58%。
@misc{deryo2026egocontact,
title = {{EgoContact: Evaluating Physical Intelligence with Egocentric Video}},
author = {Adam Deryło and Alexey Gavryushin and Daniel Cremers and Marc Pollefeys and Xi Wang},
year = {2026},
eprint = {2610.39167},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.39167},
}
10 long-horizon and 30 stage-level bimanual dexterous tasks with physically grounded success conditions per stage, so partial progress is measurable; includes verified teleoperated demos.10 个长时序和 30 个阶段级双手灵巧任务，每个阶段都有物理判定的成功条件，可衡量中间进度；提供经过验证的遥操作示范和统一接口。
@misc{ping2026dexarena,
title = {{DexArena: A Unified Benchmark for Long-Horizon Dexterous Manipulation}},
author = {Haonan Ping and Jianshu Hu and Yimin Wang and Yi Zhu and Pai Liang and Qiyan Jin and Qirui Bao and Ying Qian and Xinyang Ding and Hongfei Qiao and Tianle Xue and Yucheng Bai and Soo Wei Jie and Shuguang Li and Yutong Ban},
year = {2026},
eprint = {2611.02629},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.02629},
}
Extends robot piano to one to five Shadow Hands, with robust, safe and multi-agent RL settings: at least 912 task settings and 36 baselines. Reports that top RL algorithms and LLM agents still struggle.把机器人弹钢琴扩展到 1 至 5 只 Shadow 手，并支持鲁棒、安全和多智能体强化学习：至少 912 种任务设置、36 个基线算法。作者报告前沿算法与大模型智能体在高难任务上仍很吃力。
@misc{xu2026omnipiano,
title = {{OmniPiano: Diverse Dexterous Piano-Playing Challenges for Standard, Robust, Safe, and Multi-Agent RL}},
author = {Fan Xu and Fan Yang and Yifan Xu and Ruiyuan Zhang and Yimou Wu and Nuo Yi and Muning Wen and Guang Chen and Costas Spanos and Shangding Gu},
year = {2026},
eprint = {2611.00243},
archivePrefix = {arXiv},
primaryClass = {cs.AI},
url = {https://arxiv.org/abs/2611.00243},
}
Sim platform where a fine-tuned VLA generates nearly 8,000 successful garment trajectories, re-rendered for RGB policies. Those beat policies trained on the original human data, but unseen garments stay hard.仿真平台：用微调后的 VLA 生成近 8000 条成功的衣物操作轨迹，并重新渲染供 RGB 策略训练，效果优于直接用人类数据训练，但泛化到未见衣物仍是难题。
@misc{huang2026clothdojo,
title = {{ClothDojo: Learned Data Generation and Benchmarking for Bimanual Garment Flattening and Folding}},
author = {Zilong Huang and Zipeng Ye and Caicheng Wang and Yuchen Xie and Jiaxiang Luo},
year = {2026},
eprint = {2611.01190},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.01190},
}
Suites isolate generalisation across spatial layout, object instance, object category and subtask composition, plus bimanual BiMESA. The best of seven baselines reaches only 52% and 49%.五组评测分别隔离空间布局、物体实例、物体类别和子任务组合上的泛化能力，并提供双臂版 BiMESA。七种基线中最好的也只有 52% 和 49%，仍有很大提升空间。
@misc{wilcox2026mesa,
title = {{MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics}},
author = {Albert Wilcox and Frank Chang and Nhi Nguyen and Aishani Chakraborty and Jeremy A. Collins and Vaibhav Saxena and Benjamin Joffe and Siddharth Karamcheti and Animesh Garg},
year = {2026},
eprint = {2611.05567},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.05567},
}
Synthetic T-shirt folding and unfolding data: 6 robots, 1K shirts, 120K episodes. Policies trained only on it are reported above 90% success on unseen real shirts.合成 T 恤折叠与展开数据：6 种机器人、1000 件 T 恤、12 万条轨迹。报告仅用其训练的策略在未见过的真实 T 恤上成功率超过 90%。
@misc{chen2026foldnet,
title = {{FoldNet++: a Large-Scale Synthetic Dataset for Robotic T-Shirt Folding and Unfolding}},
author = {Yuxing Chen and Zhiyuan Wei and Bowen Xiao and Zhizheng Zhang and He Wang},
year = {2026},
eprint = {2609.12433},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.12433},
}
Human demos of creative tool use (repurposing objects when the usual tool is missing) from 1 ego + 8 exo cameras with hand, tool and object labels; benchmarks reasoning, 3D/4D reconstruction and policy learning.记录人类创造性使用工具（缺少常规工具时改用其他物体）的示范，含一路第一视角和八路第三视角相机及手、工具、物体标注；基准涵盖推理、三维/四维重建与策略学习。
@misc{elmoghany2026tooldex,
title = {{ToolDex: An Ego-Exo Human Demonstration Dataset for Creative Physical Tool Use}},
author = {Mohamed Elmoghany and Kehan Li and Ruoqu Chen and Tarun Chiruvolu and Shutong Zhang and Ryan Wang and Stephen Tian and Yaru Niu and Michelle Guo and Ding Zhao and Weiyu Liu and Mengdi Xu and Li Fei-Fei and Jiajun Wu},
year = {2026},
eprint = {2611.09036},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09036},
}
Manipulation benchmark where simulated humans cue goals, finish parts of the task, interrupt or share the workspace. Reports that VLA and IL policies drop when humans act, and higher success does not mean safer motion.让仿真人类参与其中的操作基准：人会提示目标、完成部分任务、打断或与机器人共处同一工作区。报告显示 VLA 与模仿学习策略在有人活动时性能下降，且成功率更高并不意味着更安全。
@misc{zhou2026harp,
title = {{HARP-Bench: Benchmarking Robot Manipulation Around Active Humans}},
author = {Qinhong Zhou and Riley Zong and Chunru Lin and Diyi Hu and Shivraj Singh Bhatti and Hongxin Zhang and Lin Qiu and Hanqing Zeng and Chuang Gan},
year = {2026},
eprint = {2611.11123},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11123},
}
'Stochastic challenges': reproducible real scenes with tasks sampled from a shared distribution, so robots in different labs compare by success rate like a hidden test set. Shown with two robots in two labs.提出“随机挑战”评测范式：可复现的真实场景 + 从同一分布随机抽取任务，让不同实验室的机器人能像隐藏测试集一样按成功率比较。在一个移动操作基准上用两个实验室的两台机器人做了演示。
@misc{glaer2026benchmarking,
title = {{Benchmarking Real-World Robotic Manipulation through Stochastic Challenges}},
author = {Elias Gläßer and Gülhan Şikaroğlu and Pavlo Kupyn and Marc Pollefeys and Zuria Bauer and Hermann Blum},
year = {2026},
eprint = {2611.12804},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12804},
}
A portable hand-held rig for collecting demonstrations with a high-DoF robot hand, using the same hand hardware for collection and deployment, so there is no retargeting from human hands and no full robot setup needed.便携手持式数据采集装置，面向高自由度机器人灵巧手；采集与部署使用同一只机械手，无需从人手动作重定向，也无需搭建完整机器人系统。
@misc{xie2026dexport,
title = {{DexPort: A Hand-Held Skill Interface for Dexterous Hand Data Collection}},
author = {Amber Xie and Jiawei Fu and Xiaofeng Guo and Carmelo Sferrazza and Guanya Shi and Haozhi Qi},
year = {2026},
eprint = {2610.32969},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.32969},
}
Fully linkage-driven humanoid hand with 3 actuated DoF per finger; a flexure hinge and spring-steel drive rods replace many tiny parts, so it prints on a budget 3D printer at near human-hand size.完全连杆驱动的人形灵巧手，每根手指 3 个主动自由度；用柔性铰链和弹簧钢驱动杆取代大量微型零件，可用入门级 3D 打印机制造，尺寸接近真人手。
@misc{adwani2026phd,
title = {{PHD-Hand: A Printable Humanoid Dexterous Hand with Linkage-Driven 3-DOA Digits}},
author = {Neel Adwani and Muhaiminul Islam Akash and Saffia Ouali and Ethan Liu and Cong Wang},
year = {2026},
eprint = {2610.29030},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.29030},
}
One operator drives a Unitree Go2 + Piper arm with a leader arm and two handheld controllers, with sensorless force feedback; records a unified 11-D action reused at deployment (shown by fine-tuning MolmoAct2).单人即可同时操控 Unitree Go2 四足与 Piper 机械臂：主端机械臂加两个手柄，无需力传感器即可力反馈；ROS 2 管线记录与部署共用的 11 维统一动作，并通过微调 MolmoAct2 验证从采集到部署的全流程。
@misc{langella2026locopilot,
title = {{LocoPilot: Single-Operator Bilateral Teleoperation for Loco-Manipulation Policy Learning}},
author = {Antonio Langella and Giovanni Minelli and Giovanni B. Dessy and Lorenzo Amatucci and Giulio Turrisi and Victor Barasuol and Claudio Semini},
year = {2026},
eprint = {2610.27998},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.27998},
}
Fully open UMI-style handheld gripper with a low-cost stereo camera for metric, low-drift trajectories, plus UMI-Loc-Bench-v1 for localization accuracy; hardware, software and data under a permissive license.完全开源的 UMI 式手持夹爪：3D 打印、标准 ISO 接口，用低成本双目相机获得公制尺度、低漂移的轨迹，并提出手持采集设备定位精度基准 UMI-Loc-Bench-v1；硬件、软件与数据以可商用许可证发布。
@misc{murali2026openjaws,
title = {{OpenJaws: An Open Hardware UMI Framework with Stereo Localization for Generalist Manipulation}},
author = {Adithyavairavan Murali and Yash Oza and Karanbir Chahal and Neel Jawale and Silas De Vries and Hesam Rabeti and Erwin Coumans and Balakumar Sundaralingam and Stan Birchfield},
year = {2026},
eprint = {2610.36598},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.36598},
}
Open-source wearable extra thumb: a tendon-driven finger with motors moved to the forearm; an imitation policy maps palm-camera images to joint positions, inferring intent without explicit triggers.开源的可穿戴“第六指”：紧凑的双腱驱动手指，电机移到前臂以减轻末端重量；模仿学习策略直接把掌心相机图像映射为关节位置，从视觉中推断使用者意图，无需显式手势触发，软硬件全部开源。
@misc{hou2026egothumb,
title = {{EgoThumb: A Vision-Driven Supernumerary Robotic Thumb for Autonomous Grasp and Holding Assistance}},
author = {Jianyu Hou and Jiawei Fu and Weikang Wan and Minghui Jiang and Jiajun Li and Hao Su},
year = {2026},
eprint = {2610.37582},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.37582},
}
Open-source bimanual UMI with camera, LiDAR-inertial sensing, compute and power on board: no external tracking or calibration; parallel SLAM outputs training-ready data. Releases 40 h of data and hardware files.全栈开源的双手 UMI 野外数据采集系统：手持设备集成相机、激光雷达-惯性传感、板载算力与电源，无需外部追踪或现场标定；异步 SLAM 流水线直接产出可训练数据集，并发布 40 小时数据、硬件设计与复现指南，用 π0.5 和 MolmoAct2 微调验证数据质量。
@misc{velamoor2026umi,
title = {{UMI-Live for In-The-Wild Bimanual Robotics Data Collection}},
author = {Govind Velamoor and Adrian Lin and Sam Jung and Spencer Mateega and Brandon Y. Yang and Amish Sethi},
year = {2026},
eprint = {2610.39385},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.39385},
}
Instead of one universal gripper: a fixed robot-side coupling plus cheap tool-side adapters with tactile sensors turn household tools into swappable end effectors; imitation policies cook tofu and flip meat.不追求万能夹爪，而是用固定的机器人侧接口加低成本、带触觉传感器的工具侧适配器，把日常厨具变成可自动更换的末端执行器，兼容多种机器人平台；配套遥操作采集示范，模仿学习策略完成煎豆腐、拌沙拉、翻肉等工具任务。
@misc{zhang2026swapbot,
title = {{SwapBot: Tool-Centric Robot Manipulation via Interchangeable End-Effectors and Tactile Sensing}},
author = {Yunchu Zhang and Ray Song and Junhan Jiao and Liyiming Ke and Yuchen Yang and Ruihan Gao and Emma Romig and Siddhartha Srinivasa and Yiyue Luo},
year = {2026},
eprint = {2610.40646},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.40646},
}
An open-source 56-DOF mobile bimanual robot with 7-DOF arms and 17-DOF hands for about US$8.1k in parts, with Apple Vision Pro teleoperation and a demonstrated imitation learning pipeline.开源 56 自由度移动双臂机器人，含 7 自由度手臂和 17 自由度灵巧手，零件成本约 8100 美元，支持 Apple Vision Pro 遥操作，并演示了模仿学习全流程。
@misc{cheng2026mabel,
title = {{MABEL: An Open Anthropomorphic Mobile Bimanual Platform for Teleoperation and Learning}},
author = {Qilong Cheng and Rooholla Khorrambakht and Mohamed Yassine Kabouri and Quang Nam Nguyen and Luca Macesanu and Haoran Zhou and Yipeng Wang and Ludovic Righetti},
year = {2026},
eprint = {2611.04083},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.04083},
}
Pretrains a grasping VLA almost only on SynGrasp-1B, a billion-frame photorealistic sim dataset, co-trained with web data; reports zero-shot sim-to-real open-vocabulary grasping and few-shot adaptation.几乎只用十亿帧级的照片级仿真抓取数据集 SynGrasp-1B（并联合网络语义数据）预训练抓取 VLA；论文报告可零样本从仿真迁移到真实开放词汇抓取，并支持少样本适配。
@misc{deng2025graspvla,
title = {{GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Data}},
author = {Shengliang Deng and Mi Yan and Songlin Wei and Haixin Ma and Yuxin Yang and Jiayi Chen and Zhiqi Zhang and Taoyu Yang and Xuheng Zhang and Wenhao Zhang and Heming Cui and Zhizheng Zhang and He Wang},
year = {2025},
eprint = {2505.03233},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2505.03233},
}
A dexterous VLA pretrained on human hand motion from mocap, VR and RGB video ('physical instruction tuning'), with part-level motion tokens, then aligned to 3D space and post-trained for real robot hands.灵巧手 VLA：先在动捕、VR 与普通视频中的人手动作上预训练（"物理指令微调"），使用部件级动作 token，再做三维空间对齐，并针对真实机器人灵巧手后训练。
@misc{luo2025being,
title = {{Being-H0: Vision-Language-Action Pretraining from Large-Scale Human Videos}},
author = {Hao Luo and Yicheng Feng and Wanpeng Zhang and Sipeng Zheng and Ye Wang and Haoqi Yuan and Jiazheng Liu and Chaoyi Xu and Qin Jin and Zongqing Lu},
year = {2025},
eprint = {2507.15597},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2507.15597},
}
A hierarchical VLA in which a fine-tuned VLM draws a coarse 2D path for the end effector on the image and a low-level 3D policy follows it, so the VLM can be trained on cheap off-domain data such as sim or videos.分层 VLA：微调后的 VLM 在图像上画出末端执行器的粗略二维路径，再由底层三维策略跟随执行；因此 VLM 可以用仿真、视频等低成本的非本域数据训练。
@misc{li2025hamster,
title = {{HAMSTER: Hierarchical Action Models for Open-World Robot Manipulation}},
author = {Yi Li and Yuquan Deng and Jesse Zhang and Joel Jang and Marius Memmel and Raymond Yu and Caelan Reed Garrett and Fabio Ramos and Dieter Fox and Anqi Li and Abhishek Gupta and Ankit Goyal},
year = {2025},
eprint = {2502.05485},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2502.05485},
}
Trains VLAs on egocentric human videos, predicting wrist and hand actions, because human data scales beyond robot data and is richer in scenes and tasks.用第一视角人类视频训练 VLA，预测手腕与手部动作；人类数据规模更大，场景和任务也更丰富。
@misc{yang2025egovla,
title = {{EgoVLA: Learning Vision-Language-Action Models from Egocentric Human Videos}},
author = {Ruihan Yang and Qinxi Yu and Yecheng Wu and Rui Yan and Borui Li and An-Chieh Cheng and Xueyan Zou and Yunhao Fang and Xuxin Cheng and Ri-Zhao Qiu and Hongxu Yin and Sifei Liu and Song Han and Yao Lu and Xiaolong Wang},
year = {2025},
eprint = {2507.12440},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2507.12440},
}
A pretrained VLM plans and a diffusion controller grasps with a dexterous hand, working on domain-invariant features from foundation models; reports 90%+ grasp success across thousands of unseen cluttered scenes.由预训练 VLM 做高层规划、扩散控制器驱动灵巧手抓取，底层输入为基础模型提取的域不变特征；报告在数千个未见过的杂乱场景中抓取成功率超过 90%。
@misc{zhong2025dexgraspvla,
title = {{DexGraspVLA: A Vision-Language-Action Framework Towards General Dexterous Grasping}},
author = {Yifan Zhong and Xuchuan Huang and Ruochong Li and Ceyao Zhang and Zhang Chen and Tianrui Guan and Fanlian Zeng and Ka Num Lui and Yuyao Ye and Yitao Liang and Yaodong Yang and Yuanpei Chen},
year = {2025},
eprint = {2502.20900},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2502.20900},
}
Pretrains a task-agnostic latent action space in a low-fidelity simulator so real-world RL on high-DoF mobile manipulators becomes safe and efficient.在低保真仿真器中预训练与任务无关的潜在动作空间，让高自由度移动操作机器人的真实世界强化学习安全高效。
@misc{hu2025slac,
title = {{SLAC: Safe and Efficient Real-Robot Reinforcement Learning via Unsupervised Simulation Pre-Training}},
author = {Jiaheng Hu and Peter Stone and Roberto Martín-Martín},
year = {2025},
eprint = {2506.04147},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2506.04147},
}
Uses egocentric human demonstrations as cross-embodiment data for humanoid manipulation policies, narrowing the human-robot gap in data and model design.把第一视角人类示教作为人形操作策略的跨形态数据，从数据和模型两方面缩小人与机器人的差距。
@misc{qiu2025humanoid,
title = {{Humanoid Policy ~ Human Policy}},
author = {Ri-Zhao Qiu and Shiqi Yang and Xuxin Cheng and Chaitanya Chawla and Jialong Li and Tairan He and Ge Yan and David J. Yoon and Ryan Hoque and Lars Paulsen and Ge Yang and Jian Zhang and Sha Yi and Guanya Shi and Xiaolong Wang},
year = {2025},
eprint = {2503.13441},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2503.13441},
}
Active perception learned from human demonstrations: a 6-DoF robotic neck and a VR interface let bimanual robots search, track and focus.从人类示教学习主动感知：6 自由度机械颈加 VR 界面，让双臂机器人能搜索、跟踪和聚焦目标。
@misc{xiong2025vision,
title = {{Vision in Action: Learning Active Perception from Human Demonstrations}},
author = {Haoyu Xiong and Xiaomeng Xu and Jimmy Wu and Yifan Hou and Jeannette Bohg and Shuran Song},
year = {2025},
eprint = {2506.15666},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2506.15666},
}
H-RDT pretrains a diffusion transformer on human manipulation data to work around the morphology mismatch of cross-robot datasets.H-RDT 用人类操作数据预训练扩散 Transformer，规避跨机器人数据集的形态差异问题。
@misc{bi2025h,
title = {{H-RDT: Human Manipulation Enhanced Bimanual Robotic Manipulation}},
author = {Hongzhe Bi and Lingxuan Wu and Tianwei Lin and Hengkai Tan and Zhizhong Su and Hang Su and Jun Zhu},
year = {2025},
eprint = {2507.23523},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2507.23523},
}
Procedural dexterous grasp synthesis reporting orders-of-magnitude speedups, without carefully tuned energy functions, including for irregular tool-like objects.过程式灵巧抓取合成，报告提速数个量级，无需精调能量函数，并适用于不规则的工具类物体。
SAM2Act builds on a visual foundation model and adds a memory architecture for multitask manipulation and tasks that need spatial memory.SAM2Act 基于视觉基础模型并加入记忆架构，用于多任务操作和需要空间记忆的任务。
@misc{fang2025sam,
title = {{SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation}},
author = {Haoquan Fang and Markus Grotz and Wilbert Pumacay and Yi Ru Wang and Dieter Fox and Ranjay Krishna and Jiafei Duan},
year = {2025},
eprint = {2501.18564},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2501.18564},
}
A slow latent diffusion policy plans action chunks while a fast policy reacts to tactile/force feedback; reports >35% higher scores than visual imitation baselines on three contact-rich tasks.慢速扩散策略在隐空间规划动作块，快速策略根据触觉/力反馈高频调整；论文报告在三个接触丰富任务上比纯视觉模仿学习基线总分高 35% 以上。
Low-cost bilateral teleop relays arm forces to the operator, and a curriculum corrupts vision less and less over training so policies learn to attend to force.低成本双边遥操作把机械臂受力反馈给操作者；训练时对视觉的破坏逐渐减弱，让策略学会关注力信息。
@misc{liu2025factr,
title = {{FACTR: Force-Attending Curriculum Training for Contact-Rich Policy Learning}},
author = {Jason Jingzhou Liu and Yulong Li and Kenneth Shaw and Tony Tao and Ruslan Salakhutdinov and Deepak Pathak},
year = {2025},
eprint = {2502.17432},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2502.17432},
}
Picks up paper-like thin objects with a multi-fingered hand carrying high-resolution omnidirectional tactile sensors, tackling state estimation and grasp planning.用带高分辨率全向触觉传感器的多指手抓取纸张类薄物体，同时解决状态估计与抓取规划。
@misc{lin2025pp,
title = {{PP-Tac: Paper Picking Using Tactile Feedback in Dexterous Robotic Hands}},
author = {Pei Lin and Yuzhe Huang and Wanlin Li and Jianpeng Ma and Chenxi Xiao and Ziyuan Jiao},
year = {2025},
eprint = {2504.16649},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2504.16649},
}
@misc{gu2025humanoid,
title = {{Humanoid Locomotion and Manipulation: Current Progress and Challenges in Control, Planning, and Learning}},
author = {Zhaoyuan Gu and Junheng Li and Wenlan Shen and Wenhao Yu and Zhaoming Xie and Stephen McCrory and Xianyi Cheng and Abdulaziz Shamsah and Robert Griffin and C. Karen Liu and Abderrahmane Kheddar and Xue Bin Peng and Yuke Zhu and Guanya Shi and Quan Nguyen and Gordon Cheng and Huijun Gao and Ye Zhao},
year = {2025},
eprint = {2501.02116},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2501.02116},
}
VPP — fine-tunes Stable Video Diffusion on robot and human videos and feeds its one-pass predictive features to a diffusion policy; reports 4.33 average length on CALVIN ABC→D and real Franka / dexterous-hand results.VPP：在机器人与人类操作视频上微调 Stable Video Diffusion，只做一次前向提取"预测性"视觉特征再交给扩散策略；论文报告 CALVIN ABC→D 平均完成长度 4.33，并有 Franka 与灵巧手真机实验。
@misc{hu2024video,
title = {{Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations}},
author = {Yucheng Hu and Yanjiang Guo and Pengchao Wang and Xiaoyu Chen and Yen-Jen Wang and Jianke Zhang and Koushil Sreenath and Chaochao Lu and Jianyu Chen},
year = {2024},
eprint = {2412.14803},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2412.14803},
}
UVA — one shared latent for video and actions with two light diffusion heads; masked training lets the same model act as policy, video generator and forward/inverse dynamics model, while keeping policy inference fast.UVA：视频与动作共享一个潜表示，配两个轻量扩散头解码；通过掩码训练，同一模型既能当策略，也能做视频生成和正/逆动力学，且策略推理仍然较快。
Genie Envisioner (AgiBot) — a video-diffusion world model (GE-Base) plus a flow-matching action decoder (GE-Act), an action-conditioned neural simulator (GE-Sim) and the EWMBench benchmark in one platform.智元 Genie Envisioner：一个平台集成视频扩散世界模型 GE-Base、流匹配动作解码器 GE-Act、动作条件神经仿真器 GE-Sim，以及世界模型评测基准 EWMBench。
@misc{liao2025genie,
title = {{Genie Envisioner: A Unified World Foundation Platform for Robotic Manipulation}},
author = {Yue Liao and Pengfei Zhou and Siyuan Huang and Donglin Yang and Shengcong Chen and Yuxin Jiang and Yue Hu and Jingbin Cai and Si Liu and Jianlan Luo and Liliang Chen and Shuicheng Yan and Maoqing Yao and Guanghui Ren},
year = {2025},
eprint = {2508.05635},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2508.05635},
}
DSRL — adapts a frozen diffusion policy by running RL over its input noise, not its weights; needs only forward passes, works with π0, and reports real-robot success rising from 20% to 90% in under 50 online episodes.DSRL：不改扩散策略的权重，而是对其输入的初始噪声做强化学习来"引导"输出动作；只需前向推理、可用于 π0，论文报告真机任务在不到 50 次在线交互内成功率从 20% 升至 90%。
@misc{wagenmaker2025steering,
title = {{Steering Your Diffusion Policy with Latent Space Reinforcement Learning}},
author = {Andrew Wagenmaker and Mitsuhiko Nakamoto and Yunchu Zhang and Seohong Park and Waleed Yagoub and Anusha Nagabandi and Abhishek Gupta and Sergey Levine},
year = {2025},
eprint = {2506.15799},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2506.15799},
}
Isaac Sim platform where an LLM builds tasks and layouts from a task-oriented scene graph over ~10K objects; its 200-scenario benchmark finds modular foundation-model systems generalize better than end-to-end policies.基于 Isaac Sim 的平台：LLM 依据任务导向场景图，从约 1 万个物体中生成任务与布局；其 200 个场景的基准发现，结合基础模型的模块化系统比端到端策略泛化更好。
@misc{gao2025genmanip,
title = {{GenManip: LLM-driven Simulation for Generalizable Instruction-Following Manipulation}},
author = {Ning Gao and Yilun Chen and Shuai Yang and Xinyi Chen and Yang Tian and Hao Li and Haifeng Huang and Hanqing Wang and Tai Wang and Jiangmiao Pang},
year = {2025},
eprint = {2506.10966},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2506.10966},
}
829 hours of egocentric Apple Vision Pro video with tracked 3D poses for every hand joint, across 194 tabletop tasks from tying shoelaces to folding laundry; includes hand-trajectory prediction benchmarks.用 Apple Vision Pro 采集的 829 小时第一视角视频，配有每个手部关节的三维姿态，覆盖从系鞋带到叠衣服的 194 种桌面任务，并提供手部轨迹预测基准。
@misc{hoque2025egodex,
title = {{EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Video}},
author = {Ryan Hoque and Peide Huang and David J. Yoon and Mouli Sivapurapu and Jian Zhang},
year = {2025},
eprint = {2505.11709},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2505.11709},
}
A whole-body teleoperation interface plus a learning method for a wheeled bimanual robot doing long household tasks that need reach, balance and navigation.面向轮式双臂机器人的全身遥操作接口与学习方法，用于需要伸展、平衡与导航的长程家务任务。
@misc{jiang2025behavior,
title = {{BEHAVIOR Robot Suite: Streamlining Real-World Whole-Body Manipulation for Everyday Household Activities}},
author = {Yunfan Jiang and Ruohan Zhang and Josiah Wong and Chen Wang and Yanjie Ze and Hang Yin and Cem Gokmen and Shuran Song and Jiajun Wu and Li Fei-Fei},
year = {2025},
eprint = {2503.05652},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2503.05652},
}
A 3D-printable, 17-DoF tendon-driven hand with fingertip tactile sensing, under CHF 2,000 in materials; design files and code are open, and the paper reports long continuous runs for teleoperation and learning.可 3D 打印的 17 自由度腱驱动灵巧手，指尖带触觉传感，材料成本低于 2000 瑞士法郎；设计文件与代码开源，论文报告了可长时间连续运行用于遥操作与学习。
@misc{christoph2025orca,
title = {{ORCA: An Open-Source, Reliable, Cost-Effective, Anthropomorphic Robotic Hand for Uninterrupted Dexterous Task Learning}},
author = {Clemens C. Christoph and Maximilian Eberlein and Filippos Katsimalis and Arturo Roberti and Aristotelis Sympetheros and Michel R. Vogt and Davide Liconti and Chenyu Yang and Barnabas Gavin Cangan and Ronan J. Hinchet and Robert K. Katzschmann},
year = {2025},
eprint = {2504.04259},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2504.04259},
}
Exoskeleton demos are converted into pseudo-robot demonstrations; the paper reports its RISE-2 policy trained on them alone, with no robot data, performs comparably to one trained on teleoperated robot data.将外骨骼示教转换为“伪机器人”示教；论文报告仅用这些数据（无任何机器人数据）训练的 RISE-2 策略，效果与用遥操作机器人数据训练的策略相当。
@misc{fang2025airexo,
title = {{AirExo-2: Scaling up Generalizable Robotic Imitation Learning with Low-Cost Exoskeletons}},
author = {Hongjie Fang and Chenxi Wang and Yiming Wang and Jingjing Chen and Shangning Xia and Jun Lv and Zihao He and Xiyan Yi and Yunhan Guo and Xinyu Zhan and Lixin Yang and Weiming Wang and Cewu Lu and Hao-Shu Fang},
year = {2025},
eprint = {2503.03081},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2503.03081},
}
DexUMI — an exoskeleton matched to the robot hand's fingertip workspace, plus inpainting that swaps the human hand for the robot's; reports 86% average success (Inspire, XHand) and ~3.2× faster collection than teleop.DexUMI：按目标机器人手的指尖工作空间设计的可穿戴手部外骨骼，再用视频修补把画面中的人手换成机器人手；论文报告在 Inspire 与 XHand 上平均成功率 86%，采集效率约为遥操作的 3.2 倍。
@misc{xu2025dexumi,
title = {{DexUMI: Using Human Hand as the Universal Manipulation Interface for Dexterous Manipulation}},
author = {Mengda Xu and Han Zhang and Yifan Hou and Zhenjia Xu and Linxi Fan and Manuela Veloso and Shuran Song},
year = {2025},
eprint = {2505.21864},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2505.21864},
}
An open-source teleoperation glove buildable in hours for under $600, with 21-DoF hand capture, 5-DoF cable-driven force feedback and fingertip haptics; reports the feedback matters for contact-rich tasks without vision.开源遥操作手套，几小时即可组装、成本低于 600 美元：21 自由度手部动捕、5 自由度绳驱力反馈和指尖振动触觉；报告在无视觉输入时力/触觉反馈对接触密集任务至关重要，采集数据可训练模仿学习策略。
@misc{zhang2025doglove,
title = {{DOGlove: Dexterous Manipulation with a Low-Cost Open-Source Haptic Force Feedback Glove}},
author = {Han Zhang and Songbo Hu and Zhecheng Yuan and Huazhe Xu},
year = {2025},
eprint = {2502.07730},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2502.07730},
}
A human-sized, tendon-driven hand (5 fingers, 15 underactuated DoF) from 3D-printed and off-the-shelf parts; joint-to-actuator control is learned from MANUS-glove mocap data instead of hand-modelled.人手大小的腱驱动灵巧手（5 指、15 个欠驱动自由度），由 3D 打印件和现成零件制成；关节到电机的控制映射用 MANUS 数据手套的动捕数据学习得到，而非手工建模。
@misc{zorin2025ruka,
title = {{RUKA: Rethinking the Design of Humanoid Hands with Learning}},
author = {Anya Zorin and Irmak Guzey and Billy Yan and Aadhithya Iyer and Lisa Kondrich and Nikhil X. Bhattasali and Lerrel Pinto},
year = {2025},
eprint = {2504.13165},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2504.13165},
}
@misc{wang2024dexcap,
title = {{DexCap: Scalable and Portable Mocap Data Collection System for Dexterous Manipulation}},
author = {Chen Wang and Haochen Shi and Weizhuo Wang and Ruohan Zhang and Li Fei-Fei and C. Karen Liu},
year = {2024},
eprint = {2403.07788},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2403.07788},
}
ReKep — a VLM writes manipulation tasks as Python cost functions over 3D keypoints (relational keypoint constraints), solved by a hierarchical optimizer in real time; zero-shot, no task-specific training.ReKep：由视觉语言模型把操作任务写成定义在三维关键点上的 Python 代价函数（关系关键点约束），再由分层优化器实时求解；零样本，无需针对任务训练。
@misc{huang2024rekep,
title = {{ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulation}},
author = {Wenlong Huang and Chen Wang and Yunzhu Li and Ruohan Zhang and Li Fei-Fei},
year = {2024},
eprint = {2409.01652},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2409.01652},
}
Treats egocentric human video with 3D hand tracking (Project Aria glasses) as robot data and co-trains one policy on both; reports an extra hour of human hand data helps more than an extra hour of robot data.把佩戴 Project Aria 眼镜采集、带三维手部跟踪的第一视角人类视频当作机器人数据，与机器人数据共同训练同一个策略；论文报告多 1 小时人手数据比多 1 小时机器人数据更有用。
@misc{kareer2024egomimic,
title = {{EgoMimic: Scaling Imitation Learning via Egocentric Video}},
author = {Simar Kareer and Dhruv Patel and Ryan Punamiya and Pranay Mathur and Shuo Cheng and Chen Wang and Judy Hoffman and Danfei Xu},
year = {2024},
eprint = {2410.24221},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2410.24221},
}
Seer forecasts the next visual state and feeds it to an inverse dynamics model that outputs actions, trained end to end and pretrained on DROID; reports 4.28 avg. length on CALVIN ABC-D plus LIBERO and real-robot gains.Seer 先预测未来视觉状态，再由逆动力学模型据此输出动作，端到端训练并在 DROID 上预训练；报告 CALVIN ABC-D 平均任务长度 4.28，在 LIBERO-LONG 和真机任务上也有提升。
@misc{tian2024predictive,
title = {{Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulation}},
author = {Yang Tian and Sizhe Yang and Jia Zeng and Ping Wang and Dahua Lin and Hao Dong and Jiangmiao Pang},
year = {2024},
eprint = {2412.15109},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2412.15109},
}
3D Diffuser Actor conditions diffusion policies on 3D scene features aggregated from one or more camera views.3D Diffuser Actor 用单视角或多视角聚合的三维场景特征来条件化扩散策略。
@misc{ke2024d,
title = {{3D Diffuser Actor: Policy Diffusion with 3D Scene Representations}},
author = {Tsung-Wei Ke and Nikolaos Gkanatsios and Katerina Fragkiadaki},
year = {2024},
eprint = {2402.10885},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2402.10885},
}
HDP splits a policy into a high-level next-best-pose planner and a low-level goal-conditioned diffusion policy that generates motion to reach it.HDP 把策略分为高层“下一最佳位姿”规划与低层目标条件扩散策略，由后者生成到达该位姿的运动。
@misc{ma2024hierarchical,
title = {{Hierarchical Diffusion Policy for Kinematics-Aware Multi-Task Robotic Manipulation}},
author = {Xiao Ma and Sumit Patidar and Iain Haughton and Stephen James},
year = {2024},
eprint = {2403.03890},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2403.03890},
}
Distils a diffusion teacher into a consistency model so visuomotor policies run fast on robots without high-end GPUs.把扩散教师模型蒸馏为一致性模型，让视觉运动策略在没有高端 GPU 的机器人上快速运行。
@misc{prasad2024consistency,
title = {{Consistency Policy: Accelerated Visuomotor Policies via Consistency Distillation}},
author = {Aaditya Prasad and Kevin Lin and Jimmy Wu and Linqi Zhou and Jeannette Bohg},
year = {2024},
eprint = {2405.07503},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2405.07503},
}
ManiCM applies consistency models to 3D diffusion policies, removing most denoising steps for real-time inference.ManiCM 把一致性模型用于三维扩散策略，去掉大部分去噪步骤以实现实时推理。
@misc{gao2024manicm,
title = {{ManiCM: Real-time 3D Diffusion Policy via Consistency Model for Robotic Manipulation}},
author = {Zifeng Gao and Guanxing Lu and Tianxing Chen and Wenxun Dai and Ziwei Wang and Chao Shang and Wenbo Ding and Yansong Tang},
year = {2024},
eprint = {2406.01586},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2406.01586},
}
Streaming Diffusion Policy speeds up action synthesis using variable-noise diffusion, aimed at tasks that need fast reactive policies.Streaming Diffusion Policy 用可变噪声扩散加速动作生成，面向需要快速反应的任务。
@misc{heg2024streaming,
title = {{Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Models}},
author = {Sigmund H. Høeg and Yilun Du and Olav Egeland},
year = {2024},
eprint = {2406.04806},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2406.04806},
}
Sparse Diffusion Policy reuses sparse expert modules across tasks instead of one universal policy, for multitask and continual learning.Sparse Diffusion Policy 在任务间复用稀疏专家模块而非单一通用策略，用于多任务和持续学习。
@misc{wang2024sparse,
title = {{Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learning}},
author = {Yixiao Wang and Yifei Zhang and Mingxiao Huo and Ran Tian and Xiang Zhang and Yichen Xie and Chenfeng Xu and Pengliang Ji and Wei Zhan and Mingyu Ding and Masayoshi Tomizuka},
year = {2024},
eprint = {2407.01531},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2407.01531},
}
MDT is a diffusion transformer that learns versatile behaviour from multimodal goals with few language annotations, using two self-supervised objectives.MDT 是扩散 Transformer，借助两个自监督目标，在语言标注很少时从多模态目标学习多样行为。
ThinkGrasp uses GPT-4o reasoning to pick which object and part to grasp in heavy clutter, a plug-and-play vision-language system.ThinkGrasp 借助 GPT-4o 推理，在严重杂乱中决定抓哪个物体的哪个部位，是即插即用的视觉-语言系统。
@misc{qian2024thinkgrasp,
title = {{ThinkGrasp: A Vision-Language System for Strategic Part Grasping in Clutter}},
author = {Yaoyao Qian and Xupeng Zhu and Ondrej Biza and Shuo Jiang and Linfeng Zhao and Haojie Huang and Yu Qi and Robert Platt},
year = {2024},
eprint = {2407.11298},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2407.11298},
}
@misc{duan2024manipulate,
title = {{Manipulate-Anything: Automating Real-World Robots using Vision-Language Models}},
author = {Jiafei Duan and Wentao Yuan and Wilbert Pumacay and Yi Ru Wang and Kiana Ehsani and Dieter Fox and Ranjay Krishna},
year = {2024},
eprint = {2406.18915},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2406.18915},
}
AHA is a vision-language model for detecting and reasoning about manipulation failures, a gap in earlier VLM work.AHA 是用于检测和推理操作失败的视觉-语言模型，填补此前 VLM 研究的空白。
@misc{duan2024aha,
title = {{AHA: A Vision-Language-Model for Detecting and Reasoning Over Failures in Robotic Manipulation}},
author = {Jiafei Duan and Wilbert Pumacay and Nishanth Kumar and Yi Ru Wang and Shulin Tian and Wentao Yuan and Ranjay Krishna and Dieter Fox and Ajay Mandlekar and Yijie Guo},
year = {2024},
eprint = {2410.00371},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2410.00371},
}
@misc{zheng2024survey,
title = {{A Survey of Embodied Learning for Object-Centric Robotic Manipulation}},
author = {Ying Zheng and Lei Yao and Yuejiao Su and Yi Zhang and Yi Wang and Sicheng Zhao and Yiyi Zhang and Lap-Pui Chau},
year = {2024},
eprint = {2408.11537},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2408.11537},
}
Multi-GraspSet, a large multi-hand grasp dataset with automatic contact annotations, and an LLM that generates language-guided grasps for different robot hands in one model.发布带自动接触标注的大规模多手型抓取数据集 Multi-GraspSet，并用一个基于大模型的统一框架，根据语言指令为不同机器人手生成抓取。
@misc{li2024multi,
title = {{Multi-GraspLLM: A Multimodal LLM for Multi-Hand Semantic Guided Grasp Generation}},
author = {Haosheng Li and Weixin Mao and Weipeng Deng and Chenyu Meng and Haoqiang Fan and Tiancai Wang and Yoshie Osamu and Ping Tan and Hongan Wang and Xiaoming Deng},
year = {2024},
eprint = {2412.08468},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2412.08468},
}
@misc{huang2024d,
title = {{3D-ViTac: Learning Fine-Grained Manipulation with Visuo-Tactile Sensing}},
author = {Binghao Huang and Yixuan Wang and Xinyi Yang and Yiyue Luo and Yunzhu Li},
year = {2024},
eprint = {2410.24091},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2410.24091},
}
An 'ear-in-hand' gripper records audio with vision during human demos in the wild; contact sound then helps policies when vision alone is ambiguous.“手上长耳朵”：夹爪在野外人类演示时同步录制音频与视觉；当视觉信息不足时，接触声帮助策略完成接触密集操作。
@misc{liu2024maniwav,
title = {{ManiWAV: Learning Robot Manipulation from In-the-Wild Audio-Visual Data}},
author = {Zeyi Liu and Cheng Chi and Eric Cousineau and Naveen Kuppuswamy and Benjamin Burchfiel and Shuran Song},
year = {2024},
eprint = {2406.19464},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2406.19464},
}
A recurrent graph network estimates object state and latent physics from visuo-tactile history; the learned dynamics model drives MPC for tasks like dense packing.循环图网络从视触觉历史估计物体状态与潜在物理属性，学到的动力学模型用于 MPC，完成密集装箱等任务。
@misc{ai2024robopack,
title = {{RoboPack: Learning Tactile-Informed Dynamics Models for Dense Packing}},
author = {Bo Ai and Stephen Tian and Haochen Shi and Yixuan Wang and Cheston Tan and Yunzhu Li and Jiajun Wu},
year = {2024},
eprint = {2407.01418},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2407.01418},
}
GR-1 — a GPT-style model that predicts both actions and future frames; pre-training on Ego4D video prediction before robot fine-tuning lifts CALVIN unseen-scene success from 53.3% to 85.4% (as reported).GR-1：GPT 式模型同时预测动作与未来画面；先在 Ego4D 上做视频预测预训练、再用机器人数据微调，论文报告 CALVIN 未见场景成功率从 53.3% 提升到 85.4%。
@misc{wu2023unleashing,
title = {{Unleashing Large-Scale Video Generative Pre-training for Visual Robot Manipulation}},
author = {Hongtao Wu and Ya Jing and Chilam Cheang and Guangzeng Chen and Jiafeng Xu and Xinghang Li and Minghuan Liu and Hang Li and Tao Kong},
year = {2023},
eprint = {2312.13139},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2312.13139},
}
@misc{liu2024ok,
title = {{OK-Robot: What Really Matters in Integrating Open-Knowledge Models for Robotics}},
author = {Peiqi Liu and Yaswanth Orru and Jay Vakil and Chris Paxton and Nur Muhammad Mahi Shafiullah and Lerrel Pinto},
year = {2024},
eprint = {2401.12202},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2401.12202},
}
An open-source stack for training manipulation policies with RL on a real arm — learned reward classifiers, reset-free training and an impedance controller.在真实机械臂上用强化学习训练操作策略的开源软件栈：学习式奖励分类器、免重置训练与阻抗控制器。
@misc{luo2024serl,
title = {{SERL: A Software Suite for Sample-Efficient Robotic Reinforcement Learning}},
author = {Jianlan Luo and Zheyuan Hu and Charles Xu and You Liang Tan and Jacob Berg and Archit Sharma and Stefan Schaal and Chelsea Finn and Abhishek Gupta and Sergey Levine},
year = {2024},
eprint = {2401.16013},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2401.16013},
}
DPPO — fine-tunes a pre-trained diffusion policy with PPO by treating the denoising steps as part of the MDP; reports stable gains over other RL fine-tuning methods and sim-to-real transfer.DPPO：把去噪步骤视为 MDP 的一部分，用 PPO 微调预训练扩散策略；报告称比其他强化学习微调方法更稳定，并实现了仿真到真机迁移。
@misc{ren2024diffusion,
title = {{Diffusion Policy Policy Optimization}},
author = {Allen Z. Ren and Justin Lidard and Lars L. Ankile and Anthony Simeonov and Pulkit Agrawal and Anirudha Majumdar and Benjamin Burchfiel and Hongkai Dai and Max Simchowitz},
year = {2024},
eprint = {2409.00588},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2409.00588},
}
Extends MimicGen to humanoids with two dexterous hands, turning 60 human demos into 21K simulated ones across bimanual tasks, plus a real-to-sim-to-real pipeline tested on real can sorting.把 MimicGen 扩展到双灵巧手人形机器人，从 60 条人类演示生成 2.1 万条双臂仿真演示，并给出在真实分拣罐子任务上验证的 real-to-sim-to-real 流程。
@misc{jiang2024dexmimicgen,
title = {{DexMimicGen: Automated Data Generation for Bimanual Dexterous Manipulation via Imitation Learning}},
author = {Zhenyu Jiang and Yuqi Xie and Kevin Lin and Zhenjia Xu and Weikang Wan and Ajay Mandlekar and Linxi Fan and Yuke Zhu},
year = {2024},
eprint = {2410.24185},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2410.24185},
}
Benchmark of 40 home tasks for mobile bimanual robots, from reaching to kitchen cleaning, each with human-collected demonstrations.面向移动双臂机器人的基准，含 40 个家庭任务（从抓取到厨房清洁），均配人类采集的演示。
@misc{chernyadev2024bigym,
title = {{BiGym: A Demo-Driven Mobile Bi-Manual Manipulation Benchmark}},
author = {Nikita Chernyadev and Nicholas Backshall and Xiao Ma and Yunfan Lu and Younggyo Seo and Stephen James},
year = {2024},
eprint = {2407.07788},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2407.07788},
}
GPU-accelerated Home Assistant Benchmark with realistic low-level control; the authors report over 3x the speed of prior magical-grasp versions.GPU 加速的 Home Assistant Benchmark，采用真实低层控制；作者报告速度超过此前“魔法抓取”实现的 3 倍。
@misc{shukla2024maniskill,
title = {{ManiSkill-HAB: A Benchmark for Low-Level Manipulation in Home Rearrangement Tasks}},
author = {Arth Shukla and Stone Tao and Hao Su},
year = {2024},
eprint = {2412.13211},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2412.13211},
}
@misc{pumacay2024colosseum,
title = {{THE COLOSSEUM: A Benchmark for Evaluating Generalization for Robotic Manipulation}},
author = {Wilbert Pumacay and Ishika Singh and Jiafei Duan and Ranjay Krishna and Jesse Thomason and Dieter Fox},
year = {2024},
eprint = {2402.08191},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2402.08191},
}
100 task categories over 2,000+ objects that need common sense, world knowledge and multi-step reasoning, not just instruction following; evaluates VLAs, embodied agents and VLMs in one suite.100 类任务、2000 多个物体，任务需要常识、世界知识与多步推理，而不只是照指令执行；在同一套件中评测 VLA、具身智能体与 VLM。
@misc{zhang2024vlabench,
title = {{VLABench: A Large-Scale Benchmark for Language-Conditioned Robotics Manipulation with Long-Horizon Reasoning Tasks}},
author = {Shiduo Zhang and Zhe Xu and Peiju Liu and Xiaopeng Yu and Yuan Li and Qinghui Gao and Zhaoye Fei and Zhangyue Yin and Zuxuan Wu and Yu-Gang Jiang and Xipeng Qiu},
year = {2024},
eprint = {2412.18194},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2412.18194},
}
Apple Vision Pro teleoperation of two dexterous robot hands with real-time retargeting, collision and singularity avoidance, and low-cost vibration haptics for the operator.用 Apple Vision Pro 实时遥操作双灵巧手，带实时重定向、碰撞与奇异点规避，并为操作者提供低成本振动触觉反馈。
@misc{ding2024bunny,
title = {{Bunny-VisionPro: Real-Time Bimanual Dexterous Teleoperation for Imitation Learning}},
author = {Runyu Ding and Yuzhe Qin and Jiyue Zhu and Chengzhe Jia and Shiqi Yang and Ruihan Yang and Xiaojuan Qi and Xiaolong Wang},
year = {2024},
eprint = {2407.03162},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2407.03162},
}
One modular teleop interface that mixes RGB-D body tracking, VR controllers, keyboard and joystick for whole-body mobile manipulators (Tiago++, HSR, Fetch), in sim and real.模块化遥操作接口，可组合 RGB-D 人体追踪、VR 手柄、键盘与摇杆，控制 Tiago++、HSR、Fetch 等移动操作机器人做全身动作，仿真和真机都支持。
@misc{dass2024telemoma,
title = {{TeleMoMa: A Modular and Versatile Teleoperation System for Mobile Manipulation}},
author = {Shivin Dass and Wensi Ai and Yuqian Jiang and Samik Singh and Jiaheng Hu and Ruohan Zhang and Peter Stone and Ben Abbatematteo and Roberto Martín-Martín},
year = {2024},
eprint = {2403.07869},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2403.07869},
}
HATO — bimanual setup from two UR5e arms and repurposed prosthetic hands with fingertip touch sensors, VR teleop for data collection, and visuotactile policies for tasks like pouring and handovers.HATO：用两台 UR5e 机械臂和改装的假肢手（指尖带触觉传感器）搭建双臂系统，用 VR 遥操作采集数据，并训练视觉-触觉策略完成倒酒、交接等任务。
@misc{lin2024learning,
title = {{Learning Visuotactile Skills with Two Multifingered Hands}},
author = {Toru Lin and Yu Zhang and Qiyang Li and Haozhi Qi and Brent Yi and Sergey Levine and Jitendra Malik},
year = {2024},
eprint = {2404.16823},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2404.16823},
}
An inexpensive open-source mobile base with powered casters, so it is fully holonomic; carries arbitrary arms and comes with phone teleoperation for collecting mobile-manipulation demos for imitation learning.低成本开源移动底盘，采用主动脚轮实现全向（完整约束）运动，可搭载任意机械臂，并配有手机遥操作界面，方便为模仿学习采集移动操作示教数据。
@misc{wu2024tidybot,
title = {{TidyBot++: An Open-Source Holonomic Mobile Manipulator for Robot Learning}},
author = {Jimmy Wu and William Chong and Robert Holmberg and Aaditya Prasad and Yihuai Gao and Oussama Khatib and Shuran Song and Szymon Rusinkiewicz and Jeannette Bohg},
year = {2024},
eprint = {2412.10447},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2412.10447},
}
Robot-free UMI gripper demos train the manipulation policy; a sim-trained whole-body controller tracks its end-effector trajectories on a quadruped. Reports >70% success on all tasks and zero-shot reuse of an arm policy.用手持 UMI 夹爪采集无需机器人的示教来训练操作策略，再由仿真训练的全身控制器在四足机器人上跟踪其输出的末端轨迹；报告所有任务成功率超过 70%，并能零样本复用原为固定机械臂训练的策略。
@misc{ha2024umi,
title = {{UMI-on-Legs: Making Manipulation Policies Mobile with Manipulation-Centric Whole-body Controllers}},
author = {Huy Ha and Yihuai Gao and Zipeng Fu and Jie Tan and Shuran Song},
year = {2024},
eprint = {2407.10353},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2407.10353},
}
@misc{shaw2024bimanual,
title = {{Bimanual Dexterity for Complex Tasks}},
author = {Kenneth Shaw and Yulong Li and Jiahui Yang and Mohan Kumar Srirama and Ray Liu and Haoyu Xiong and Russell Mendonca and Deepak Pathak},
year = {2024},
eprint = {2411.13677},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2411.13677},
}
LLMs write code that composes 3D affordance/constraint maps, which a planner turns into trajectories — no robot training.大模型编写代码组合 3D 可供性/约束地图，再由规划器生成轨迹，无需机器人训练。
@misc{huang2023voxposer,
title = {{VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models}},
author = {Wenlong Huang and Chen Wang and Ruohan Zhang and Yunzhu Li and Jiajun Wu and Li Fei-Fei},
year = {2023},
eprint = {2307.05973},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2307.05973},
}
Dobb-E collects home demonstrations with a cheap reacher-grabber "stick", pretrains on them, then adapts to a new household task from a few minutes of data.Dobb-E 用低成本的手持夹取“棒”采集家庭示范数据并预训练，在新家庭任务上只需几分钟数据即可适配。
@misc{shafiullah2023bringing,
title = {{On Bringing Robots Home}},
author = {Nur Muhammad Mahi Shafiullah and Anant Rai and Haritheja Etukuru and Yiqian Liu and Ishan Misra and Soumith Chintala and Lerrel Pinto},
year = {2023},
eprint = {2311.16098},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2311.16098},
}
RVT — re-renders the point cloud into a few virtual views and uses a multi-view Transformer to predict keyframe poses; reported to train and run much faster than voxel-based PerAct on RLBench. Repo also hosts RVT-2.RVT：把点云重新渲染为若干虚拟视角，再用多视角 Transformer 预测关键帧位姿；论文报告在 RLBench 上训练与推理都比基于体素的 PerAct 快得多。代码库同时包含 RVT-2。
@misc{goyal2023rvt,
title = {{RVT: Robotic View Transformer for 3D Object Manipulation}},
author = {Ankit Goyal and Jie Xu and Yijie Guo and Valts Blukis and Yu-Wei Chao and Dieter Fox},
year = {2023},
eprint = {2306.14896},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2306.14896},
}
Act3D samples 3D feature fields at adaptive resolutions to get high spatial precision without a costly high-resolution voxel grid.Act3D 用自适应分辨率采样三维特征场，无需昂贵的高分辨率体素也能获得高空间精度。
@misc{li2023crossway,
title = {{Crossway Diffusion: Improving Diffusion-based Visuomotor Policy via Self-supervised Learning}},
author = {Xiang Li and Varun Belagali and Jinghuan Shang and Michael S. Ryoo},
year = {2023},
eprint = {2307.01849},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2307.01849},
}
Uses an LLM plus sampling-based planners to generate language-labelled robot data, then distils it into one language-conditioned visuomotor policy.用大模型加采样规划器生成带语言标注的机器人数据，再蒸馏为一个语言条件视觉运动策略。
@misc{ha2023scaling,
title = {{Scaling Up and Distilling Down: Language-Guided Robot Skill Acquisition}},
author = {Huy Ha and Pete Florence and Shuran Song},
year = {2023},
eprint = {2307.14535},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2307.14535},
}
SuSIE uses a pretrained image-editing diffusion model as a high-level planner that proposes intermediate subgoal images.SuSIE 用预训练图像编辑扩散模型作高层规划器，提出中间子目标图像。
@misc{black2023zero,
title = {{Zero-Shot Robotic Manipulation with Pretrained Image-Editing Diffusion Models}},
author = {Kevin Black and Mitsuhiko Nakamoto and Pranav Atreya and Homer Walke and Chelsea Finn and Aviral Kumar and Sergey Levine},
year = {2023},
eprint = {2310.10639},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2310.10639},
}
@misc{hu2023look,
title = {{Look Before You Leap: Unveiling the Power of GPT-4V in Robotic Vision-Language Planning}},
author = {Yingdong Hu and Fanqi Lin and Tong Zhang and Li Yi and Yang Gao},
year = {2023},
eprint = {2311.17842},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2311.17842},
}
@misc{gao2023physically,
title = {{Physically Grounded Vision-Language Models for Robotic Manipulation}},
author = {Jensen Gao and Bidipta Sarkar and Fei Xia and Ted Xiao and Jiajun Wu and Brian Ichter and Anirudha Majumdar and Dorsa Sadigh},
year = {2023},
eprint = {2309.02561},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2309.02561},
}
RotateIt trains in simulation with privileged object information, then distills to a policy that rotates diverse objects in hand using vision, touch and proprioception on a real hand.RotateIt 先在仿真中利用特权物体信息训练，再蒸馏为仅用视觉、触觉和本体感觉的策略，在真实灵巧手上旋转多种物体。
@misc{qi2023general,
title = {{General In-Hand Object Rotation with Vision and Touch}},
author = {Haozhi Qi and Brent Yi and Sudharshan Suresh and Mike Lambeta and Yi Ma and Roberto Calandra and Jitendra Malik},
year = {2023},
eprint = {2309.09979},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2309.09979},
}
Covers a robot hand with cheap binary (contact / no contact) touch sensors and trains an RL policy in simulation that rotates unseen objects in hand on a real robot without any vision.在灵巧手上铺设廉价的二值（有/无接触）触觉传感器，在仿真中用强化学习训练策略，迁移到真机后无需视觉即可在手中旋转未见过的物体。
@misc{yin2023rotating,
title = {{Rotating without Seeing: Towards In-hand Dexterity through Touch}},
author = {Zhao-Heng Yin and Binghao Huang and Yuzhe Qin and Qifeng Chen and Xiaolong Wang},
year = {2023},
eprint = {2303.10880},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2303.10880},
}
Turns tactile readings into points in the same 3D frame as the camera point cloud and the hand model, so one visuotactile point-cloud input drives in-hand object rotation.将触觉读数转换为与相机点云、手部模型处于同一三维坐标系的点，用统一的视触觉点云输入驱动手内物体旋转。
@misc{yuan2023robot,
title = {{Robot Synesthesia: In-Hand Manipulation with Visuotactile Sensing}},
author = {Ying Yuan and Haichuan Che and Yuzhe Qin and Binghao Huang and Zhao-Heng Yin and Kang-Won Lee and Yi Wu and Soo-Chul Lim and Xiaolong Wang},
year = {2023},
eprint = {2312.01853},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2312.01853},
}
NeuralFeels fuses vision and touch from a multi-fingered hand into an online neural field, reconstructing an object's shape and tracking its pose while the hand manipulates it.NeuralFeels 将多指手的视觉与触觉在线融合为神经场，在操作过程中同时重建物体形状并跟踪其位姿。
@misc{suresh2023neural,
title = {{Neural feels with neural fields: Visuo-tactile perception for in-hand manipulation}},
author = {Sudharshan Suresh and Haozhi Qi and Tingfan Wu and Taosha Fan and Luis Pineda and Mike Lambeta and Jitendra Malik and Mrinal Kalakrishnan and Roberto Calandra and Michael Kaess and Joseph Ortiz and Mustafa Mukadam},
year = {2023},
eprint = {2312.13469},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2312.13469},
}
Learns tactile policies from demonstrations given with human hands, closing the gap of teleop where operators steer by sight while the policy needs touch.从人手直接演示学习触觉策略，弥补遥操作中“人靠视觉控制、策略却要用触觉”的落差。
@misc{yu2023mimictouch,
title = {{MimicTouch: Leveraging Multi-modal Human Tactile Demonstrations for Contact-rich Manipulation}},
author = {Kelin Yu and Yunhai Han and Qixian Wang and Vaibhav Saxena and Danfei Xu and Ye Zhao},
year = {2023},
eprint = {2310.16917},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2310.16917},
}
TRILL collects VR demonstrations and uses a whole-body controller to turn task-space commands into joint torques, for data-efficient loco-manipulation imitation.TRILL 用 VR 采集演示，由全身控制器把任务空间指令转为关节力矩，实现数据高效的人形移动操作模仿。
@misc{seo2023deep,
title = {{Deep Imitation Learning for Humanoid Loco-manipulation through Human Teleoperation}},
author = {Mingyo Seo and Steve Han and Kyutae Sim and Seung Hyeon Bang and Carlos Gonzalez and Luis Sentis and Yuke Zhu},
year = {2023},
eprint = {2309.01952},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2309.01952},
}
VAMP — vectorizes collision checking with CPU SIMD so sampling-based planners like RRT-Connect solve arm motion-planning queries in microseconds to milliseconds, no GPU needed.VAMP：用 CPU SIMD 指令把碰撞检测向量化，使 RRT-Connect 等基于采样的规划器在微秒到毫秒级求解机械臂运动规划，无需 GPU。
Discretizes each action dimension and predicts Q-values per dimension autoregressively with a Transformer, plus a conservative regularizer; trained offline on large real robot datasets mixing demos and autonomous data.将每个动作维度离散化，用 Transformer 自回归地逐维预测 Q 值，并加入保守正则；在混合演示与自主采集数据的大规模真实机器人数据集上离线训练。
@misc{chebotar2023q,
title = {{Q-Transformer: Scalable Offline Reinforcement Learning via Autoregressive Q-Functions}},
author = {Yevgen Chebotar and Quan Vuong and Alex Irpan and Karol Hausman and Fei Xia and Yao Lu and Aviral Kumar and Tianhe Yu and Alexander Herzog and Karl Pertsch and Keerthana Gopalakrishnan and Julian Ibarz and Ofir Nachum and Sumedh Sontakke and Grecia Salazar and Huong T Tran and Jodilyn Peralta and Clayton Tan and Deeksha Manjunath and Jaspiar Singht and Brianna Zitkovich and Tomas Jackson and Kanishka Rao and Chelsea Finn and Sergey Levine},
year = {2023},
eprint = {2309.10150},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2309.10150},
}
Turns a handful of human demos into thousands by splitting them into object-centric segments and replaying those segments in new scenes; widely reused to scale imitation data.把少量人类演示切分为以物体为中心的片段，并在新场景中变换重放，从而扩增为数千条演示；被广泛用于扩充模仿学习数据。
@misc{mandlekar2023mimicgen,
title = {{MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrations}},
author = {Ajay Mandlekar and Soroush Nasiriany and Bowen Wen and Iretiayo Akinola and Yashraj Narang and Linxi Fan and Yuke Zhu and Dieter Fox},
year = {2023},
eprint = {2310.17596},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2310.17596},
}
Camera-only teleoperation for arm + dexterous hand pairs - tracks the operator's hand and retargets it to many different robot hands, in sim or on real hardware.仅用相机即可遥操作"机械臂 + 灵巧手"：跟踪操作者手部并重定向到多种不同的机器人手，可用于仿真或真机。
@misc{qin2023anyteleop,
title = {{AnyTeleop: A General Vision-Based Dexterous Robot Arm-Hand Teleoperation System}},
author = {Yuzhe Qin and Wei Yang and Binghao Huang and Karl Van Wyk and Hao Su and Xiaolong Wang and Yu-Wei Chao and Dieter Fox},
year = {2023},
eprint = {2307.04577},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2307.04577},
}
A portable dual-arm exoskeleton (about $600 in parts, per the paper) for joint-level teleop and cheap in-the-wild demos; reports few-minute teleop plus wild data rivals much longer teleop-only training.便携式双臂外骨骼（论文称零件成本约 600 美元），可做关节级遥操作并低成本采集野外示教；报告称少量遥操作数据加野外数据即可达到甚至超过长时间纯遥操作训练的效果。
@misc{fang2023airexo,
title = {{AirExo: Low-Cost Exoskeletons for Learning Whole-Arm Manipulation in the Wild}},
author = {Hongjie Fang and Hao-Shu Fang and Yiming Wang and Jieji Ren and Jingjing Chen and Ruo Zhang and Weiming Wang and Cewu Lu},
year = {2023},
eprint = {2309.14975},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2309.14975},
}
@misc{chen2022visual,
title = {{Visual Dexterity: In-Hand Reorientation of Novel and Complex Object Shapes}},
author = {Tao Chen and Megha Tippur and Siyang Wu and Vikash Kumar and Edward Adelson and Pulkit Agrawal},
year = {2022},
eprint = {2211.11744},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2211.11744},
}
@misc{nair2022r,
title = {{R3M: A Universal Visual Representation for Robot Manipulation}},
author = {Suraj Nair and Aravind Rajeswaran and Vikash Kumar and Chelsea Finn and Abhinav Gupta},
year = {2022},
eprint = {2203.12601},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2203.12601},
}
Over a million physically validated ShadowHand grasps on thousands of objects, synthesized by optimizing a differentiable force-closure energy; a standard training set for dexterous grasp models.通过优化可微的力封闭能量合成、并经物理仿真验证的百万级 ShadowHand 抓取，覆盖数千个物体，是灵巧抓取模型的常用训练集。
@misc{wang2022dexgraspnet,
title = {{DexGraspNet: A Large-Scale Robotic Dexterous Grasp Dataset for General Objects Based on Simulation}},
author = {Ruicheng Wang and Jialiang Zhang and Jiayi Chen and Yinzhen Xu and Puhao Li and Tengyu Liu and He Wang},
year = {2022},
eprint = {2210.02697},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2210.02697},
}
PerAct — a Perceiver Transformer over voxelized RGB-D observations that predicts the next keyframe gripper pose as voxel classification; one language-conditioned policy for many RLBench and real tasks.PerAct：在体素化的 RGB-D 观测上运行 Perceiver Transformer，把下一关键帧夹爪位姿当作体素分类来预测；一个语言条件策略即可覆盖众多 RLBench 与真机任务。
Fuses camera, contact microphone and GelSight touch with self-attention in an imitation-learned policy, studied on dense packing and pouring tasks.在模仿学习策略中用自注意力融合相机、接触麦克风和 GelSight 触觉，在密集装箱与倾倒任务上做了研究。
@misc{li2022see,
title = {{See, Hear, and Feel: Smart Sensory Fusion for Robotic Manipulation}},
author = {Hao Li and Yizhi Zhang and Junzhe Zhu and Shaoxiong Wang and Michelle A Lee and Huazhe Xu and Edward Adelson and Li Fei-Fei and Ruohan Gao and Jiajun Wu},
year = {2022},
eprint = {2212.03858},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2212.03858},
}
LLMs write Python that calls perception and control APIs, turning instructions into reactive robot programs.大模型编写调用感知与控制 API 的 Python 代码，把指令变成可响应的机器人程序。
@misc{liang2022code,
title = {{Code as Policies: Language Model Programs for Embodied Control}},
author = {Jacky Liang and Wenlong Huang and Fei Xia and Peng Xu and Karol Hausman and Brian Ichter and Pete Florence and Andy Zeng},
year = {2022},
eprint = {2209.07753},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2209.07753},
}
A neural policy trained on millions of planner-generated trajectories that moves an arm to goals straight from point clouds.在数百万条规划器生成的轨迹上训练的神经策略，可直接根据点云把机械臂移动到目标位置。
2.4M egocentric RGB-D frames (4,000+ sequences, 800 object instances in 16 categories, 610 rooms) of people handling rigid and articulated objects, with segmentation, hand and object pose labels.240 万帧第一视角 RGB-D（4000 多段序列、16 类 800 个物体实例、610 个房间），记录人操作刚体与铰接物体，带分割、手部与物体位姿标注。
@misc{liu2022hoi,
title = {{HOI4D: A 4D Egocentric Dataset for Category-Level Human-Object Interaction}},
author = {Yunze Liu and Yun Liu and Che Jiang and Kangbo Lyu and Weikang Wan and Hao Shen and Boqiang Liang and Zhoujie Fu and He Wang and Li Yi},
year = {2022},
eprint = {2203.01577},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2203.01577},
}
Affordance labels for ~1,800 household objects plus 50K intent-oriented hand-object interactions recorded on 100 of them; a common source for dexterous grasp synthesis. Follow-up OakInk2 covers bimanual tasks.为约 1800 个日常物体标注可供性，并在其中 100 个物体上记录 5 万次带意图的手-物交互；常用于灵巧抓取生成。后续 OakInk2 扩展到双手任务。
@misc{yang2022oakink,
title = {{OakInk: A Large-scale Knowledge Repository for Understanding Hand-Object Interaction}},
author = {Lixin Yang and Kailin Li and Xinyu Zhan and Fei Wu and Anran Xu and Liu Liu and Cewu Lu},
year = {2022},
eprint = {2203.15709},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2203.15709},
}
2.1M video frames of two hands manipulating articulated objects (scissors, laptops…) with accurate 3D hand and object meshes from motion capture; a reference for bimanual hand-object reconstruction.210 万帧双手操作铰接物体（剪刀、笔记本电脑等）的视频，借助动作捕捉提供精确的手与物体三维网格；双手-物体重建的常用基准。
@misc{fan2022arctic,
title = {{ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulation}},
author = {Zicong Fan and Omid Taheri and Dimitrios Tzionas and Muhammed Kocabas and Manuel Kaufmann and Michael J. Black and Otmar Hilliges},
year = {2022},
eprint = {2204.13662},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2204.13662},
}
CLIPort — combines CLIP's semantic "what" stream with Transporter's spatial "where" stream for language-conditioned tabletop manipulation; an early bridge from web vision-language models to robot actions.CLIPort：把 CLIP 提供的语义"是什么"通路与 Transporter 的空间"在哪里"通路结合，实现语言条件的桌面操作；是较早把网络视觉语言模型接入机器人动作的工作。
@misc{shridhar2021cliport,
title = {{CLIPort: What and Where Pathways for Robotic Manipulation}},
author = {Mohit Shridhar and Lucas Manuelli and Dieter Fox},
year = {2021},
eprint = {2109.12098},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2109.12098},
}
Contact-GraspNet — predicts 6-DoF parallel-jaw grasps for a whole cluttered scene from one depth view by treating observed points as grasp contacts, which shrinks the problem to 4 DoF per point.Contact-GraspNet：从单帧深度图为整个杂乱场景预测 6 自由度平行夹爪抓取；把观测到的点当作抓取接触点，使每个点只需学习 4 个自由度。
Adds fast rigid-body physics and an interactive apartment dataset (ReplicaCAD) to Habitat, plus a home-rearrangement benchmark for mobile manipulators.为 Habitat 加入快速刚体物理与可交互公寓数据集 ReplicaCAD，并提出面向移动操作机器人的家居物品重排基准。
@misc{szot2021habitat,
title = {{Habitat 2.0: Training Home Assistants to Rearrange their Habitat}},
author = {Andrew Szot and Alex Clegg and Eric Undersander and Erik Wijmans and Yili Zhao and John Turner and Noah Maestre and Mustafa Mukadam and Devendra Chaplot and Oleksandr Maksymets and Aaron Gokaslan and Vladimir Vondrus and Sameer Dharur and Franziska Meier and Wojciech Galuba and Angel Chang and Zsolt Kira and Vladlen Koltun and Jitendra Malik and Manolis Savva and Dhruv Batra},
year = {2021},
eprint = {2106.14405},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/2106.14405},
}
@misc{chao2021dexycb,
title = {{DexYCB: A Benchmark for Capturing Hand Grasping of Objects}},
author = {Yu-Wei Chao and Wei Yang and Yu Xiang and Pavlo Molchanov and Ankur Handa and Jonathan Tremblay and Yashraj S. Narang and Karl Van Wyk and Umar Iqbal and Stan Birchfield and Jan Kautz and Dieter Fox},
year = {2021},
eprint = {2104.04631},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2104.04631},
}
Transporter Nets — pick-and-place as spatial matching of deep features cropped around the pick point; very sample-efficient on rearrangement tasks, released with the Ravens benchmark.Transporter 网络：把抓取-放置建模为以抓取点为中心的深度特征的空间匹配，在重排类任务上样本效率很高，并随论文发布 Ravens 基准。
@misc{zeng2020transporter,
title = {{Transporter Networks: Rearranging the Visual World for Robotic Manipulation}},
author = {Andy Zeng and Pete Florence and Jonathan Tompson and Stefan Welker and Jonathan Chien and Maria Attarian and Travis Armstrong and Ivan Krasin and Dan Duong and Ayzaan Wahid and Vikas Sindhwani and Johnny Lee},
year = {2020},
eprint = {2010.14406},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2010.14406},
}
Open-source Gym-style benchmark for manipulating deformable objects such as cloth, rope and liquids; shows that high-dimensional, partially observable states make these tasks hard for standard RL.开源、Gym 接口的可变形物体操作基准（布料、绳子、液体等）；实验表明高维且部分可观测的状态让这类任务对常规强化学习方法很困难。
@misc{lin2020softgym,
title = {{SoftGym: Benchmarking Deep Reinforcement Learning for Deformable Object Manipulation}},
author = {Xingyu Lin and Yufei Wang and Jake Olkin and David Held},
year = {2020},
eprint = {2011.07215},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2011.07215},
}
@misc{openai2019solving,
title = {{Solving Rubik's Cube with a Robot Hand}},
author = {OpenAI and Ilge Akkaya and Marcin Andrychowicz and Maciek Chociej and Mateusz Litwin and Bob McGrew and Arthur Petron and Alex Paino and Matthias Plappert and Glenn Powell and Raphael Ribas and Jonas Schneider and Nikolas Tezak and Jerry Tworek and Peter Welinder and Lilian Weng and Qiming Yuan and Wojciech Zaremba and Lei Zhang},
year = {2019},
eprint = {1910.07113},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/1910.07113},
}
A Shadow Hand reorients a cube using RL trained purely in randomized simulation — a sim-to-real milestone.完全在随机化仿真中用强化学习训练，Shadow 灵巧手在真实世界翻转方块，是 sim-to-real 里程碑。
@misc{openai2018learning,
title = {{Learning Dexterous In-Hand Manipulation}},
author = {OpenAI and Marcin Andrychowicz and Bowen Baker and Maciek Chociej and Rafal Jozefowicz and Bob McGrew and Jakub Pachocki and Arthur Petron and Matthias Plappert and Glenn Powell and Alex Ray and Jonas Schneider and Szymon Sidor and Josh Tobin and Peter Welinder and Lilian Weng and Wojciech Zaremba},
year = {2018},
eprint = {1808.00177},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/1808.00177},
}
Learns a compact vision + force/torque representation with self-supervised objectives, then trains a peg-insertion policy with RL on top of it on a real robot.用自监督目标学习视觉 + 力/力矩的紧凑联合表征，再在其上用强化学习在真实机器人上训练插孔策略。
@misc{lee2018making,
title = {{Making Sense of Vision and Touch: Self-Supervised Learning of Multimodal Representations for Contact-Rich Tasks}},
author = {Michelle A. Lee and Yuke Zhu and Krishnan Srinivasan and Parth Shah and Silvio Savarese and Li Fei-Fei and Animesh Garg and Jeannette Bohg},
year = {2018},
eprint = {1810.10191},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/1810.10191},
}
Off-policy Q-learning with a cross-entropy-method action optimizer, trained on several hundred thousand real grasp attempts collected by a fleet of robots; reports 96% grasp success on unseen objects.使用交叉熵方法优化动作的离策略 Q 学习，在机器人集群采集的数十万次真实抓取上训练；报告对未见物体的抓取成功率为 96%。
@misc{kalashnikov2018qt,
title = {{QT-Opt: Scalable Deep Reinforcement Learning for Vision-Based Robotic Manipulation}},
author = {Dmitry Kalashnikov and Alex Irpan and Peter Pastor and Julian Ibarz and Alexander Herzog and Eric Jang and Deirdre Quillen and Ethan Holly and Mrinal Kalakrishnan and Vincent Vanhoucke and Sergey Levine},
year = {2018},
eprint = {1806.10293},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/1806.10293},
}
The YCB object set — 73 everyday objects with scans, textured meshes and simulation models, plus benchmarking protocols; widely used in grasping and pose-estimation work.YCB 物体集：73 个日常物体，附扫描数据、带纹理网格与仿真模型，以及基准测试规程；被抓取与位姿估计研究广泛使用。
@misc{calli2015benchmarking,
title = {{Benchmarking in Manipulation Research: The YCB Object and Model Set and Benchmarking Protocols}},
author = {Berk Calli and Aaron Walsman and Arjun Singh and Siddhartha Srinivasa and Pieter Abbeel and Aaron M. Dollar},
year = {2015},
eprint = {1502.03143},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/1502.03143},
}
Heng Fang, Zehui Guo, Yuehan Zhang, Mingyang Du et al.
Abstract摘要
Trained action policies write manipulation skill into network weights, tying what is learned to the distribution it is learned from. In-context learning (ICL) avoids this coupling but has not matched trained policies, because a demonstration recorded in another scene carries no feedback from the world the robot currently faces. We observe that what an agent learns in context is semantic, recording task steps and object roles rather than visual appearance, so experience gathered in one world transfers to another. Based on this insight, we propose World In-Context Learning (WICL), a paradigm in …
@misc{fang2026world,
title = {{World In-Context Learning for Agentic Robot Manipulation}},
author = {Heng Fang and Zehui Guo and Yuehan Zhang and Mingyang Du and Dingkang Liang and Xiang Bai},
year = {2026},
eprint = {2611.12885},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2611.12885},
}
Bing Han, Yichuan Peng, Haoyu Dong, Yuanmeng Hu et al.
Abstract摘要
Tactile sensing is an indispensable modality for contact-rich dexterous manipulation, a capability that hints at the ultimate frontiers of embodied intelligence. However, commercially available tactile sensors are still at a relatively early stage of development, with each type of sensor offering irreplaceable strengths alongside its own drawbacks. A key challenge is thus to free tactile understanding from sensor-specific physics, so that embodied models can effectively leverage heterogeneous tactile data. To this end, we develop LingBot-Touch, a tactile foundation model characterized by the …
@misc{han2026learning,
title = {{Learning Heterogeneous Tactile Representation for General Dexterous Manipulation}},
author = {Bing Han and Yichuan Peng and Haoyu Dong and Yuanmeng Hu and Yongyi Jia and Fangjing Wang and Kecheng Zheng and Yiming Zhang and others},
year = {2026},
eprint = {2611.12806},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12806},
}
Elias Gläßer, Gülhan Şikaroğlu, Pavlo Kupyn, Marc Pollefeys et al.
Abstract摘要
A challenging bottleneck in manipulation research is reproducibility and comparability of methods in real-world environments. Since manipulation is an interactive task, it cannot be evaluated on a static dataset of observations. Benchmarking in simulation is a common remedy, but cannot cover the whole complexity of real-world tests. This paper proposes a new paradigm of stochastic challenges. Like the well-known YCB objects, it is based on reproducible real-world scenes and therefore allows system-level innovation and comparability between different platforms. Due to their stochastic nature, …
@misc{glaer2026benchmarking,
title = {{Benchmarking Real-World Robotic Manipulation through Stochastic Challenges}},
author = {Elias Gläßer and Gülhan Şikaroğlu and Pavlo Kupyn and Marc Pollefeys and Zuria Bauer and Hermann Blum},
year = {2026},
eprint = {2611.12804},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12804},
}
Emanuele Cuzzocrea, Giuseppina Iannotti, Luca Rossini, Arturo Laurenzi et al.
Abstract摘要
Robots are increasingly expected to execute natural-language manipulation tasks in real-world environments. This is particularly relevant in household assistance and logistics, where clutter requires reasoning about both task objectives and execution feasibility. This challenge becomes even more pronounced for bimanual humanoid platforms, where coordinating two manipulators introduces additional interaction and kinematic constraints. We present a zero-shot vision-language framework for execution-aware bimanual manipulation that jointly reasons about task objectives and physical feasibility. …
Ruilin Zhang, Xuanshuo Liu, Junzhe Dai, Kefei Wu et al.
Abstract摘要
Modern visuotactile sensors primarily sense contact by imaging the deformation of a compliant interface. Using techniques such as photometric stereo, they can recover detailed contact geometry, often represented as surface normals or depth. However, force-related information typically still needs to be inferred indirectly through calibration, geometric reconstruction, or learned mappings. Mechanochromic photonic-crystal (PC) materials offer an alternative physical encoding mechanism: contact loading changes their periodic microstructure and spectral response, directly transforming local …
@misc{zhang2026pctac,
title = {{PCTac: Mechanochromic Photonic-Crystal Tactile Sensor for Dexterous Manipulation}},
author = {Ruilin Zhang and Xuanshuo Liu and Junzhe Dai and Kefei Wu and Xuhao Qin and Ziyuan Jiao and Wanlin Li and Yurui Qu and others},
year = {2026},
eprint = {2611.12383},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12383},
}
Meraj Mammadov, Pedro Zuidberg Dos Martires, Johannes Andreas Stork
Abstract摘要
Reinforcement learning (RL) from raw sensory inputs such as images or onboard sensors can be challenging due to high-dimensional observations and sparse rewards. A common strategy is to train a teacher policy with access to privileged state information and then distill it into a student policy that acts from raw inputs alone. However, when the teacher relies on information unavailable to the student, exact imitation may be impossible, creating an irreducible imitation gap. Existing approaches typically mitigate this mismatch through careful reward shaping or additional reinforcement learning …
@misc{mammadov2026rail,
title = {{RAIL: Representation-Aligned Imitation Learning for Student-Compatible Teacher Policies}},
author = {Meraj Mammadov and Pedro Zuidberg Dos Martires and Johannes Andreas Stork},
year = {2026},
eprint = {2611.12251},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/2611.12251},
}
Jinghe Wang, Xinrui Cao, Duo Wu, Chenghao Gu et al.
Abstract摘要
Robotic manipulation has advanced rapidly in mapping visual observations and language instructions to actions. However, reliable manipulation in open-world environments requires robust spatial understanding: the ability to identify targets in complex layouts (spatial reasoning) and adapt to scenes that continuously evolve through interaction (spatial memory). Yet existing embodied benchmarks fall short in evaluating spatial understanding: they often introduce strong visual cues that bypass spatial reasoning, and focus on static scenes that offer little insight into an agent's capacity for …
@misc{wang2026geomind,
title = {{GeoMind: A Benchmark for Spatial Understanding in Robotic Manipulation}},
author = {Jinghe Wang and Xinrui Cao and Duo Wu and Chenghao Gu and Yong Zhong and Linjia Kang and Tianyi Xiong and Zhi Wang},
year = {2026},
eprint = {2611.12175},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12175},
}
Xianlong Mai, Rui Yu, Rui Chen, Daniele Leonardis et al.
Abstract摘要
Hand kinesthetic feedback is crucial for restoring and enhancing human perception of force and displacement in robotic hand teleoperation. Wide perception range of force feedback is vital in rendering objects of different stiffness, especially for high-stiffness rendering. However, existing approaches struggle to achieve wide range of force feedback while maintaining compact and wearable design. This paper presents a hand exoskeleton system employing magnetorheological (MR) brakes to provide a wide range kinesthetic feedback during teleoperation. The MR brake provides a braking torque ranging …
@misc{mai2026compact,
title = {{Compact and Wide-Force-Range Magnetorheological Hand Exoskeleton for Kinesthetic Feedback in Teleoperation}},
author = {Xianlong Mai and Rui Yu and Rui Chen and Daniele Leonardis and Wenbo Li and Domenico Chiaradia and Antonio Frisoli and Shuaishuai Sun},
year = {2026},
eprint = {2611.12130},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12130},
}
Jinyi Chen, Hongyi Li, Lei Jiang, Xianwei Liu et al.
Abstract摘要
Dexterous hands enable versatile manipulation, but policies trained in simulation often degrade on hardware owing to non-idealities, and the high-dimensional action space makes real-world adaptation costly. Piano playing, which requires precise and coordinated multifinger motion, offers a representative testbed. We propose GAP (Generalized Adaptation Prior for Dexterous Piano Playing), a sample-efficient adaptation framework that learns how to adapt, rather than how to act, during simulation pretraining and transfers this adaptation prior to hardware. Between repeated executions, GAP uses …
@misc{chen2026gap,
title = {{GAP: Generalized Adaptation Prior for Dexterous Piano Playing}},
author = {Jinyi Chen and Hongyi Li and Lei Jiang and Xianwei Liu and Yanhong Liang and Yulin Chen and Yucheng Tao and Yongbin Jin and others},
year = {2026},
eprint = {2611.12111},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12111},
}
The motion of an articulated part in cluttered surroundings is constrained not only by an object's intrinsic joint limits but also by extrinsic obstacles. However, existing approaches primarily consider object articulation properties alone. Here we present Motion Constraint Prediction Network (MoCo-Net), a two-stream neural network that jointly predicts the continuous motion ranges of articulated parts and obstacle heatmaps highlighting potential collision regions from partial-view point cloud observation. By capturing both internal and external motion constraints, MoCo-Net produces more …
Rumi Loghmani, Abraham George, Amir Barati Farimani
Abstract摘要
Dexterous grasping in robotics relies on distinct hand configurations depending on object geometry. We introduce BetaGrasp, a 3D diffusion policy that conditions a DP3-style point-cloud policy on a four-class grasp label (crimp, sloper, pinch, and jug), running on a Franka arm with a 16-DoF LEAP hand. We also introduce a climbing-hold benchmark: a spring testbed whose ratchet gives a slip-force reading in newtons. In our analysis the policy's point-cloud encoder did not separate the four holds, so the scan alone gave it little to choose a grasp from. Without the label (a DP3-style baseline), …
@misc{loghmani2026grasp,
title = {{Grasp-Taxonomy Conditioning as a Mode Switch for Dexterous Grasping of Climbing Holds}},
author = {Rumi Loghmani and Abraham George and Amir Barati Farimani},
year = {2026},
eprint = {2611.12017},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.12017},
}
Jingxuan Chen, Junshi Luo, Chencheng Dong, Xiaojian Nie et al.
Abstract摘要
Recent Vision-Language-Action (VLA) models have begun incorporating thermal sensing, but existing approaches typically treat thermal images as ordinary visual inputs, providing limited support for temperature-aware relational reasoning in manipulation. To address this limitation, we propose Thermo-VLA with a ThermoRes Encoder, which decomposes thermal images into cold and hot attribute representations and injects them into the RGB token through language-conditioned residual modulation, enabling temperature-driven target selection and continuous manipulation. We further construct six …
@misc{chen2026thermo,
title = {{Thermo-VLA: A Thermal-Aware Vision-Language-Action Model with Gated Residual Modulation for Robotic Manipulation}},
author = {Jingxuan Chen and Junshi Luo and Chencheng Dong and Xiaojian Nie and Tongge Huang and Qiang Huang},
year = {2026},
eprint = {2611.11981},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11981},
}
Flipping is a fundamental robotic manipulation skill that requires reorienting a target object in a stable pose on a planar support surface by rotating it 180$^\circ$ about an axis parallel to the support surface -- think of flipping a coin from heads to tails. Although flipping is a common objective in industrial tasks such as parts inspection, mechanical assembly, and tool use, coding a bimanual robot with grippers to flip diverse objects is a significant challenge. In this paper, we show that a coding agent can autonomously develop and refine object-flipping strategies through iterative …
Bingzhang Wang, Nathan Zhao, Jialeng Ni, Jiajin Zheng et al.
Abstract摘要
Vision-language-action (VLA) models pair a vision-language backbone with an action head, but in closed-loop control the backbone sits on the control-critical path: while it computes, the robot keeps moving, so each executed action is conditioned on an increasingly stale observation. A larger backbone therefore directly blocks reactivity, especially in humanoid dexterous manipulation. We present DASH, a Decoupled Asynchronous System for Humanoids that removes this coupling. DASH splits the policy into a slow vision-language model (System 2) and a fast flow-matching action head (System 1) that …
@misc{wang2026decoupled,
title = {{Decoupled Asynchronous System for Humanoids (DASH): Latency-Agnostic Scaling of Vision-Language-Action Models for Humanoid Manipulation}},
author = {Bingzhang Wang and Nathan Zhao and Jialeng Ni and Jiajin Zheng and Yin Zhang and Kehua Chen and Jian Yao},
year = {2026},
eprint = {2611.11854},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11854},
}
Sung-Wook Lee, Xuhui Kang, Dylan Losey, Yen-Ling Kuo
Abstract摘要
While imitation learning has made substantial progress toward generalizable robot manipulation through scaling, viewpoint robustness remains a major bottleneck. Visuomotor policies rely heavily on visual inputs, yet images provide an inherently ambiguous signal under viewpoint change: similar scenes may appear markedly different across viewpoints, while distinct states may share certain visual overlaps. A policy must therefore infer not only the task-relevant state, but also how it is visually projected under a particular camera configuration. This entanglement of viewpoint-dependent …
@misc{lee2026pair,
title = {{PAIR: Proprioception-Aligned Image-Ray Representations for Viewpoint-Robust Manipulation}},
author = {Sung-Wook Lee and Xuhui Kang and Dylan Losey and Yen-Ling Kuo},
year = {2026},
eprint = {2611.11798},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11798},
}
Yunheng Wang, Jiaxi Zhang, Wei Lu, Yuetong Fang et al.
Abstract摘要
Humanoid loco-manipulation requires continuous coordination of high-dimensional whole-body motion to realize task objectives. While vision-language-action models and world action models transfer semantic and visual-dynamics priors to robot control, these priors do not necessarily capture the structure of robot motion. We introduce Kinematic Action Models (KAMs), a paradigm that treats whole-body motion as a direct pretraining target and uses a pretrained motion generator as the primary action-generation backbone. KAMs ground learned temporal and cross-body coordination priors through task …
@misc{wang2026kimodo,
title = {{Kimodo-Policy: Learning Kinematic Action Models for Humanoid Loco-Manipulation}},
author = {Yunheng Wang and Jiaxi Zhang and Wei Lu and Yuetong Fang and Taowen Wang and Yixiao Feng and Shuning Zhang and Junzhe Xu and others},
year = {2026},
eprint = {2611.11554},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11554},
}
Estimating the motion of deformable bodies in soft robotics from proprioceptive sensing is challenging due to partial observability, as internal observations provide only indirect access to global motion, contact, and shape evolution. In such settings, instantaneous observations are insufficient, and accurate estimation requires temporally consistent latent representations. This paper studies deformable motion estimation under proprioceptive partial observability from a latent dynamics perspective. We argue that a key requirement for long-horizon estimation is latent temporal consistency, …
Zhongxi Chen, Yi Yang, Xinyi Chen, Mingzhe Ye et al.
Abstract摘要
Dexterous manipulation requires rich spatial foresight over both the surrounding scene and the robot hand. Existing world-action models often acquire such predictive knowledge through explicit future prediction. We introduce DexForesight, a framework that learns spatial foresight implicitly by injecting dynamics-aware world representations into the latent space of a dexterous policy. During training, DexForesight learns an action-conditioned spatial world model from offline scene depth, kinematic hand geometry, and demonstration trajectories. The world model predicts 3D hand-point motion and …
@misc{chen2026dexforesight,
title = {{DexForesight: Grounding Dexterous Policies with Implicit Spatial Foresight}},
author = {Zhongxi Chen and Yi Yang and Xinyi Chen and Mingzhe Ye and Yaping Li and Zhihong Liu and Congsheng Xu and Haonan Wang and others},
year = {2026},
eprint = {2611.11166},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11166},
}
Qinhong Zhou, Riley Zong, Chunru Lin, Diyi Hu et al.
Abstract摘要
Recent robot manipulation benchmarks have enabled systematic evaluation of general-purpose robot policies across diverse objects, scenes, instructions, and embodiments. However, most existing benchmarks still assume that the robot is the only active agent and that task semantics are determined primarily by the instruction and object state. This assumption overlooks a common source of real-world variation: humans actively sharing the workspace. In such settings, humans may reveal an implicit goal, complete part of the task, temporarily change object availability, or constrain safe robot motion …
@misc{zhou2026harp,
title = {{HARP-Bench: Benchmarking Robot Manipulation Around Active Humans}},
author = {Qinhong Zhou and Riley Zong and Chunru Lin and Diyi Hu and Shivraj Singh Bhatti and Hongxin Zhang and Lin Qiu and Hanqing Zeng and others},
year = {2026},
eprint = {2611.11123},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.11123},
}
Lei Shu, Alberta Longhini, Leonidas Guibas, Gordon Wetzstein et al.
Abstract摘要
We present SoftPAT3D, a physics-augmented framework for text-to-3D scene generation in which deformable objects, such as cloth, stuffed toys and ropes, are generated as physically valid, simulation-ready bodies and placed in intersection-free contact with the furniture that supports them, with one another and with themselves. Existing physics-augmented pipelines simulate rigid bodies only and cannot accommodate a generated deformable object, which text-to-3D generation returns as a geometrically invalid mesh, sometimes self-intersecting, that has no rest shape, no material model and no …
@misc{shu2026softpat,
title = {{SoftPAT3D: Physics-Augmented Text-to-3D Scene Generation with Deformables}},
author = {Lei Shu and Alberta Longhini and Leonidas Guibas and Gordon Wetzstein and Jeannette Bohg},
year = {2026},
eprint = {2611.11087},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2611.11087},
}
Mayur Deshmukh, Aaron Tung, Junyi Li, Nikash Bhardwaj et al.
Abstract摘要
Humanoid robots must coordinate locomotion and manipulation to navigate complex environments and perform everyday tasks. Learning such behaviours from human demonstrations requires accurate whole-body motion paired with egocentric vision, particularly for lower-body movements and foot placements that are frequently invisible from a first-person perspective. However, existing egocentric datasets often lack high-fidelity optical motion capture or systematic coverage of diverse loco-manipulation behaviours. To address this gap, we introduce RoboMoCap, a large-scale multimodal dataset combining …
@misc{deshmukh2026robomocap,
title = {{RoboMoCap: A Large-Scale Whole-Body Motion Capture Dataset with Egocentric Vision for Robot Learning}},
author = {Mayur Deshmukh and Aaron Tung and Junyi Li and Nikash Bhardwaj and Ademi Adeniji and Vincent Liu and Michael Cho and Jiafei Duan},
year = {2026},
eprint = {2611.10994},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10994},
}
Dong Fang, Feixuan Huang, Yuanxin Zhong, Yuxiang Gao et al.
Abstract摘要
Dexterous manipulation with world action models (WAMs) relies on hand configuration and contact information, yet joint-state and whole-hand tactile observations may be partially or entirely missing. RGB images provide scene context and hand-object geometry but are insufficient to reliably recover these missing modalities. We further observe a correlation between hand state and tactile information. Motivated by this observation, we introduce BAT-WAM, a framework that bridges visual, joint-state, and tactile representations for WAMs. BAT-WAM learns separate codebooks and decoders for the …
@misc{fang2026bat,
title = {{BAT-WAM: Bridging Action and Touch in World-Action Models for Dexterous Hands}},
author = {Dong Fang and Feixuan Huang and Yuanxin Zhong and Yuxiang Gao and Qi Song and Yunlong Wang and Xiaosong Jia and Yu-Gang Jiang},
year = {2026},
eprint = {2611.10967},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10967},
}
Sungjun Heo, Sungjun Yang, Heedon Jeong, Junhyoung Ha et al.
Abstract摘要
Selective data acquisition in Dataset Aggregation (DAgger) on robots requires deciding which learner rollouts to correct and when to intervene. In manipulation with ordered phases, router entropy from a phase-supervised soft mixture-of-experts (MoE) is insufficient: the same value can reflect either a normal adjacent-phase handoff or support for nonadjacent or earlier phases. We propose MoE-DAgger, which uses causal router history to distinguish these cases without providing the policy with a phase label, progress variable, or hard phase command during execution. The acquisition rule allows …
@misc{heo2026moe,
title = {{MoE-DAgger: Transition-Aware Router-Entropy Gating for Selective Human Correction in Phase-Structured Manipulation}},
author = {Sungjun Heo and Sungjun Yang and Heedon Jeong and Junhyoung Ha and Jeong hwan Jeon},
year = {2026},
eprint = {2611.10825},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10825},
}
Taeyun Kim, Jinsoo Shin, Taehwa Kim, Taeyoon Kim et al.
Abstract摘要
Policies for dexterous manipulation depend on design choices made before training: which sensors enter the observation, where they are placed, and how their readings are arranged for the encoder. These choices are rarely examined. For tactile sensing, simulation studies place taxels uniformly over the hand, and this placement has become a default without being tested. We test this default under a single controlled protocol. The protocol holds the teacher, the training procedure, and the candidate taxels fixed and varies only the sensor configuration and the representation. We evaluate it on …
@misc{kim2026which,
title = {{Which Taxels, Not How Many: Tactile Sensor Placement for Dexterous Policies}},
author = {Taeyun Kim and Jinsoo Shin and Taehwa Kim and Taeyoon Kim and Jinyoung Choi},
year = {2026},
eprint = {2611.10647},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10647},
}
Reliable navigation with a single monocular limited field-of-view camera is challenging because the lack of depth measurements and limited spatial awareness complicate collision checking. Existing approaches to tackle these challenges fall into two categories. Model-based planners can incorporate explicit collision-avoidance guidance which leads to better generalization but are often computationally expensive and require manual tuning of the algorithmic parameters. In contrast, end-to-end approaches trained by imitation learning can be flexible but typically encode safety only implicitly and …
Self-improvement on compositional manipulation tasks requires deciding what to reuse, where to improve, and which changes to retain. We present ReSkill, a selective self-improvement framework that concentrates revision on capability gaps. A coding agent constructs an object-centric task graph from an instruction and a skill library, then uses node-level execution records to locate failures. At the task level, it revises stage organization, hand-off conditions, and recovery routes; at the skill level, it strengthens perception, motion, manipulation, and verification capabilities that …
@misc{sun2026reskill,
title = {{ReSkill: Selective Self-Improvement for Robot Manipulation}},
author = {Zheng Sun and Zhihao Li and Zhuo Li and Wenrui Liu and Xin Wang and Fei Chen},
year = {2026},
eprint = {2611.10510},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10510},
}
Learned manipulation policies often rely on wrist-mounted cameras for close-range visual feedback that complements external views of the workspace. However, contact-rich manipulation requires establishing and maintaining contacts that can remain hidden even from wrist cameras. Tactile sensors mounted on fingers can reliably and directly measure contact, without needing visual access. In this work, we show that this feedback enables policies to recover the performance advantage of wrist cameras using scene and egocentric vision together with touch. Our approach incorporates two low-cost …
@misc{yang2026tactile,
title = {{Tactile Sensors Are an Effective Substitute for Wrist Cameras in Contact-Rich or Occluded Manipulation}},
author = {Aijia Yang and Arsh Tangri and Sergey Zakharov and Krishna Murthy Jatavallabhula},
year = {2026},
eprint = {2611.10490},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10490},
}
Rongqian Chen, Yiding Chen, Jijie Zhou, Tom Z. Jiahao
Abstract摘要
Most robotic handwriting rigidly mounts the pen and reduces writing to trajectory tracking. A freely held pen instead couples grasp stability, tip motion, and contact force. We present Grip, Trace, Write, a force-aware simulation environment and learning framework for high-DoF handwriting. A two-branch actor separates pen retention from arm-wrist writing while preserving their closed-loop coupling. Training learns a grasp controller, initializes writing from inverse-kinematics demonstrations by behavior cloning, and improves writing through reinforcement learning with rewards on ink, force, …
@misc{chen2026grip,
title = {{Grip, Trace, Write: Force-Aware Imitation and Reinforcement Learning for High-DoF Robotic Handwriting}},
author = {Rongqian Chen and Yiding Chen and Jijie Zhou and Tom Z. Jiahao},
year = {2026},
eprint = {2611.10451},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10451},
}
Qiyang Yan, Yulin Liu, Akash Sharma, Haozhi Qi et al.
Abstract摘要
Bimanual dexterous manipulation requires coordinated arm and hand control, yet specifying such coordination manually is challenging and often task-specific. In this work, we explore whether complex bimanual dexterity can emerge from a simple object-centric learning objective. We train reinforcement learning policies under a shared object-centric reward, varying their training environments rather than tuning separate rewards for each behavior. We find that the policies discover pre-grasp manipulation, grasping and regrasping, in-hand adjustment and handover to reach object-pose goals without …
@misc{yan2026simduoreal,
title = {{SimDuoReal: Object-Centric Bimanual Policy for Emergent Dexterity}},
author = {Qiyang Yan and Yulin Liu and Akash Sharma and Haozhi Qi and Guanya Shi and C. Karen Liu},
year = {2026},
eprint = {2611.10410},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10410},
}
Visuomotor policies can handle many robot manipulation tasks, but understanding their failures remains difficult. Unfamiliar conditions and small action errors can lead to failure, yet watching the robot alone tells us little about what went wrong inside the policy. We introduce MIND (Mechanistic Interpretability for Neural Diagnosis), a method that looks inside a trained policy to detect failures during execution and identify the internal concepts involved. MIND uses a sparse autoencoder to break down the policy's internal activations into features that can be studied individually. We give …
@misc{ahmed2026detecting,
title = {{Detecting and Explaining Robot Policy Failures from Sparse Internal Features}},
author = {Riad Ahmed and Sujosh Nag and Moniruzzaman Akash and Momotaz Begum},
year = {2026},
eprint = {2611.10400},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10400},
}
Aryaman Gupta, Xiaoyue Wu, Rana Soltani Zarrin, Mark Van der Merwe
Abstract摘要
Generative robot policies based on diffusion and flow matching can model complex, multimodal behaviors, but remain susceptible to covariate shift during closed-loop execution. Existing approaches mitigate such failures through additional demonstrations, policy fine-tuning, or test-time guidance, but often require human intervention, engineered rewards, or other task-specific signals. We introduce GUARD, an anomaly-informed Q-steering framework that leverages the policy's training distribution to construct a dense, failure-anticipatory steering signal. GUARD pairs sparse task rewards with …
@misc{gupta2026guard,
title = {{GUARD: Anomaly-Informed Q-Steering for Generative Robot Policies in Dexterous Manipulation}},
author = {Aryaman Gupta and Xiaoyue Wu and Rana Soltani Zarrin and Mark Van der Merwe},
year = {2026},
eprint = {2611.10343},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.10343},
}
Chen Chen, Omid Bagherieh, Naveen Nikamanth Appiah Balaji, Jesse T. Gonzalez et al.
Abstract摘要
Recent work on robot manipulation has focused on specialist policies trained on demonstrations. We show that generalist AI models can control manipulators more reliably than specialist policies such as vision-language-action models (VLAs) and world-action models (WAMs). We systematically analyze agentic manipulation policies (AMPs), in which a generalist AI model commands the manipulator directly, adapting its actions as the task unfolds. We present AMP1, a reference harness that runs unchanged across single-arm, bimanual, and mobile manipulation. AMP1 achieves state-of-the-art results on the …
@misc{chen2026agentic,
title = {{On Agentic Manipulation Policies}},
author = {Chen Chen and Omid Bagherieh and Naveen Nikamanth Appiah Balaji and Jesse T. Gonzalez and Negin Heravi and Fahad Islam and Xueting Li and Jean-Baptiste Regli and others},
year = {2026},
eprint = {2611.09892},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09892},
}
Loco-manipulation, which requires mobile manipulators to coordinate locomotion and manipulation over long horizons, is a key capability for large-object rearrangement in applications such as service robotics and warehouse automation. Despite recent progress, reliably executing these tasks remains challenging because performance depends on complex robot-object and object-environment interaction dynamics. Existing approaches address this challenge either through model-based methods, which often require accurate dynamics models, or through learning-based methods, which reduce this requirement …
@misc{tu2026grasp,
title = {{Grasp-TAGS: Task-Aware Grasp Selection via Critic Distillation for Prehensile Loco-Manipulation}},
author = {Xun Tu and Minghao Zou and Karthik Desingh},
year = {2026},
eprint = {2611.09864},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09864},
}
When should a robot look at the cup, and when at the portafilter? Spatial-reasoning benchmarks evaluate whether a model can describe a scene, but manipulation needs something more specific: the ability to acquire the particular evidence that makes the current command executable. A view that is informative for one task can be useless for another, even at the same workstation. We introduce Seeing What Matters, a benchmark for task-conditioned actionable spatial belief in simulated tabletop manipulation. Across three domains--coffee workstation, drawer/cabinet, and microwave/appliance--the …
Adarsh Salagame, Henry Noyes, Haroon Hublikar, Alireza Ramezani
Abstract摘要
Snake robots can leverage their hyper-redundant morphology for both locomotion and manipulation through distributed body contact but whole-body loco-manipulation is particularly challenging since the entire body serves simultaneously as propulsive mechanism and manipulation tool Prior work on wheeled legged and snake robot platforms has decoupled these functions through dedicated appendages body segmentation or grippers We present closed-loop loco-manipulation with our snake robot COBRA without this distinction such that any part of the body may make contact We develop a contact-implicit …
@misc{salagame2026autonomous,
title = {{Autonomous Whole-body Loco-manipulation Using a Snake Robot}},
author = {Adarsh Salagame and Henry Noyes and Haroon Hublikar and Alireza Ramezani},
year = {2026},
eprint = {2611.09777},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09777},
}
Tianshu Wu, Tianxiang Gui, Yue Chen, Tingxu Liu et al.
Abstract摘要
Scaling up visual humanoid loco-manipulation skill acquisition requires reducing the human effort needed as tasks and spatial configurations expand, particularly in demonstration collection and task-specific training design. We present SUGARv2, a unified framework for learning diverse visual humanoid loco-manipulation skills through interaction reference synthesis and a shared training pipeline. To obtain diverse references at low cost, we adapt reusable task templates to different spatial configurations, synthesizing coordinated robot and object motions. To reduce task-specific reward design …
@misc{wu2026sugarv,
title = {{SUGARv2: Learning Diverse Visual Humanoid Loco-Manipulation Skills from Synthesized Interaction References}},
author = {Tianshu Wu and Tianxiang Gui and Yue Chen and Tingxu Liu and Ziye Peng and Guangjie Yang and Xiangqi Kong and Ye Shi and others},
year = {2026},
eprint = {2611.09266},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.09266},
}
Grasp-feasibility predictors report near-perfect accuracy on Task and Motion Planning benchmarks, but high accuracy on these benchmarks does not verify that a predictor uses the collision geometry that the motion planner checks. We propose a planner-verified counterfactual (CF) test: holding the grasp pose fixed, we insert a solid obstacle into the approach corridor, add its points to the model input, and let a planner re-derive the label, which flips from feasible to infeasible. Across eight predictors spanning point clouds, voxels, depth, and object tokens, CF-pair AUROC falls from high …
We present RevoSim, a scalable and high-fidelity multimodal tactile simulator for the BrainCo Revo3 hand that jointly models 285 piezoresistive taxels and five visuotactile fingertips. RevoSim efficiently generates synchronized pressure, depth, RGB, and marker-displacement observations; on a single RTX 5090, its depth-conditioned marker-displacement configuration supports up to 4,096 parallel environments. We train in-hand manipulation and grasping policies entirely in simulation, establishing a foundation for future sim-to-real transfer. Project page: …
@misc{zhang2026revosim,
title = {{RevoSim: Scalable Multimodal Tactile Simulation for Dexterous Manipulation}},
author = {Ke Zhang and Lixin Xu and Ziyi Wang and Xiyue Dong and Jie Tan and Lianghu Shao and Xianpeng Wang and Chuanyu Li and others},
year = {2026},
eprint = {2611.08939},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08939},
}
ZhenYu Zou, Xing Liu, ShiChuang Wang, YunLong Dong et al.
Abstract摘要
Humanoid robots are expected to take over strenuous material handling in manufacturing, logistics, and healthcare, where operating existing wheeled carts would let them move heavy loads without modifying the workplace. However, humanoid cart pushing remains challenging because the bimanual grasp closes a force-coupled loop in which the contact wrenches that drive the cart also disturb the robot's balance, and payload changes alter the coupled dynamics. We propose ForPC, a force-aware whole-body control framework that regulates cart velocity through two-stage reinforcement learning. A …
@misc{zou2026forpc,
title = {{ForPC: Force-Aware Whole-Body Control for Precise Humanoid Cart Pushing}},
author = {ZhenYu Zou and Xing Liu and ShiChuang Wang and YunLong Dong and LiNan Deng and Feng Hua and Jun Wan and Cheng Cheng},
year = {2026},
eprint = {2611.08664},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08664},
}
Meituan Robotics Team, Caicheng Wang, Chen Zhao, Chengpeng Jin et al.
Abstract摘要
We present Meituan-Robotics-0 (MR0), a vision-language-action foundation model and embodied-agent system for stationary-base desktop manipulation. MR0 combines 147 million vision-language samples with 15,060 hours of curated robot trajectories spanning 12 embodiments. A three-stage training recipe aligns multimodal representations with robot behavior through discrete action-token pre-training, transfers these priors to continuous cross-embodiment control through flow-matching mid-training, and adapts the policy through real-world post-training. We introduce Rollback, which pairs replaced …
@misc{team2026meituan,
title = {{Meituan-Robotics-0: A Vision-Language-Action Foundation Model for Desktop Manipulation}},
author = {Meituan Robotics Team and Caicheng Wang and Chen Zhao and Chengpeng Jin and Duo Xu and Fengjiao Chen and Guanglu Wan and Heqing Yang and others},
year = {2026},
eprint = {2611.08577},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2611.08577},
}