HuggingFace Daily Papers·· 2 天前
HandsOn：面向可泛化灵巧操作的大规模仿真基准
HandsOn: A Large-Scale Simulation Benchmark for Generalizable Dexterous Manipulation
AI 导读
研究团队推出 HandsOn 仿真平台与基准，支持五种机械臂、六种灵巧手和约 200 个刚体、铰接及可变形资产，含 18,000 条人类演示与 66 个经人工核验的任务。表现最好的多任务策略在 Core-50 上成功率 29.1%；GPT-6 未经基准微调使用 move_eef 控制手掌与手指，在 660 次试验中 Core-50 仅 5.2%、Transfer-16 仅 1.9%。
正文 · 原文
Authors:Tianwei Luo, Yiming He, Hongjin Wu, Ting Pan, Debin Ren, Haorui Guo, Junlin Yang, Yuxin Zuo, Zhehao Kou, Kaisen Yang, Zhuoqun Lai, Yi Ding, Jiale Yu, Yuchen Zhang, Zijie Qiu, Haozhan Li, Ning Ding
Abstract:General-purpose manipulation requires reliable behavior across embodiments, objects, and task demands. We introduce HandsOn, a composable simulation platform and benchmark for cross-embodiment dexterous manipulation. It supports five robotic arms, six dexterous hands, and approximately 200 rigid, articulated, and deformable assets in randomized scenes. The benchmark pairs 18,000 curated human demonstrations with 66 human-verified tasks: Core-50 supports demonstration-based learning and execution tests, while Transfer-16 evaluates asset and relation transfer, skill composition, and deformable manipulation without target-task training demonstrations. Qualitative real-robot rollouts illustrate sim-to-real transfer. Across diffusion, vision-language-action, world-action, and prompted VLM-based methods, the best trained multi-task policy achieves 29.1% Core-50 success. Without benchmark finetuning, GPT-6 using move_eef for palm-and-finger control achieves 5.2% on Core-50 and 1.9% on Transfer-16 over 660 trials. Trained-policy stage diagnostics expose failures in contact establishment, action coordination, and maintaining achieved goals, with additional sequencing errors in composed tasks. HandsOn provides a shared basis for investigating manipulation generalization and the limits of reliable physical interaction.
| Comments: | 56 pages, including appendices |
| Subjects: | Robotics (cs.RO) |
| Cite as: | arXiv:2610.26765 [cs.RO] |
| (or arXiv:2610.26765v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.26765 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tianwei Luo [view email]
[v1]
Mon, 19 Oct 2026 17:22:48 UTC (32,802 KB)
来源：HuggingFace Daily Papers · arxiv.org