从监督式玩耍中获取灵巧性
Dexterity from Supervised Play
AI总结：
该研究提出DSP框架，通过人类监督玩耍学习策略，自主收集数据训练世界模型，结合MPC与自玩蒸馏策略，在真实世界欠驱动推压等任务中，性能优于事后行为克隆且数据量远少于强化学习。
摘要：
机器人如何自主收集学习灵巧操作所需的大量接触式交互经验？我们提出DSP：Dexterity from supervised play（从监督式玩耍中获取灵巧性），该框架中机器人通过行为克隆从数分钟的人类玩耍中学习玩耍策略，随后自主部署该策略以收集数小时的交互数据来训练世界模型。推理阶段，同一玩耍策略通过为学习到的世界模型下的评估提出候选动作序列，进而指导模型预测控制（MPC）。为加速推理，我们还引入了一种在学习到的世界模型内通过自玩来提炼闭环、目标条件策略的方法。我们在欠驱动推压和多指手的任意到任意手内姿态重定向任务中，于仿真和真实世界中评估该框架。真实世界实验显示，我们的方法在成功率上比hindsight behavior cloning（事后行为克隆）高出69.6%，且在交互数据量减少490倍的情况下，达到与强化学习相当的成功率。这些结果表明，监督式玩耍可自主收集交互数据、指导基于模型的规划，并实现无需任何任务特定演示或仿真到真实迁移的策略学习。
英文摘要：
How can robots autonomously collect the contact-rich experience needed to learn dexterous manipulation? We present DSP: Dexterity from supervised play, a framework in which a robot learns a play policy from minutes of human play through behavior cloning, then deploys it autonomously to collect hours of interaction data to train a world model. During inference, the same play policy then guides model-predictive control (MPC) by proposing candidate action sequences for evaluation under the learned world model. To accelerate inference, we also introduce a method for distilling a closed-loop, goal-conditioned policy through self-play within the learned world model. We evaluate this framework on underactuated pushing and any-to-any in-hand pose reorientation with a multifingered hand, in simulation and real world. In real world experiment, our method outperforms hindsight behavior cloning by 69.6% in success rate and achieves comparable success to reinforcement learning with 490 times less interaction data. These results demonstrate that supervised play can autonomously collect interaction data, guide model-based planning, and enable policy learning without any task-specific demonstrations or sim-to-real transfer.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
