arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.04436cs.ROcs.AIcs.LGcs.SYeess.SY
收藏

Learning Human-to-Humanoid Real-Time Whole-Body Teleoperation(学习人-人形机器人实时全身遥操作)

Learning Human-to-Humanoid Real-Time Whole-Body Teleoperation

Carnegie Mellon University(卡内基梅隆大学)

Tairan He, Zhengyi Luo, Wenli Xiao, Chong Zhang, Kris Kitani, Changliu Liu, Guanya Shi

AI总结:

该研究提出基于强化学习的H2O框架,通过“sim-to-data”流程构建优化运动数据集,训练运动模仿器并零样本迁移至真实人形机器人,仅用RGB相机实现了多种动态全身动作的实时遥操作,为相关领域首个演示成果。

摘要:

我们提出Human to Humanoid(H2O,人到人形机器人)框架,这是一种基于强化学习(RL)的框架,仅用一台RGB相机即可实现全尺寸人形机器人的实时全身遥操作。为了构建面向人形机器人的大规模人类动作重定向运动数据集,我们提出了一种可扩展的“sim-to-data(仿真到数据)”流程,借助特权运动模仿器对可行运动进行筛选与挑选。随后,我们利用这些经过优化的运动数据在仿真环境中训练出鲁棒的实时人形机器人运动模仿器,并以零样本方式将其迁移到真实人形机器人上。我们在真实场景中成功实现了动态全身动作的遥操作,涵盖行走、后跳、踢腿、转身、挥手、推搡、拳击等动作。据我们所知,这是首个实现基于学习的实时全身人形机器人遥操作的演示成果。

英文摘要:

We present Human to Humanoid (H2O), a reinforcement learning (RL) based framework that enables real-time whole-body teleoperation of a full-sized humanoid robot with only an RGB camera. To create a large-scale retargeted motion dataset of human movements for humanoid robots, we propose a scalable "sim-to-data" process to filter and pick feasible motions using a privileged motion imitator. Afterwards, we train a robust real-time humanoid motion imitator in simulation using these refined motions and transfer it to the real humanoid robot in a zero-shot manner. We successfully achieve teleoperation of dynamic whole-body motions in real-world scenarios, including walking, back jumping, kicking, turning, waving, pushing, boxing, etc. To the best of our knowledge, this is the first demonstration to achieve learning-based real-time whole-body humanoid teleoperation.

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。

↑