EgoMimic: Scaling Imitation Learning via Egocentric Video

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。

Paper论文 · Code代码 · Project项目页 ★ 230

BibTeX

@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},
}

Manipulation & Dexterous Hands操作与灵巧手 Robot Learning: RL & Imitation机器人学习：强化学习与模仿学习Robot arm机械臂realbimanualhuman-videoegocentricimitation

Authors (from arXiv): 作者（来自 arXiv）：Simar Kareer, Dhruv Patel, Ryan Punamiya, Pranay Mathur, Shuo Cheng, Chen Wang, Judy Hoffman, Danfei Xu

More from: 同机构更多条目：Georgia Tech佐治亚理工学院

EgoMimic: Scaling Imitation Learning via Egocentric Video

More in Manipulation & Dexterous Hands更多操作与灵巧手条目