EgoEngine: From Egocentric Human Videos to High-Fidelity Dexterous Robot Demonstrations
Converts egocentric human videos into robot data: a video with the human replaced by a robot plus a feasible robot action trajectory. Claims the first zero-shot dexterous visuomotor policy with no real-robot demos.把第一人称人类视频转为机器人数据：生成把人替换成机器人的视频，以及满足可行性约束的机器人动作轨迹。声称首次在无真机演示下零样本训练出灵巧视觉运动策略。
BibTeX
@misc{liu2026egoengine,
title = {{EgoEngine: From Egocentric Human Videos to High-Fidelity Dexterous Robot Demonstrations}},
author = {Yangcen Liu and Shuo Cheng and Xinchen Yin and Woo Chul Shin and Alfred Cueva and Yiran Yang and Zhenyang Chen and Chuye Zhang and Danfei Xu},
year = {2026},
eprint = {2606.12604},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.12604},
}Robot Learning: RL & Imitation机器人学习：强化学习与模仿学习Dexterous hand灵巧手human-videoegocentricdata-generationdexterousreal