MimicPlay: Long-Horizon Imitation Learning by Watching Human Play

TL;DR

MimicPlay leverages human play videos to learn long-horizon robot manipulation tasks efficiently, reducing the need for extensive demonstrations by extracting latent plans to guide low-level control, achieving superior real-world performance.

Contribution

Introduces MimicPlay, a hierarchical framework that uses human play data to improve long-horizon imitation learning for robots with minimal demonstrations.

Findings

Outperforms state-of-the-art methods in success rate

Demonstrates strong generalization to new tasks

Shows robustness to environmental disturbances

Abstract

Imitation learning from human demonstrations is a promising paradigm for teaching robots manipulation skills in the real world. However, learning complex long-horizon tasks often requires an unattainable amount of demonstrations. To reduce the high data requirement, we resort to human play data - video sequences of people freely interacting with the environment using their hands. Even with different morphologies, we hypothesize that human play data contain rich and salient information about physical interactions that can readily facilitate robot policy learning. Motivated by this, we introduce a hierarchical learning framework named MimicPlay that learns latent plans from human play data to guide low-level visuomotor control trained on a small number of teleoperated demonstrations. With systematic evaluations of 14 long-horizon manipulation tasks in the real world, we show that MimicPlay outperforms state-of-the-art imitation learning methods in task success rate, generalization ability, and robustness to disturbances. Code and videos are available at https://mimic-play.github.io