ViViDex: Learning Vision-based Dexterous Manipulation from Human Videos
This addresses the challenge of enabling multi-fingered robot hands to manipulate diverse objects in realistic scenarios, representing a strong specific gain in robotics.
The authors tackled the problem of learning vision-based dexterous manipulation policies from noisy human videos without privileged object information, and their method ViViDex outperformed state-of-the-art approaches on three tasks in simulation and on a real robot.
In this work, we aim to learn a unified vision-based policy for multi-fingered robot hands to manipulate a variety of objects in diverse poses. Though prior work has shown benefits of using human videos for policy learning, performance gains have been limited by the noise in estimated trajectories. Moreover, reliance on privileged object information such as ground-truth object states further limits the applicability in realistic scenarios. To address these limitations, we propose a new framework ViViDex to improve vision-based policy learning from human videos. It first uses reinforcement learning with trajectory guided rewards to train state-based policies for each video, obtaining both visually natural and physically plausible trajectories from the video. We then rollout successful episodes from state-based policies and train a unified visual policy without using any privileged information. We propose coordinate transformation to further enhance the visual point cloud representation, and compare behavior cloning and diffusion policy for the visual policy training. Experiments both in simulation and on the real robot demonstrate that ViViDex outperforms state-of-the-art approaches on three dexterous manipulation tasks.