ViViDex: Learning Vision-based Dexterous Manipulation from Human Videos

TL;DR

This paper introduces ViViDex, a framework that learns vision-based dexterous manipulation policies from human videos, overcoming noise and privileged information limitations to achieve superior performance in simulation and real-world tasks.

Contribution

ViViDex combines reinforcement learning with trajectory-guided rewards and a coordinate transformation to train unified visual policies from human videos without privileged information.

Findings

Outperforms state-of-the-art methods in three manipulation tasks

Effective in both simulation and real robot experiments

Improves visual policy learning from noisy human videos

Abstract

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.