UniDoorManip: Learning Universal Door Manipulation Policy Over Large-scale and Diverse Door Manipulation Environments

TL;DR

This paper introduces UniDoorManip, a comprehensive framework and environment for training a universal door manipulation policy that generalizes across diverse door types, geometries, and mechanisms using large-scale datasets and realistic simulation settings.

Contribution

The work presents a novel environment, a large-scale diverse door dataset, and a three-stage disentangled learning framework for universal door manipulation.

Findings

The framework achieves strong performance across various door types.

The environment and dataset enable better generalization to real-world scenarios.

Extensive experiments validate the effectiveness of the proposed approach.

Abstract

Learning a universal manipulation policy encompassing doors with diverse categories, geometries and mechanisms, is crucial for future embodied agents to effectively work in complex and broad real-world scenarios. Due to the limited datasets and unrealistic simulation environments, previous works fail to achieve good performance across various doors. In this work, we build a novel door manipulation environment reflecting different realistic door manipulation mechanisms, and further equip this environment with a large-scale door dataset covering 6 door categories with hundreds of door bodies and handles, making up thousands of different door instances. Additionally, to better emulate real-world scenarios, we introduce a mobile robot as the agent and use the partial and occluded point cloud as the observation, which are not considered in previous works while possessing significance for real-world implementations. To learn a universal policy over diverse doors, we propose a novel framework disentangling the whole manipulation process into three stages, and integrating them by training in the reversed order of inference. Extensive experiments validate the effectiveness of our designs and demonstrate our framework's strong performance. Code, data and videos are avaible on https://unidoormanip.github.io/.