Part-Guided 3D RL for Sim2Real Articulated Object Manipulation

TL;DR

This paper introduces a part-guided 3D reinforcement learning framework that enables robots to manipulate unseen articulated objects using visual feedback, combining 2D segmentation and hierarchical 3D representations for improved efficiency and generalization.

Contribution

It presents a novel part-guided 3D RL approach that learns manipulation policies without demonstrations and generalizes across multiple object categories and instances.

Findings

Effective in simulation and real-world tests

Improves training efficiency with 2D segmentation and 3D representation

Demonstrates strong generalization to novel objects

Abstract

Manipulating unseen articulated objects through visual feedback is a critical but challenging task for real robots. Existing learning-based solutions mainly focus on visual affordance learning or other pre-trained visual models to guide manipulation policies, which face challenges for novel instances in real-world scenarios. In this paper, we propose a novel part-guided 3D RL framework, which can learn to manipulate articulated objects without demonstrations. We combine the strengths of 2D segmentation and 3D RL to improve the efficiency of RL policy training. To improve the stability of the policy on real robots, we design a Frame-consistent Uncertainty-aware Sampling (FUS) strategy to get a condensed and hierarchical 3D representation. In addition, a single versatile RL policy can be trained on multiple articulated object manipulation tasks simultaneously in simulation and shows great generalizability to novel categories and instances. Experimental results demonstrate the effectiveness of our framework in both simulation and real-world settings. Our code is available at https://github.com/THU-VCLab/Part-Guided-3D-RL-for-Sim2Real-Articulated-Object-Manipulation.