DexRepNet++: Learning Dexterous Robotic Manipulation with Geometric and Spatial Hand-Object Representations

TL;DR

This paper introduces DexRep, a novel hand-object representation for dexterous robotic manipulation, improving generalization and success rates across various tasks and object categories in simulation and real-world settings.

Contribution

DexRep is a new representation capturing surface features and spatial relations, enabling more effective policy learning for complex manipulation tasks.

Findings

Achieves 87.9% success rate on unseen objects in simulation.

Boosts success rates by 20-40% over existing representations.

Demonstrates effective sim-to-real transfer with minimal gap.

Abstract

Robotic dexterous manipulation is a challenging problem due to high degrees of freedom (DoFs) and complex contacts of multi-fingered robotic hands. Many existing deep reinforcement learning (DRL) based methods aim at improving sample efficiency in high-dimensional output action spaces. However, existing works often overlook the role of representations in achieving generalization of a manipulation policy in the complex input space during the hand-object interaction. In this paper, we propose DexRep, a novel hand-object interaction representation to capture object surface features and spatial relations between hands and objects for dexterous manipulation skill learning. Based on DexRep, policies are learned for three dexterous manipulation tasks, i.e. grasping, in-hand reorientation, bimanual handover, and extensive experiments are conducted to verify the effectiveness. In simulation, for grasping, the policy learned with 40 objects achieves a success rate of 87.9% on more than 5000 unseen objects of diverse categories, significantly surpassing existing work trained with thousands of objects; for the in-hand reorientation and handover tasks, the policies also boost the success rates and other metrics of existing hand-object representations by 20% to 40%. The grasp policies with DexRep are deployed to the real world under multi-camera and single-camera setups and demonstrate a small sim-to-real gap.