Robust Manipulation Primitive Learning via Domain Contraction

TL;DR

This paper introduces a bi-level method combining domain contraction and parameter-augmented policy learning to improve robustness and generalization in contact-rich robotic manipulation tasks.

Contribution

It unifies domain randomization and adaptation for robust manipulation primitives with instance-specific adaptation.

Findings

Outperforms existing methods in hitting, pushing, reorientation tasks

Generates policies robust to diverse physical parameters

Maintains generalization while improving task success rates

Abstract

Contact-rich manipulation plays an important role in human daily activities, but uncertain parameters pose significant challenges for robots to achieve comparable performance through planning and control. To address this issue, domain adaptation and domain randomization have been proposed for robust policy learning. However, they either lose the generalization ability across diverse instances or perform conservatively due to neglecting instance-specific information. In this paper, we propose a bi-level approach to learn robust manipulation primitives, including parameter-augmented policy learning using multiple models, and parameter-conditioned policy retrieval through domain contraction. This approach unifies domain randomization and domain adaptation, providing optimal behaviors while keeping generalization ability. We validate the proposed method on three contact-rich manipulation primitives: hitting, pushing, and reorientation. The experimental results showcase the superior performance of our approach in generating robust policies for instances with diverse physical parameters.