Sampling-based Exploration for Reinforcement Learning of Dexterous Manipulation

TL;DR

This paper introduces a sampling-based exploration method using RRT algorithms to improve reinforcement learning for complex dexterous manipulation tasks, enabling policies that transfer well to real robots.

Contribution

It presents a novel exploration approach combining RRT variants with reinforcement learning to handle high-dimensional manipulation problems without passive supports.

Findings

Effective manipulation policies for complex objects

Successful transfer to real robotic systems

Enhanced sample efficiency in exploration

Abstract

In this paper, we present a novel method for achieving dexterous manipulation of complex objects, while simultaneously securing the object without the use of passive support surfaces. We posit that a key difficulty for training such policies in a Reinforcement Learning framework is the difficulty of exploring the problem state space, as the accessible regions of this space form a complex structure along manifolds of a high-dimensional space. To address this challenge, we use two versions of the non-holonomic Rapidly-Exploring Random Trees algorithm; one version is more general, but requires explicit use of the environment's transition function, while the second version uses manipulation-specific kinematic constraints to attain better sample efficiency. In both cases, we use states found via sampling-based exploration to generate reset distributions that enable training control policies under full dynamic constraints via model-free Reinforcement Learning. We show that these policies are effective at manipulation problems of higher difficulty than previously shown, and also transfer effectively to real robots. Videos of the real-hand demonstrations can be found on the project website: https://sbrl.cs.columbia.edu/