Reward-Based Environment States for Robot Manipulation Policy Learning

TL;DR

This paper introduces a reward-based state representation for robot manipulation that improves policy learning efficiency, achieving up to 97% success in simulated tasks with deep reinforcement learning.

Contribution

The paper proposes a novel, compact state representation based on predicted rewards from an image classifier, enhancing robot manipulation policy learning.

Findings

Achieved up to 97% task success in simulation

Effective with deep reinforcement learning algorithms

Simplifies state representation for manipulation tasks

Abstract

Training robot manipulation policies is a challenging and open problem in robotics and artificial intelligence. In this paper we propose a novel and compact state representation based on the rewards predicted from an image-based task success classifier. Our experiments, using the Pepper robot in simulation with two deep reinforcement learning algorithms on a grab-and-lift task, reveal that our proposed state representation can achieve up to 97% task success using our best policies.