DrS: Learning Reusable Dense Rewards for Multi-Stage Tasks

TL;DR

DrS introduces a data-driven method to learn reusable dense rewards for multi-stage tasks, reducing human effort and improving RL performance across diverse unseen tasks.

Contribution

We propose DrS, a novel approach that learns dense rewards from sparse rewards and demonstrations, enabling reward reuse in unseen multi-stage tasks.

Findings

Reused learned rewards improve RL performance and sample efficiency.

Rewards achieve comparable results to human-engineered rewards.

Method generalizes well across 1000+ task variants.

Abstract

The success of many RL techniques heavily relies on human-engineered dense rewards, which typically demand substantial domain expertise and extensive trial and error. In our work, we propose DrS (Dense reward learning from Stages), a novel approach for learning reusable dense rewards for multi-stage tasks in a data-driven manner. By leveraging the stage structures of the task, DrS learns a high-quality dense reward from sparse rewards and demonstrations if given. The learned rewards can be \textit{reused} in unseen tasks, thus reducing the human effort for reward engineering. Extensive experiments on three physical robot manipulation task families with 1000+ task variants demonstrate that our learned rewards can be reused in unseen tasks, resulting in improved performance and sample efficiency of RL algorithms. The learned rewards even achieve comparable performance to human-engineered rewards on some tasks. See our project page (https://sites.google.com/view/iclr24drs) for more details.