Reinforcement Learning for Improving Agent Design

TL;DR

This paper introduces a method for jointly learning an agent's physical design and policy within a modified reinforcement learning framework, leading to improved task performance and potential design insights.

Contribution

It proposes a novel approach to optimize both agent structure and policy simultaneously, extending the OpenAI Gym framework for environment parameter learning.

Findings

Agents learn more effective body structures for tasks.

Joint optimization accelerates policy learning.

Design principles can be uncovered through joint learning.

Abstract

In many reinforcement learning tasks, the goal is to learn a policy to manipulate an agent, whose design is fixed, to maximize some notion of cumulative reward. The design of the agent's physical structure is rarely optimized for the task at hand. In this work, we explore the possibility of learning a version of the agent's design that is better suited for its task, jointly with the policy. We propose an alteration to the popular OpenAI Gym framework, where we parameterize parts of an environment, and allow an agent to jointly learn to modify these environment parameters along with its policy. We demonstrate that an agent can learn a better structure of its body that is not only better suited for the task, but also facilitates policy learning. Joint learning of policy and structure may even uncover design principles that are useful for assisted-design applications. Videos of results at https://designrl.github.io/