Opportunities and Challenges from Using Animal Videos in Reinforcement Learning for Navigation

TL;DR

This paper explores how animal videos can be used to enhance reinforcement learning for navigation, addressing challenges and proposing solutions to improve efficiency and performance in sparse reward environments.

Contribution

It introduces a method to incorporate animal videos into RL, demonstrating improved navigation performance over traditional algorithms without such observations.

Findings

Animal videos can enhance RL navigation performance

Weighted policy optimization effectively leverages observational data

Proposed solutions address key challenges in learning from videos

Abstract

We investigate the use of animal videos (observations) to improve Reinforcement Learning (RL) efficiency and performance in navigation tasks with sparse rewards. Motivated by theoretical considerations, we make use of weighted policy optimization for off-policy RL and describe the main challenges when learning from animal videos. We propose solutions and test our ideas on a series of 2D navigation tasks. We show how our methods can leverage animal videos to improve performance over RL algorithms that do not leverage such observations.