Parrot: Data-Driven Behavioral Priors for Reinforcement Learning

TL;DR

This paper introduces Parrot, a pre-training approach for reinforcement learning that leverages behavioral priors learned from diverse tasks to enable rapid adaptation to new tasks, especially in robotic manipulation with image inputs.

Contribution

It proposes a novel pre-training method for RL that captures complex behaviors from previous tasks to facilitate quick learning of new tasks without limiting exploration.

Findings

Outperforms prior methods significantly in robotic manipulation tasks

Effective in environments with image observations and sparse rewards

Enables rapid adaptation to new tasks using behavioral priors

Abstract

Reinforcement learning provides a general framework for flexible decision making and control, but requires extensive data collection for each new task that an agent needs to learn. In other machine learning fields, such as natural language processing or computer vision, pre-training on large, previously collected datasets to bootstrap learning for new tasks has emerged as a powerful paradigm to reduce data requirements when learning a new task. In this paper, we ask the following question: how can we enable similarly useful pre-training for RL agents? We propose a method for pre-training behavioral priors that can capture complex input-output relationships observed in successful trials from a wide range of previously seen tasks, and we show how this learned prior can be used for rapidly learning new tasks without impeding the RL agent's ability to try out novel behaviors. We demonstrate the effectiveness of our approach in challenging robotic manipulation domains involving image observations and sparse reward functions, where our method outperforms prior works by a substantial margin.