Visual Reinforcement Learning with Imagined Goals

TL;DR

This paper introduces a visual reinforcement learning method enabling agents to learn general-purpose, goal-conditioned skills from raw images through self-supervised practice, goal relabeling, and efficient off-policy algorithms, demonstrated on real robots.

Contribution

It combines unsupervised representation learning with goal-conditioned reinforcement learning, introducing a retroactive goal relabeling scheme for improved sample efficiency.

Findings

Outperforms prior techniques in real-world robotic experiments

Learns policies directly from raw image observations and goals

Efficient off-policy learning with visual inputs

Abstract

For an autonomous agent to fulfill a wide range of user-specified goals at test time, it must be able to learn broadly applicable and general-purpose skill repertoires. Furthermore, to provide the requisite level of generality, these skills must handle raw sensory input such as images. In this paper, we propose an algorithm that acquires such general-purpose skills by combining unsupervised representation learning and reinforcement learning of goal-conditioned policies. Since the particular goals that might be required at test-time are not known in advance, the agent performs a self-supervised "practice" phase where it imagines goals and attempts to achieve them. We learn a visual representation with three distinct purposes: sampling goals for self-supervised practice, providing a structured transformation of raw sensory inputs, and computing a reward signal for goal reaching. We also propose a retroactive goal relabeling scheme to further improve the sample-efficiency of our method. Our off-policy algorithm is efficient enough to learn policies that operate on raw image observations and goals for a real-world robotic system, and substantially outperforms prior techniques.