Unsupervised Skill-Discovery and Skill-Learning in Minecraft

TL;DR

This paper explores unsupervised skill discovery in high-dimensional pixel-based environments like Minecraft, using variational and contrastive learning to improve skill learning and representation, with mixed success in complex maps.

Contribution

It introduces a method combining variational and contrastive techniques for unsupervised skill discovery in pixel-based RL environments, addressing scalability issues.

Findings

Representations and policies work for simple environments.

Scaling to complex maps remains challenging.

Alternative inputs like relative position can improve performance.

Abstract

Pre-training Reinforcement Learning agents in a task-agnostic manner has shown promising results. However, previous works still struggle in learning and discovering meaningful skills in high-dimensional state-spaces, such as pixel-spaces. We approach the problem by leveraging unsupervised skill discovery and self-supervised learning of state representations. In our work, we learn a compact latent representation by making use of variational and contrastive techniques. We demonstrate that both enable RL agents to learn a set of basic navigation skills by maximizing an information theoretic objective. We assess our method in Minecraft 3D pixel maps with different complexities. Our results show that representations and conditioned policies learned from pixels are enough for toy examples, but do not scale to realistic and complex maps. To overcome these limitations, we explore alternative input observations such as the relative position of the agent along with the raw pixels.