Unsupervised Skill-Discovery and Skill-Learning in Minecraft
This is an incremental improvement for researchers in reinforcement learning and unsupervised learning, addressing scalability issues in skill discovery.
The paper tackled the problem of learning and discovering meaningful skills in high-dimensional pixel spaces for reinforcement learning agents in Minecraft, but found that their method using unsupervised skill discovery and self-supervised learning only worked for toy examples and did not scale to complex maps.
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.