Latent Space Policies for Hierarchical Reinforcement Learning

TL;DR

This paper introduces a hierarchical reinforcement learning framework where each layer learns diverse strategies using maximum entropy objectives and latent variables, enabling complex task solving without restricting lower layers.

Contribution

It proposes a novel hierarchical policy method with latent variables and invertible mappings, allowing higher layers to control lower layers directly and improving performance on complex tasks.

Findings

Hierarchical policies outperform single-layer policies on benchmarks.

Adding layers enables solving more complex sparse-reward tasks.

Latent space control enhances policy expressivity and flexibility.

Abstract

We address the problem of learning hierarchical deep neural network policies for reinforcement learning. In contrast to methods that explicitly restrict or cripple lower layers of a hierarchy to force them to use higher-level modulating signals, each layer in our framework is trained to directly solve the task, but acquires a range of diverse strategies via a maximum entropy reinforcement learning objective. Each layer is also augmented with latent random variables, which are sampled from a prior distribution during the training of that layer. The maximum entropy objective causes these latent variables to be incorporated into the layer's policy, and the higher level layer can directly control the behavior of the lower layer through this latent space. Furthermore, by constraining the mapping from latent variables to actions to be invertible, higher layers retain full expressivity: neither the higher layers nor the lower layers are constrained in their behavior. Our experimental evaluation demonstrates that we can improve on the performance of single-layer policies on standard benchmark tasks simply by adding additional layers, and that our method can solve more complex sparse-reward tasks by learning higher-level policies on top of high-entropy skills optimized for simple low-level objectives.