Improving Stochastic Action-Constrained Reinforcement Learning via Truncated Distributions

TL;DR

This paper introduces efficient numerical methods for accurately estimating key properties of truncated normal distributions in action-constrained reinforcement learning, leading to improved policy performance.

Contribution

It proposes novel numerical approximations and sampling strategies for truncated distributions, enhancing policy updates in constrained RL settings.

Findings

Significant performance improvements on benchmark environments

Accurate estimation of entropy and log-probability is crucial

Efficient sampling reduces computational overhead

Abstract

In reinforcement learning (RL), it is often advantageous to consider additional constraints on the action space to ensure safety or action relevance. Existing work on such action-constrained RL faces challenges regarding effective policy updates, computational efficiency, and predictable runtime. Recent work proposes to use truncated normal distributions for stochastic policy gradient methods. However, the computation of key characteristics, such as the entropy, log-probability, and their gradients, becomes intractable under complex constraints. Hence, prior work approximates these using the non-truncated distributions, which severely degrades performance. We argue that accurate estimation of these characteristics is crucial in the action-constrained RL setting, and propose efficient numerical approximations for them. We also provide an efficient sampling strategy for truncated policy distributions and validate our approach on three benchmark environments, which demonstrate significant performance improvements when using accurate estimations.