Action Mapping for Reinforcement Learning in Continuous Environments with Constraints

TL;DR

This paper introduces an action mapping strategy for deep reinforcement learning in continuous constrained environments, improving training efficiency by leveraging feasibility models to focus on feasible actions.

Contribution

The paper proposes a novel DRL training method using action mapping that decouples feasibility assessment from policy learning, enhancing performance in constrained continuous spaces.

Findings

Significant improvement in training efficiency with action mapping.

Effective even with imperfect feasibility models.

Enhanced performance in constrained environments.

Abstract

Deep reinforcement learning (DRL) has had success across various domains, but applying it to environments with constraints remains challenging due to poor sample efficiency and slow convergence. Recent literature explored incorporating model knowledge to mitigate these problems, particularly through the use of models that assess the feasibility of proposed actions. However, integrating feasibility models efficiently into DRL pipelines in environments with continuous action spaces is non-trivial. We propose a novel DRL training strategy utilizing action mapping that leverages feasibility models to streamline the learning process. By decoupling the learning of feasible actions from policy optimization, action mapping allows DRL agents to focus on selecting the optimal action from a reduced feasible action set. We demonstrate through experiments that action mapping significantly improves training performance in constrained environments with continuous action spaces, especially with imperfect feasibility models.