Last-Iterate Global Convergence of Policy Gradients for Constrained Reinforcement Learning

TL;DR

This paper introduces a new gradient-based primal-dual framework for constrained reinforcement learning, providing global convergence guarantees and extending to risk-based constraints, with empirical validation on control problems.

Contribution

It proposes a novel exploration-agnostic algorithm C-PG with last-iterate convergence guarantees and extends it to risk-sensitive constraints in CRL.

Findings

C-PG achieves global last-iterate convergence under weak assumptions.

The algorithms effectively handle risk-based constraints.

Numerical results outperform state-of-the-art baselines.

Abstract

Constrained Reinforcement Learning (CRL) tackles sequential decision-making problems where agents are required to achieve goals by maximizing the expected return while meeting domain-specific constraints, which are often formulated as expected costs. In this setting, policy-based methods are widely used since they come with several advantages when dealing with continuous-control problems. These methods search in the policy space with an action-based or parameter-based exploration strategy, depending on whether they learn directly the parameters of a stochastic policy or those of a stochastic hyperpolicy. In this paper, we propose a general framework for addressing CRL problems via gradient-based primal-dual algorithms, relying on an alternate ascent/descent scheme with dual-variable regularization. We introduce an exploration-agnostic algorithm, called C-PG, which exhibits global last-iterate convergence guarantees under (weak) gradient domination assumptions, improving and generalizing existing results. Then, we design C-PGAE and C-PGPE, the action-based and the parameter-based versions of C-PG, respectively, and we illustrate how they naturally extend to constraints defined in terms of risk measures over the costs, as it is often requested in safety-critical scenarios. Finally, we numerically validate our algorithms on constrained control problems, and compare them with state-of-the-art baselines, demonstrating their effectiveness.