Actor-Free Continuous Control via Structurally Maximizable Q-Functions

TL;DR

This paper introduces an actor-free, value-based reinforcement learning method for continuous control that achieves competitive performance and stability without learning a separate actor, especially excelling in constrained action environments.

Contribution

The work presents a novel actor-free Q-learning framework with structural maximization, improving stability and efficiency in continuous control tasks compared to traditional actor-critic methods.

Findings

Performs on par with state-of-the-art baselines in standard tasks.

Outperforms actor-critic methods in constrained action environments.

Demonstrates stable and sample-efficient learning without an actor.

Abstract

Value-based algorithms are a cornerstone of off-policy reinforcement learning due to their simplicity and training stability. However, their use has traditionally been restricted to discrete action spaces, as they rely on estimating Q-values for individual state-action pairs. In continuous action spaces, evaluating the Q-value over the entire action space becomes computationally infeasible. To address this, actor-critic methods are typically employed, where a critic is trained on off-policy data to estimate Q-values, and an actor is trained to maximize the critic's output. Despite their popularity, these methods often suffer from instability during training. In this work, we propose a purely value-based framework for continuous control that revisits structural maximization of Q-functions, introducing a set of key architectural and algorithmic choices to enable efficient and stable learning. We evaluate the proposed actor-free Q-learning approach on a range of standard simulation tasks, demonstrating performance and sample efficiency on par with state-of-the-art baselines, without the cost of learning a separate actor. Particularly, in environments with constrained action spaces, where the value functions are typically non-smooth, our method with structural maximization outperforms traditional actor-critic methods with gradient-based maximization. We have released our code at https://github.com/USC-Lira/Q3C.