Sparse Threats, Focused Defense: Criticality-Aware Robust Reinforcement Learning for Safe Autonomous Driving
This work addresses safety-critical robustness for autonomous driving systems, representing an incremental improvement over existing adversarial training methods.
The paper tackles the vulnerability of reinforcement learning in autonomous driving to perturbations by proposing CARRL, a criticality-aware robust RL method that models adversarial training as a general-sum game to handle sparse safety-critical risks, resulting in at least a 22.66% reduction in collision rates compared to state-of-the-art baselines.
Reinforcement learning (RL) has shown considerable potential in autonomous driving (AD), yet its vulnerability to perturbations remains a critical barrier to real-world deployment. As a primary countermeasure, adversarial training improves policy robustness by training the AD agent in the presence of an adversary that deliberately introduces perturbations. Existing approaches typically model the interaction as a zero-sum game with continuous attacks. However, such designs overlook the inherent asymmetry between the agent and the adversary and then fail to reflect the sparsity of safety-critical risks, rendering the achieved robustness inadequate for practical AD scenarios. To address these limitations, we introduce criticality-aware robust RL (CARRL), a novel adversarial training approach for handling sparse, safety-critical risks in autonomous driving. CARRL consists of two interacting components: a risk exposure adversary (REA) and a risk-targeted robust agent (RTRA). We model the interaction between the REA and RTRA as a general-sum game, allowing the REA to focus on exposing safety-critical failures (e.g., collisions) while the RTRA learns to balance safety with driving efficiency. The REA employs a decoupled optimization mechanism to better identify and exploit sparse safety-critical moments under a constrained budget. However, such focused attacks inevitably result in a scarcity of adversarial data. The RTRA copes with this scarcity by jointly leveraging benign and adversarial experiences via a dual replay buffer and enforces policy consistency under perturbations to stabilize behavior. Experimental results demonstrate that our approach reduces the collision rate by at least 22.66\% across all cases compared to state-of-the-art baseline methods.