SARAD: LLM-Based Safety-Aware Hybrid Reinforcement Learning with Collision Prediction for Autonomous Driving

ArXi:2605.28583v1 Announce Type: cross Ensuring both safety and efficiency in decision-making for autonomous driving systems remains a fundamental challenge. Traditional Deep Reinforcement Learning (DRL) suffers from unsafe random exploration and slow convergence, while Large Language Models (LLMs) nstrate inherent latency in real-time inference operations. To address these limitations, this paper proposes SARAD, a novel safety-aware hybrid framework that synergizes LLMs and DRL for autonomous driving.