VL-SAFE: Vision-Language Guided Safety-Aware Reinforcement Learning with World Models for Autonomous Driving

-- Reinforcement learning (RL) - based autonomous driving policy learning faces critical limitations such as low sample efficiency and poor generalization; its reliance on online interactions and trial - and - error learning is especially unacceptable in safety - critical scenarios. Existing methods including s afe RL often fail to capture the true semantic meaning of "safety" in complex driving contexts, leading to either overly conservative driving behavior or constraint violations . To address these challenges, we propose VL - SAFE, a world model - based safe RL framework with Vision - Language model ( VLM) - as - safety - guidance paradigm, designed for offline safe policy learning. Specifically, we construct offline datasets containing data collected by expert agents and labeled with safety scores derived from VLMs. A world model is trained to generate imagined rollouts together with safety estimations, allowing the agent to perform safe planning without interacting with the real environment. Based on these imagined trajectories and safety evaluations, actor - critic le arning is conducted under VLM - based safety guidance to optimize the driving policy more safely and efficiently. Extensive evaluations demonstrate that VL - SAFE achieves superior sample efficiency, generalization, safety, and overall performance compared to existing baselines. To the best of our knowledge, this is the first work that introduces a VLM - guided world model - based approach for safe autonomous driving. The demo video and code can be accessed at: https://ys - qu.github.io/vlsafe - website/ Index Terms -- vision - language models; world model; safe reinforcement learning; autonomous driving C urrent transportation management methods have produced significant improvements [3], [4] . Autonomous driving, with its advanced intelligence and automation [5], offers a promising solution . At its core, autonomous driving aims to perceive [6], understand [7], and interact [8] with complex dynamic systems in real time. Yansong Qu is with the Lyles School of Civil and Construction Engineering, Purdue University, West Lafayette, 47907, USA (e - mail: qu 120@purdu.edu