GuardAD: Safeguarding Autonomous Driving MLLMs via Markovian Safety Logic

TL;DR

GuardAD introduces a Markovian safety logic framework for autonomous driving MLLMs, enhancing safety and robustness in dynamic traffic scenarios through continuous hazard inference and action refinement.

Contribution

It proposes a novel Neuro-Symbolic Logic Formalization and Logic-Driven Action Revision to improve safety reasoning without altering the underlying MLLMs.

Findings

Reduces accident rates by 32.07% in benchmarks.

Improves task performance by 6.85%.

Validated effectiveness through simulations and real-world tests.

Abstract

Multimodal large language models (MLLMs) are increasingly integrated into autonomous driving (AD) systems; however, they remain vulnerable to diverse safety threats, particularly in accident-prone scenarios. Recent safeguard mechanisms have shown promise by incorporating logical constraints, yet most rely on static formulations that lack temporally grounded safety reasoning over evolving traffic interactions, resulting in limited robustness in dynamic driving environments. To address these limitations, we propose GuardAD, a model-agnostic safeguard that formulates AD safety as an evolving Markovian logical state. GuardAD introduces Neuro-Symbolic Logic Formalization, which represents safety predicates over heterogeneous traffic participants and continuously induces them via n-th order Markovian Logic Induction. This design enables the inference of emerging and latent hazards beyond single-step observations. Rather than simply vetoing unsafe actions, GuardAD performs Logic-Driven Action Revision, where inferred safety states actively guide action refinement without modifying the underlying MLLM. Extensive experiments on multiple benchmarks and AD-MLLMs demonstrate that GuardAD substantially reduces accident rates (-32.07%) while slightly improving task performance (+6.85%). Moreover, closed-loop simulation evaluations, together with physical-world vehicle studies, further validate the effectiveness and potential of GuardAD.