arXiv:2608. 11451v1 Announce Type: cross Abstract: Modern end-to-end driving agents can achieve high average performance yet still violate basic traffic rules that a human driver would never miss.
arXiv:2610.09294v1 Announce Type: cross Abstract: Rapid progress in AI agents has brought growing attention to agent safety, with extensive evaluation focused on digital environments. As agents move...
The paper introduces CoPhy, a Cognitive‑Physical reinforcement learning framework for autonomous driving that addresses the limitations of end‑to‑end models by combining a cognitive foundation and a foresighted physical environment. CoPhy distills vision‑language model knowledge into a BEV encoder, then removes the VLM to keep cognitive capability at zero inference cost while allowing optional human language commands. An auto‑regressive BEV world model predicts future semantic maps conditioned on candidate actions, providing an interpretable sandbox for safety metrics, and the driving policy is optimized with a dual‑reward mechanism that enforces safety constraints and intent compliance. Extensive experiments show CoPhy achieves state‑of‑the‑art results on NAVSIM v1 and v2 and enables safer, intent‑controlled driving.
CoSafe is a modular, ADS-agnostic framework that extends runtime safety assurance from ego-centric takeover to scene-level cooperative intervention in mixed-traffic autonomous driving. It uses vehicle-to-everything communication to allow roadside units to monitor hazards and jointly mitigate risks involving multiple vehicles. Evaluation on the InterDrive benchmark shows that CoSafe improves driving scores by 70.03%, reduces safety violations by 87.80%, and accurately identifies unsafe autonomous vehicles with an F3 score of 0.930 while meeting real-time constraints.
The paper introduces NegWAM, a safety post‑training framework for unified world‑action models (WAMs) used in end‑to‑end autonomous driving. NegWAM generates negative trajectories that are close to expert paths but contain safety risks, using them as repulsive anchors to steer predictions away from dangerous behavior while encoding scene‑trajectory compatibility via rider tokens. The method achieves a 4.1% relative improvement on the NAVSIM v2 navhard benchmark and 93.3 EPDMS on navtest, demonstrating strong safety performance without sacrificing routine driving, and shows cross‑benchmark generalization in zero‑shot closed‑loop tests on HUGSIM.
OPTED is a method for on‑policy fine‑tuning of end‑to‑end driving models that separates reinforcement learning from the policy update. A privileged teacher trained with RL on vectorized inputs (HD‑maps and bounding boxes) supervises the pre‑trained student during closed‑loop post‑training. Applied to the camera‑based models TransFuser and VaVAM in AlpaSim, OPTED boosts driving scores by 1.6× and 9.5×, respectively, while requiring roughly three orders of magnitude fewer simulator interactions than direct RL post‑training.
arXiv:2604. 15579v2 Announce Type: replace-cross Abstract: There is increasing interest in integrating AI agents that invoke tools into domain-specific commercial software, where unintended tool calls can cause serious security and safety incidents.
The paper introduces Closed-Loop Aware Planning for Autonomous Driving (CLAP‑AD), a new planning head that learns a driving path via imitation while independently scoring candidate speeds using a learned function guided by safety constraints. Unlike traditional open‑loop imitation learning, CLAP‑AD allows explicit test‑time control over speed selection without requiring speed‑conditioned training. Experiments on Bench2Drive‑Speed, Bench2Drive, and NAVSIM v2 demonstrate significant improvements in speed‑adherence and overall driving performance compared to state‑of‑the‑art baselines.
Agent safety is moving from content moderation toward preventing operational failures before tool-using agents act. We propose Janus, a foresight-oriented framework for long-horizon agent safety that trains guards to anticipate delayed risks from partial trajectories.
arXiv:2607. 19913v1 Announce Type: new Abstract: Agent safety is moving from content moderation toward preventing operational failures before tool-using agents act.
arXiv:2606. 23938v1 Announce Type: new Abstract: Driving VLA models incorporating Chain-of-Thought (CoT) reasoning are attractive because they leverage pretrained VLM representations and expose intermediate decisions in natural language, yet current rationales often lack the step-by-step decision semantics needed to keep the rationale causally connected to the planned motion.
arXiv:2609.10377v1 Announce Type: cross Abstract: Safety is a fundamental requirement for autonomous driving, yet existing end-to-end driving models still lack explicit risk-aware learning capacities...
FPD-Drive is an end‑to‑end autonomous driving planner that uses future‑privileged multi‑target distillation to transfer diverse, high‑quality trajectory targets from a teacher model with access to future observations to a student policy. It also introduces safety‑aware score correction to improve candidate selection by adjusting rankings between safe and unsafe trajectories. The approach achieves state‑of‑the‑art performance on NAVSIM v1 and v2 benchmarks and demonstrates strong zero‑shot closed‑loop results on HUGSIM.