Get Started
Home
Topics
Search
Library
Research questionHow can autonomous-driving planners be stress-tested in realistic closed-loop scenarios that expose failures missed by nominal benchmarks?Nominal benchmark performance can miss failures that emerge when traffic participants create rare, interacting hazards. Closed-loop evaluation needs scenarios that remain realistic while probing those behaviors.
AI
Alignment & Safety
Diffusion Models
Evaluation & Benchmarks
Multi-agent Systems
Robotics
Latest papersRecent research connected to this question, newest first.One Diffusion Model, Two Roles: Guided Trajectory Planning and Safety-Critical Scenario Generation in Closed-Loop SimulationThe source uses a single pretrained diffusion traffic model for ego-motion planning and controllable scenario generation. Inference-time guidance steers selected agents toward aggressive cut-ins, lead-vehicle braking, and combined longitudinal-lateral interactions. The evidence comes from closed-loop nuPlan simulations with independent black-box planners and reports planner degradation and exposed failure modes under these scenarios.research paper · Sep 4, 2026
Related questions
How can low-cost platforms enable reproducible testing of end-to-end autonomous-driving policies across simulation and physical vehicles?How can hybrid autonomous-driving planners align trajectory generation and refinement when their objectives differ?How can vehicle-infrastructure cooperative driving be evaluated for error accumulation under occlusion without costly full simulation?How can autonomous-driving scenario generators reliably induce collisions at a requested region of the target vehicle?