AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

ScenePilot: Controllable Boundary-Driven Critical Scenario Generation for Autonomous Driving

Qiyu Ruan, YUXUAN WANG, He Li, Zhenning Li, Cheng-zhong Xu

ICML 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Data-driven traffic simulation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add physical-feasibility constraint to AV-risk-only boundary (ablation)ScenePilot · SafeBench Red-light RunningCR: 0.37 -> 0.93Improved
add AV-risk constraint to physical-feasibility-only boundary (ablation)ScenePilot · SafeBench Red-light RunningCR: 0.75 -> 0.93Improved
add physical-feasibility constraint to AV-risk-only boundary (ablation)ScenePilot · SafeBench Right TurnCR: 0.12 -> 0.91Improved
add AV-risk constraint to physical-feasibility-only boundary (ablation)ScenePilot · SafeBench Right TurnCR: 0.74 -> 0.91Improved
use both boundaries rather than AV-risk only (ablation)ScenePilot · SafeBench Red-light RunningGap Coverage Score: 0.096 -> 0.209Improved
use both boundaries rather than physical-feasibility only (ablation)ScenePilot · SafeBench Right TurnGap Coverage Score: 0.133 -> 0.351Improved
use ScenePilot instead of ChatScene as best baselinescenario generator · SafeBench (8 base scenarios)CR: 0.831 -> 0.893Improved
use ScenePilot instead of KING on heterogeneous AV stacksscenario generator · SafeBench Scenario 6CR: 0.380 -> 0.490Improved
add 30 or 50 benign background vehiclesScenePilot · SafeBench Scenario 6CR: 0.89 -> 0.91No clear effect
add 50 benign background vehiclesScenePilot · SafeBench Scenario 6OS: 0.507 -> 0.492No clear effect
adversarially fine-tune ego on ScenePilot scenesSAC ego policy · SafeBench held-out routesCR: 0.854 -> 0.072Improved
fine-tune on ScenePilot scenes instead of ChatScene scenesSAC ego policy · SafeBench held-out routesOS: 0.869 -> 0.898Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy