Controllable Collision Scenario Generation via Collision Pattern Prediction

2025-10-14

AI Summary

Autonomous driving safety testing urgently requires controllable generation of hazardous collision scenarios; however, real-world data collection is challenging, and existing simulation methods lack precise control over collision types and time-to-arrival (TTA). To address this, we formulate a novel task—controllable collision scenario generation—and propose an interpretable collision pattern representation. We introduce COLLIDE, the first large-scale, balanced collision dataset. Building upon it, we develop a collision pattern prediction framework that jointly models spatiotemporal configurations and employs adversarial trajectory generation to synthesize diverse, high-fidelity collision trajectories grounded in real driving logs. Experiments demonstrate that our method significantly outperforms baselines in collision rate, collision-type accuracy, and TTA control precision. Generated scenarios effectively expose vulnerabilities in motion planners, and fine-tuning planners on these scenarios improves their robustness. Our work establishes a new paradigm for safety validation of autonomous driving systems.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Web data generation and simulation
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Evaluating the safety of autonomous vehicles (AVs) requires diverse, safety-critical scenarios, with collisions being especially important yet rare and unsafe to collect in the real world. Therefore, the community has been focusing on generating safety-critical scenarios in simulation. However, controlling attributes such as collision type and time-to-accident (TTA) remains challenging. We introduce a new task called controllable collision scenario generation, where the goal is to produce trajectories that realize a user-specified collision type and TTA, to investigate the feasibility of automatically generating desired collision scenarios. To support this task, we present COLLIDE, a large-scale collision scenario dataset constructed by transforming real-world driving logs into diverse collisions, balanced across five representative collision types and different TTA intervals. We propose a framework that predicts Collision Pattern, a compact and interpretable representation that captures the spatial configuration of the ego and the adversarial vehicles at impact, before rolling out full adversarial trajectories. Experiments show that our approach outperforms strong baselines in both collision rate and controllability. Furthermore, generated scenarios consistently induce higher planner failure rates, revealing limitations of existing planners. We demonstrate that these scenarios fine-tune planners for robustness improvements, contributing to safer AV deployment in different collision scenarios.

Problem

• Generating diverse safety-critical collision scenarios for AV testing
• Controlling collision type and time-to-accident attributes in simulation
• Overcoming real-world data scarcity for autonomous vehicle safety evaluation

Innovation

• Predicts collision patterns before trajectory rollout
• Generates controllable collision scenarios via simulation
• Uses interpretable spatial configuration representation