Controllable Collision Scenario Generation Via Collision Pattern Prediction
Pin-Lun Chen, Chi-Hsi Kung, Che-Han Chang, Wei-Chen Chiu, Yi-Ting Chen
AI summary
Problem
Existing simulation methods for generating safety-critical collision scenarios lack explicit controllability over key attributes like collision type and time-to-accident, hindering systematic autonomous vehicle safety evaluation.
Approach
The framework predicts a compact collision pattern representing the relative spatial configuration of vehicles at impact, conditioned on user-specified collision type and timing, then uses a polynomial planner to generate a kinematically feasible attacker trajectory.
Key results
- Introduced COLLIDE dataset with 8,586 balanced collision scenarios annotated by type and TTA
- Achieved 81% average collision rate across five types, outperforming strong conditional baselines
- Exposed planner blind spots by inducing higher failure rates in rule-based motion planners
- Improved planner robustness through fine-tuning on generated safety-critical scenarios
Why it matters
This work provides AV developers and safety evaluators with a scalable, controllable simulation tool to systematically stress-test and improve autonomous driving planners against diverse, high-risk collision scenarios.
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. Additional generated scenarios are available at this project webpage.