Seeking to Collide: Online Safety-Critical Scenario Generation for Autonomous Driving with Retrieval Augmented Large Language Models

TL;DR

This paper presents an online, retrieval-augmented large language model framework for generating safety-critical driving scenarios to improve autonomous vehicle testing, outperforming existing offline methods in exposing rare corner cases.

Contribution

Introduces a novel online LLM-based framework with retrieval augmentation for dynamic, safety-critical scenario generation in autonomous driving testing.

Findings

Reduces mean minimum time-to-collision from 1.62 to 1.08 seconds.

Achieves a 75% collision rate in evaluations.

Outperforms baseline methods in generating safety-critical scenarios.

Abstract

Simulation-based testing is crucial for validating autonomous vehicles (AVs), yet existing scenario generation methods either overfit to common driving patterns or operate in an offline, non-interactive manner that fails to expose rare, safety-critical corner cases. In this paper, we introduce an online, retrieval-augmented large language model (LLM) framework for generating safety-critical driving scenarios. Our method first employs an LLM-based behavior analyzer to infer the most dangerous intent of the background vehicle from the observed state, then queries additional LLM agents to synthesize feasible adversarial trajectories. To mitigate catastrophic forgetting and accelerate adaptation, we augment the framework with a dynamic memorization and retrieval bank of intent-planner pairs, automatically expanding its behavioral library when novel intents arise. Evaluations using the Waymo Open Motion Dataset demonstrate that our model reduces the mean minimum time-to-collision from 1.62 to 1.08 s and incurs a 75% collision rate, substantially outperforming baselines.