AnchorDrive: LLM Scenario Rollout with Anchor-Guided Diffusion Regeneration for Safety-Critical Scenario Generation

Researchers have developed AnchorDrive, a two-stage AI framework that combines large language models with diffusion models to generate realistic safety-critical scenarios for autonomous driving systems. The system uses LLMs for controllable scenario generation based on natural language instructions, then employs diffusion models to create realistic driving trajectories.

• →AnchorDrive combines LLMs and diffusion models to generate safety-critical autonomous driving scenarios with improved controllability and realism.
• →The two-stage framework first uses LLMs for semantic control through natural language instructions, then applies diffusion models for realistic trajectory generation.
• →The system addresses the challenge of rare safety-critical scenarios in real-world driving data through simulation-based synthesis.
• →Experiments on the highD dataset demonstrate superior performance in criticality, realism, and controllability metrics.
• →The framework includes a plan assessor that provides corrective feedback to improve scenario generation quality.