DiffScene: Diffusion-Based Safety-Critical Scenario Generation for Autonomous Vehicles

Abstract

The field of Autonomous Driving (AD) has witnessed significant progress in recent years. Among the various challenges faced, the safety evaluation of autonomous vehicles (AVs) stands out as a critical concern. Traditional evaluation methods are costly and inefficient, often requiring extensive driving miles in order to encounter rare safety-critical scenarios, which are distributed along the long tail of the complex real-world driving landscape. In this paper, we propose a unified framework, Diffusion-Based Safety-Critical Scenario Generation (DiffScene), to generate high-quality safety-critical scenarios, which are realistic and safety-critical for efficient AV evaluation. In particular, we propose a diffusion-based generation framework, leveraging its power of approximating the distribution of low-density spaces. We design several adversarial optimization objectives to guide the diffusion generation under predefined adversarial budgets. These objectives, such as safety-based objective, functionality-based objective, and constraint-based objective, ensure the generation of safety-critical scenarios while adhering to specific traffic constraints. Extensive experimentation has been conducted to validate the efficacy of our approach. Compared with 6 SOTA baselines, DiffScene generates scenarios that are (1) more safety-critical under different metrics, (2) more realistic under 5 distance functions, and (3) more transferable to different AV algorithms. In addition, we demonstrate that training AV algorithms with scenarios generated by DiffScene leads to significantly higher performance under safety-critical metrics. These findings highlight the potential of DiffScene in addressing the challenges of AV safety evaluation and enhancement, paving the way for safer AV development.