Search papers, labs, and topics across Lattice.
The paper introduces PtoP, a novel simulation-based testing framework for autonomous driving systems that uses Stein Variational Gradient Descent (SVGD) to generate diverse, failure-inducing initial conditions. By balancing attraction to high-risk regions and repulsion between particles, PtoP overcomes the mode collapse problem of existing search-based methods. Experiments on CARLA with industry-grade ADS (Apollo, Autoware) demonstrate that PtoP significantly improves safety violation rate, scenario diversity, and map coverage compared to baselines.
SVGD lets autonomous vehicle testers find 28% more safety violations by intelligently exploring the simulation space, even with complex systems like Apollo and Autoware.
Simulation-based testing of autonomous driving systems (ADS) must uncover realistic and diverse failures in dense, heterogeneous traffic. However, existing search-based seeding methods (e.g., genetic algorithms) struggle in high-dimensional spaces, often collapsing to limited modes and missing many failure scenarios. We present PtoP, a framework that combines adaptive random seed generation with Stein Variational Gradient Descent (SVGD) to produce diverse, failure-inducing initial conditions. SVGD balances attraction toward high-risk regions and repulsion among particles, yielding risk-seeking yet well-distributed seeds across multiple failure modes. PtoP is plug-and-play and enhances existing online testing methods (e.g., reinforcement learning--based testers) by providing principled seeds. Evaluation in CARLA on two industry-grade ADS (Apollo, Autoware) and a native end-to-end system shows that PtoP improves safety violation rate (up to 27.68%), scenario diversity (9.6%), and map coverage (16.78%) over baselines.
