Generating Critical Scenarios for Testing Automated Driving Systems
Trung-Hieu Nguyen, Truong-Giang Vuong, Hong-Nam Duong, Son Nguyen, Hieu Dinh Vo, Toshiaki Aoki, Thu-Trang Nguyen
TL;DR
AVASTRA addresses the challenge of safely evaluating autonomous driving systems by generating realistic, safety-relevant critical scenarios in simulation. It formulates scenario generation as an RL problem using a holistic state representation that combines ADS internal components and external environment, and employs a 45-action space with heuristic constraints to ensure realism. The approach uses a Double Deep Q-Network with Prioritized Experience Replay to learn environment configurations that maximize collision likelihood, measured via a probabilistic collision metric ProC. Across four road configurations on two maps, AVASTRA outperforms state-of-the-art DeepCollision and Random Search in generating more collision scenarios, while providing insights into state/action contributions, parameter effects, and training-scenario sensitivity. The results demonstrate AVASTRA’s potential to enhance safety testing for ADSs, informing both simulator-based testing strategies and future improvements in realistic scenario generation.
Abstract
Autonomous vehicles (AVs) have demonstrated significant potential in revolutionizing transportation, yet ensuring their safety and reliability remains a critical challenge, especially when exposed to dynamic and unpredictable environments. Real-world testing of an Autonomous Driving System (ADS) is both expensive and risky, making simulation-based testing a preferred approach. In this paper, we propose AVASTRA, a Reinforcement Learning (RL)-based approach to generate realistic critical scenarios for testing ADSs in simulation environments. To capture the complexity of driving scenarios, AVASTRA comprehensively represents the environment by both the internal states of an ADS under-test (e.g., the status of the ADS's core components, speed, or acceleration) and the external states of the surrounding factors in the simulation environment (e.g., weather, traffic flow, or road condition). AVASTRA trains the RL agent to effectively configure the simulation environment that places the AV in dangerous situations and potentially leads it to collisions. We introduce a diverse set of actions that allows the RL agent to systematically configure both environmental conditions and traffic participants. Additionally, based on established safety requirements, we enforce heuristic constraints to ensure the realism and relevance of the generated test scenarios. AVASTRA is evaluated on two popular simulation maps with four different road configurations. Our results show AVASTRA's ability to outperform the state-of-the-art approach by generating 30% to 115% more collision scenarios. Compared to the baseline based on Random Search, AVASTRA achieves up to 275% better performance. These results highlight the effectiveness of AVASTRA in enhancing the safety testing of AVs through realistic comprehensive critical scenario generation.