Publication Date
Fall 2024
Degree Type
Thesis
Degree Name
Master of Science (MS)
Department
Computer Engineering
Advisor
Younghee Park; Alvaro Cardenas; Jun Liu
Abstract
Autonomous vehicles utilize advanced safety features like proactive driving assistance and pre-collision alerts to minimize the risk of accidents. However, evaluating the correct functionality of these systems is complex. First, safety systems are highly sophisticated, integrating software, networking, and hardware components, many of which rely on advanced artificial intelligence and machine learning algorithms. Second, an array of dynamic factors, including numerous actors and physical variables, can influence the performance of safety mechanisms during critical scenarios. Each actor’s unique behavior introduces unpredictability, making it difficult to anticipate future states and outcomes. This thesis presents a comprehensive framework for testing autonomous vehicle safety systems, addressing the intricate challenges posed by complex environments and agent interactions. The methodology combines metric temporal logic with a probabilistic virtual scenario generator, alongside verification software, to systematically explore vast scenario spaces and identify potential safety vulnerabilities. The experiments focus on evaluating the functionality of the adaptive cruise control system under two distinct adversarial scenarios designed to challenge the system’s responses. The results provide a quantitative analysis of the safety performance of the autonomous vehicle, highlighting specific feature values that contributed to accidents, offering valuable insights into system vulnerabilities and areas for improvement.
Recommended Citation
Hernandez, Carlos O., "Uncovering Weaknesses in Autonomous Driving: a Formal Approach to Adversarial Scenario Generation and Falsification" (2024). Master's Theses. 5593.
DOI: https://doi.org/10.31979/etd.vqbt-kwk2
https://scholarworks.sjsu.edu/etd_theses/5593
