PhD Dissertation Proposal by SUN Yang | Specification-based Autonomous Driving

Autonomous driving systems (ADSs) necessitate comprehensive testing prior to deployment in autonomous vehicles. High-fidelity simulators facilitate this testing against a wide range of scenarios, including those challenging to replicate in real-world environments. While previous approaches have shown that test cases can be generated automatically, they tend to focus on weak oracles (e.g. reaching the destination without collisions) without assessing whether the journey itself was undertaken safely and satisfied the law. In this dissertation, we aim to develop a specification-based autonomous driving platform.

To achieving this goal, we start with developing LawBreaker, an automated framework for testing ADSs against real-world traffic laws, which is designed to be compatible with different scenario description languages. LawBreaker provides a rich driver-oriented specification language for describing traffic laws, and a fuzzing engine that searches for different ways of violating them by maximising specification coverage. 

In the second research work, we explore methods for creating `natural' scenarios.  This is achieved through the manipulation of the positions of commonly encountered roadside objects, without resorting to the use of `unnatural' adversarial patches.  To ensure the realism of these scenarios, they must simultaneously satisfy rules that encode regulatory guidelines about the placement of objects on public streets. Our approach involves the development of a fuzzing algorithm to identify scenarios in which the repositioning of these objects induces significant misperceptions by the AV, such as misinterpreting the color of a traffic light. The ultimate objective is to cause the AV to violate traffic laws.

Lastly, we present REDriver, a general and modular approach to runtime enforcement, in which users can specify a broad range of properties (e.g., national traffic laws) in a driver-oriented specification language based on signal temporal logic~(STL). REDriver monitors the planned trajectory of the ADS based on a quantitative semantics of STL, and uses a gradient-driven algorithm to repair the trajectory when a violation of the specification is likely.

Sun Yang is a Ph.D. candidate at SMU SCIS, supervised by Prof. SUN Jun. Yang's research focuses on evaluating and improving Autonomous Driving Systems.