Deployment of newly developed autonomous vehicle (AV) technology on public roads may result in exposure to the public of high levels of danger. A proactive metric, as opposed to a reactive metric, is required to evaluate the safety of an automated system without relying on the rate of vehicle accidents and fatalities relative to the number of miles driven. Resilience engineering (RE) is a new safety management paradigm that focuses on examining the interaction of complex systems with their environment. This chapter presents the concept of RE and the resilience assessment grid (RAG) as an evaluation tool for measuring the operational resilience of AV systems. This evaluation tool was used to test the responsiveness of a Pure Pursuit controller (PPC) and a Deep Learning controller (DLC) in the CARLA simulator. The path tracking control algorithms were evaluated using traditional reactive and proactive RE measures tested at various speeds. RE proved to be a straightforward method for enhancing performance and assessing the operational robustness of an AV controller and thus is suitable for evaluation of other AV subsystems as well as the AV system itself.
Evaluation of Autonomous Vehicle Control Strategies Using Resilience Engineering
Machine Learning and Optimization Techniques for Automotive Cyber-Physical Systems ; Chapter : 25 ; 709-733
2023-09-02
25 pages
Article/Chapter (Book)
Electronic Resource
English
Performance Evaluation of an Autonomous Vehicle Using Resilience Engineering
SAE Technical Papers | 2022
|Performance Evaluation of an Autonomous Vehicle Using Resilience Engineering
British Library Conference Proceedings | 2022
|Performance Evaluation of an Autonomous Vehicle Using Resilience Engineering
British Library Conference Proceedings | 2022
|Using Reinforcement Learning and Simulation to Develop Autonomous Vehicle Control Strategies
British Library Conference Proceedings | 2020
|