Abstract The use of Autonomous Vehicles (AVs) on our streets is soon to be a reality; increasingly, interacting with such AVs will be part of our daily routine. However, we will certainly need to assure the reliable behaviour of an AV, especially when some unexpected scenarios (e.g. harsh environments, obstacles, emergencies) are taken into account. In this article we use an intelligent agent approach to capture the high-level decision-making process within an AV and then use formal verification techniques to automatically, and strongly, analyse the required behaviours. Specifically, we use the MCAPL framework, wherein our core agent is implemented using the GWENDOLEN agent programming language, and to which we can apply model checking via the AJPF model checker. By performing such formal verification on our agent, we are able to prove that the AV’s decision-making process, embedded within the GWENDOLEN agent plans, matches our requirements. As examples, we will verify (formal) properties in order to determine whether the agent behaves in a reliable manner through three different levels of emergency displayed in a simple urban traffic environment.
Reliable Decision-Making in Autonomous Vehicles
2019-11-11
13 pages
Article/Chapter (Book)
Electronic Resource
English
Ethical decision making for autonomous vehicles
IEEE | 2020
|ETHICAL DECISION MAKING FOR AUTONOMOUS VEHICLES
British Library Conference Proceedings | 2020
|Decision-making techniques for autonomous vehicles
TIBKAT | 2023
|Autonomous Vehicles Implementing User-Based Decision Making
European Patent Office | 2024
|