Nowadays, in computer sciences, intelligent agents are becoming more and more popular. Such systems offer a logical approach to conceive a system where diverse modules interact to solve complex real world problems. Lockeed Martin Canada (LMC) and the Defence Research and Development Canada - Valcartier (DRDC Valcartier) have provided a real world multiagent application. Our main task for this application is to conceive different multiagent techniques to improve the defensive effectiveness of a Halifax class Canadian frigate against incoming missile threats.; A Halifax class frigate uses different modules that interact together to defend itself and it is necessary to propose ways to optimize the coordination, between them in order to increase their efficiency. In this thesis, we will focus mainly on multiagent planning. We first present some theoretical bases on planning, different planning approaches, then we investigate how to improve the effectiveness of defence for the frigate using different strategies. Finally, we provide different experimental results.; Our experimental results show that: (1) using missile reengagement against incoming threats is more efficient with few threats than with many threats; (2) a central coordinator coordination technique between hardkill and softkill agents is better than with two other coordination techniques; (3) a Bayesian approach for the frigate positioning enhances the frigate's chance of survival.
展开▼
