Genetic Algorithms as a feasible re-planning mechanism for Belief-Desire-Intention Agents
Gail Shaw, Etienne van der Poel · 2015
Belief-Desire-Intention (BDI) agents traditionally use a fixed, pre-defined plan library with the assumption that the environments in which they operate remain unchanged. Often agents have to operate in changing environments with the results that plans become outdated or even fail completely. The result has been much research on several approaches to re-planning mechanisms for BDI agents. Genetic Algorithms (GA) are well known as an optimization technique with wide application, however they have not been applied as a re-planning mechanism for BDI agents. This paper explores the application of GAs as a mechanism for re-planning in BDI agents in a controlled, simulated environment and shows that GAs are a feasible re-planning mechanism for BDI agents.