MACC'97 Session: Learning, Evolutionary Computing
In this paper, we propose the distributed search algorithm based on cooperative behaviors multi-agents for Vehicle Routing Problems. This algorithm is designed using multi-agent system, in which multi-agents autonomously and independently search the partial solution of problem to decrease their load of tasks usgin pheromone communication. Pheromone communication imitated from living creatures is used as the communication method between multi-agents to emerge the cooperative behavior. Especially, in order to form effective pheromone communication we define the action of agents based on observing other agents. Throughout some computer experiments, we confirm the effectiveness of proposed algorithm from viewpoint of cooperative behavior.
Vehicle Routing Problem(VRP), Traveling Salesman Problem(TSP), Combinatorial Optimization Problem, Multi-Agent
PS
file(+gzip) (in Japanese)