Traveling-salesman competition with multiple agents and simultaneous move search algorithm

摘要

In this paper, we study a traveling-salesman type competition in which several players travel around various objects distributed in a given area and compete with each other in collecting the objects. Similar competitions can be widely found in the real world such as in selling competition, in the competitive gathering of scraps, wastes, seafood or wild foods, etc. In this competition, players move simultaneously and the number of players can be more than two. We present a novel scheduling algorithm suitable for the competition in which a move toward an object along the shortest path is taken as the basic move, and the move is stopped halfway when opponents have the advantage. In the algorithm, the moves of the players can be dealt with quasi-alternately. However, a player can make several consecutive moves, and a move already made can be aborted. A position evaluation method based on the concept of territory is also presented. A simulation program is developed and competitions are held among computers or between a computer and a human being. It is confirmed that the algorithm and the position evaluation method can be effectively applied to the game model.