<研究ノート>局所知識に基づくデッドロックの解消 : 通信交渉と行動観測による移動ロボットの動作計画

概要

論文の詳細を見る
This paper discusses the problem of how to break down a deadlock situation that happens quite often in a multiple robot system. The global-information-based negotiation is applied most popularly for dealing with a deadlock problem. However, the limitation of the application has also been revealed. This is because that this kind of negotiation always needs a very strong and powerful communication system as base. Here, a local-information-based approach for solving the deadlock problem is proposed. In our system, a robot is provided with the function of action observation so that it can obtain some new information about other robots without direct communication. This paper shows how the function of action observation can make a contribution to deadlock problem solving. A robot here also does not need to know all about other robots that are involved in the same deadlock. It will select one of the robots within its sensing range and negotiate with that robot based on its sensed local information. The negotiation in our proposed approach needs less communication and simpler rules so it is useful for an autonomous and flexible multi-robot system.
1999-03-05