2000 Volume 15 Issue 1 Pages 119-128
This paper proposes a new approach to the path planning problem with dynamic change of movement time of paths. For the dynamic environmental change, we try to cope with it to modify a temporal plan dynamically without using any estimated functions. In our proposed method, firstly a temporal path plan is generated based on the initial state of the problem. Secondly, every one move to the next node, according to the observation of environmental change, the temporal plan is dynamically modified in order to reach the goal in shorter movement time, if necessary. In order to find shortcut-paths, we introduce a multi-agent model, in which agents search for them by diffusing messages to neighboring agents. Since where and whom agents send messages is determined based on the present status of the problem, modifications are executed appropriately and locally, and admissible solutions are obtained efficiently. We examine the relation between effective behavior and a variety of settings of parameters for the problem. As a result of the experiment, we show that our proposed method is more useful for the problem in which the average of delay of movement time is longer and the existence probability of delay of movement time is smaller.