Distributed Motion Planning for Learning Agents in Cellular Warehouse Problem

Abstract A distributed approach to obtain a motion sequence of transport tables for cellular warehouse problem is shown. In the proposed approach, the tables are considered to be autonomous agents, and a built-in behavior function given by ANNs and the evolved problem-oriented connection weights navigate the agents to their specified goals. To detennine the agent to be moved, a measure of the priority to move is introduced. Through various numerical experiments, we show the applicability of the proposed method and examine the contribution of the evaluation function to the learning result of behavior function.

[1]  Richard E. Korf,et al.  Planning as Search: A Quantitative Approach , 1987, Artif. Intell..

[2]  Peter J. Angeline,et al.  An evolutionary algorithm that constructs recurrent neural networks , 1994, IEEE Trans. Neural Networks.

[3]  Richard E. Korf,et al.  Depth-First Iterative-Deepening: An Optimal Admissible Tree Search , 1985, Artif. Intell..

[4]  Majid Sarrafzadeh,et al.  Discrete Warehouse Problem , 1995, Theor. Comput. Sci..

[5]  Shuji Tsukiyama,et al.  A consideration on a sliding palette problem in a two-dimensional automatic warehouse , 1990, IEEE International Symposium on Circuits and Systems.