2000 年 36 巻 12 号 p. 1152-1161
Vertical transportation in buildings is mainly dealt with elevators. In high-rise buildings, such transportation is quite expensive because the elevators occupy large floor area and labor cost is consumed only for transportation. So as to improve the quality of vertical transportation service and to reduce the cost for it, the authors have proposed to separate transportation of human and material handling in high-rise buildings, and proposed an automatic material handling system consisting of autonomous guiding vehicle (AGV)s and elevators. In the present paper, control strategies of the automatic material handling system are discussed. Concept of autonomous decentralized systems is introduced, and a control strategy of multi-agent type is adopted. For cooperation among the agents such as the AGVs and the elevators, the contract net protocol (CNP) is adopted. Performance of the system is enhanced by neural networks for estimation of accurate completion time, and genetic algorithms for refinement of the bidding scheme taking various factors into consideration in CNP.