Modified and Fuzzified General Problem Solver for "Monkey and Banana" Problem : Strategy of General Problem Solver

Abstract

Automatic operation is important for the in-service inspection of nuclear power stations or the decommission of retired plants. Master & slave control will be introduced for work-robot control. It is desirable that the slave possess problem-solving capabilities. In this paper we assume that the slave incorporates the general problem solver (GPS) algorithm. In view of having solved the "monkey and banana" problem, the slave system is regarded as a reasonable alternative which incorporates the capability of problem-solving. Basically, the GPS solves a problem by reducing the difference between the initial state and goal state, and hence the performance of GPS depends on selection of the difference to be reduced. The conventional GPS is given the order of importance of the differences in advance. In this study, the GPS was improved by making use of the rules which determine the order. When several choices are available for the given difference, a fuzzified decision to determine the necessary action is made, as demonstrated in this paper.