Application of genetic algorithm to scheduling problem of robot control computation

This paper presents a new method for solving the scheduling problem of robot control computation based on the genetic algorithm (GA). It is proved that the scheduling problem belongs to the class of NP-hard problems. For the scheduling problem, the authors have already proposed two algorithms: one is a heuristic approach and the other is based on branch-and-bound approach. These conventional algorithms, however, have the limits of their ability in quality of solutions and computational time. The scheduling problem is a typical partitioning problem: partitioning N objects into P groups to optimize an objective function. Therefore, each feasible solution is represented by a way of division of a suffix set (1,...,N) into P subsets. In the proposed GA for the scheduling problem, such a feasible solution is regarded as a phenotypic individual. Then this paper proposes a new phenotype based crossover operation named "weighted-edge crossover". By using the proposed crossover, a child inherits the desirable characteristic from both parents, keeping its structure as a feasible solution. Furthermore, in order to improve the performance of the crossover operation, this paper proposes a distance function between phenotypic individuals and uses it in the adaptive control of crossover rate. To demonstrate the efficiency of the proposed GA, comparative study with the conventional algorithms is carried out on several experiments.