Genetic algorithms for coordinating multi-agent robotic systems

Abstract

This paper provides a genetic algorithm for solving the object-sorting task. The object-sorting task is a complex multi-agent task which requires the cooperation of multiple agents for searching and moving the objects to their destinations. The agent-object sequence is utilized to represent the solutions and for performance evaluation. The object-agent sequence is developed for chromosome representation and GA operations. A distributed help model and a centralized furniture mover model are used for performance comparison. The experimental results showed that the GA is better than the two compared models. In addition, each chromosome operated in the evolution process is designed to be feasible so as to reduce the execution time.