Incorporating strategy adoption into genetic algorithm enabled multi-agent systems

Abstract

Genetic Algorithm (GA) is a widely adopted optimization technique under evolutionary optimization. Inspired by the evolutionary operators of selection, crossover and mutation, Genetic Algorithms have been used to successfully solve myriad optimization problems in a wide range of domains, including in optimizing multi-agent systems. On the other hand, Evolutionary Game Theory (EGT) is used to model social-economic systems by mimicking social evolution by adopting neighborhood strategies in a stochastic manner. In this work, an extended GA is proposed for multi-agent systems, which incorporates the strategy adoption in EGT into GA enabled multi-agent systems. The proposed extended GA algorithm is applied to an example multi-robot navigation application. The proposed algorithm gives promising results in terms of the convergence time, compared to the GA based approach. Possible applications of the proposed algorithm are also discussed, while indicating potential future research directions.