Control System Design for Unicycle Robots Using Genetic Algorithms with Pareto Immunization

Classical genetic algorithms have been used in various optimization problems in engineering and other science fields. Mobile robots have complex nonlinear dynamics and finding optimal controllers is generally a difficult task. We propose an enhancement of the classical genetic algorithm that seeks to improve search efficiency when dealing with multiple conflicting criteria. Our solution consists in an immunization mechanism. Using an updating Pareto front, we create adaptive vaccines to help the population strengthen its desirable features during evolution. Numerical simulations show that a significant improvement has been obtained in terms of required number of generations to reach a desired optimum. Moreover, the returned solutions offer more consistent closed loop performances when using the proposed Pareto-based immunization mechanism.