A design method for an image-based fuzzy trajectory tracking controller to be applied to real robots by using a genetic algorithm

This paper proposes a method for designing an image-based fuzzy controller for trajectory tracking by using a genetic algorithm (GA). Although a controller designed by a GA cannot work well on real robots as same as simulations due to modeling errors, this paper tries to reduce the influence of the modeling errors by adding motion noises to the simulations in the design process. The proposed method is evaluated through some simulations and real robot experiments for a target trajectory consisting of a straight and a sine wave. It is shown from experimental results that the target trajectory can be tracked not only in simulations but also to a real robot by introducing an appropriate evaluation function for the GA and adding the amount of motion noises properly.