A composite controller based on fuzzy rules and neural networks for soccer robot system

Robot soccer has become more attractive after years of development, and the performance of the motion controller plays a decisive role in the whole system. This paper proposes a composite motion controller based on fuzzy rules and neural networks to substitute the former PID controllers. The introduction of neural networks using back-propagation algorithms guarantees the system´s response to the steady state, while the application of fuzzy rules enhances the quality of transient response. The new controller can learn from and adapt to different environments and overcome the existence of nonlinearity in the system.