Motion controller design for two-wheeled robot based on a batch learning structure

A learning architecture with a fuzzy inference system and genetic algorithm (GA) is proposed to automatically determine a motion controller for two-wheeled robots. This architecture in each generation can be separated into three states: a control state, a system identification state, and a controller learning state. Two fuzzy inference systems are used in the proposed learning architecture. One is used to be a fuzzy controller and the other one is used to be a fuzzy identifier. The antecedent and consequent parameters of the fuzzy system are viewed as a parameter set and a fitness function is proposed in a GA method to choose an appropriate parameter set of the fuzzy system so that the selected fuzzy system has a good performance. Some practical tests are presented to illustrate that trajectories of the controlled robot from the initial point to the target position are short and straight.