Robust adaptive neuro-fuzzy control for nonholonomic mobile modular manipulators in task space

A robust adaptive neural-fuzzy (NF) controller is presented in this paper for trajectory following of nonholonomic mobile modular manipulators in task space. On the basis of modular robot concept, an integrated dynamic modeling method is proposed in consideration of the interactive motions and the nonholonomic constraints. NF systems are used as estimators to approximate dynamic model of the robot via off-line training, and adaptive on-line adjustment makes the controller be more robust. Sliding mode control and robust fuzzy logic control are introduced to suppress such errors as caused by parameter uncertainties and bounded external disturbances. Since this controller is designed in task-space directly, calculation of inverse Jacobian can be avoided. The proposed algorithm does not need exact dynamic parameters in advance and rules explosion can be avoided effectively. Simulation results for a real robot composed of a 4 degree of freedom (DOF) modular manipulator and a 3-wheeled nonholonomic mobile robot demonstrate that the proposed algorithm is effective