Self-organizing Fuzzy Radial Basis Function Neural-network Controller for Robotic Motion Control

Robotic systems are complicated, nonlinear, multiple-input multiple-output (MIMO) systems, which make the design of model-based controllers for robotic systems particularly difficult. Moreover, to achieve reasonable control performance, the dynamic coupling effects between degrees of freedom (DOFs) of the robotic systems must be overcome during the control process. Although a model-free, self-organizing fuzzy controller (SOFC) can be applied to the manipulation of complex and nonlinear systems, its parameters are difficult to select appropriately, and it mainly focuses on controlling singleinput single-output systems rather than MIMO systems. To address these problems, this study developed a self-organizing fuzzy radial basis function neural-network controller (SFRBFNC) for robotic systems. The SFRBFNC introduces a radial basis function neural-network into the SOFC to compensate for the dynamic coupling effects between the DOFs of the robotic system, as well as solve the problem caused by inappropriate selection of parameters in designing an SOFC. The SFRBFNC demonstrated control performance superior to the SOFC, as shown in experimental results from motion control tests of a 6-DOF robot.