Fuzzified PSO-SVM controller for motion control of robotic manipulator

The tremendous advantages of SVM make it one of the most emerging and proving techniques of estimating nonlinear quantities of the era. SVM performance can be enhanced by selecting optimal value of few of its kernel constant parameters for which a global optimisation technique, naming PSO can be successfully employed. For performance up gradation of PSO, fuzzy logic is used to adjust adaptively the inertia weights (w) and acceleration coefficients (c1 and c2) of PSO. This fuzzified PSO is used for obtaining optimal value of SVM parameters. Such hybrid of fuzzy logic, PSO and SVM proposed controller has not been found in literature yet. For validation purpose, the proposed controller has been applied for motion control of a robotic manipulator. Simulation results proves that the proposed controller remarkably outperforms the basic SVM and PSO-SVM controllers in terms of tracking performance and reduced 2-norm of control input. For validity, performance of various controllers has been checked in the presence of normally affecting uncertainties (payload changes and friction) of the manipulator system by incorporating the actuator dynamics in it.