A Novel Mutating PSO Based Solution For Inverse Kinematic Analysis Of Multi Degree-Of-Freedom Robot Manipulators

Abstract

A novel mutating particle swarm based optimization (MuPSO) technique for solving inverse kinematic problems of robot manipulators was proposed. The parameters of the standard PSO (SPSO) were found to produce non-converging solutions, so the parameters were modified to accommodate the complex structure of robotic inverse kinematic problems. The new parameters were also made adaptive, changing non-linearly with time. A forced mutation function was introduced capable of pushing the algorithm out of stagnation (at local optima). A mathematical model for multi degree-of-freedom (DOF) robot inverse kinematics and the proposed MuPSO was developed, the mechanism and condition for mutation were also expressly defined. The new MuPSO algorithm was found to efficiently derive solutions for complex inverse Kinematic problems and satisfactorily break out of stagnation. It was compared with a previously proposed algorithm and was found to completely dominate that solution.