On the solution of multimodal robot inverse kinematic functions using real-coded genetic algorithms

Abstract

The inverse kinematics solution of a robotic manipulator involves transcendental equations having multiple solutions. Multiplicity resolution can be achieved by incorporating performance criterion, like total joint displacement, in the solution scheme. A real-coded genetic algorithm is used in which the fitness value of the feasible solutions, satisfying the robot kinematic equations, is taken as the total joint displacement. The fitness value of the infeasible solutions is taken as the violation of the robot kinematic equations. A multimodal robot inverse kinematics function is formed from the robot kinematic equations. A niching strategy is used to obtain simultaneously the optimal solutions of this multimodal robot inverse kinematics function. These optimal solutions provide the feasible solutions satisfying the robot kinematic equations. The fitness values associated with these feasible multiple solutions provide a means for multiplicity resolution. The approach is used to obtain the inverse kinematics solution of an articulated robotic manipulator.