Solution of inverse kinematics problems of a highly kinematically redundant manipulator using genetic algorithms

This paper investigates using genetic algorithms (GAs) to solve the inverse kinematics problem for a manipulator with high kinematic redundancy. This is a global nonlinear constrained optimisation problem which involves searching for a set of manipulator joint positions to place the end effector at a specified position and orientation. The problem is complicated by the presence of obstacles that limit the postures available. The problem is first attempted using a traditional blind GA. It is then shown that performance can be improved by enhancing the GA with heuristics based on knowledge of manipulator kinematics and by prolonging diversity in the GA.