Solution of inverse kinematics by PSO based on split and merge of particles

Abstract

This paper investigates the use of particle swarm optimization (PSO) with repeating splits and merges at predetermined intervals. After a split, there is no exchange of information between the particles, which belong to different swarms. Hence, even though one particle may be trapped in a local minimum, the others are not affected. If the other particles find a better solution, the trapped particle can escape the local minimum when the particles are merged. In order to verify the efficacy of the proposed method, we applied it to the learning of a neural network for solving the inverse kinematics problem of a manipulator with uncertain parameters. The back-propagation rule requires the Jacobian of the forward kinematics, but this cannot be calculated due to uncertainties. Because PSO does not require the derivative of the objective function, it is suitable for this problem. A simulation result shows that the proposed method can obtain more accurate inverse kinematics than either global best (gbest) PSO or local best (lbest) PSO. © 2013 Springer-Verlag.