Parameter tuning of a computed-torque controller for a 5 degree of freedom robot arm using co-evolutionary particle swarm optimization

This paper proposes an accelerated co-evolutionary particle swarm optimization method to solve the problem of parameter tuning of a feedforward computed-torque controller (CTC) of a 5-DOF robot arm manipulator. The approach is based on a competitive co-evolutionary technique, which simultaneously evolves two populations of particles. The first population optimizes the feedforward CTC parameters and the second population searches for the disturbance values in the worst case. Both together generate the robust solution. The co-evolutionary PSO is modified in a way that prevents premature convergence to solutions that are located at the boundary of the parameter space. The robustness of the solution is measured by the capability of the tuned feedforward CTC to demonstrate good trajectory tracking performance in the presence of disturbances. Simulations of the closed loop system demonstrate that this control strategy provides robust control and improves trajectory tracking of a perturbed 5-DOF robot arm manipulator.