Neuro-Control of an Underactuated Manipulator by Evolutionary Computation : Numerical Simulation and Experimental Verification
Akira ABE, Takuya Sato · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 2009
This paper describes a neuro-control scheme for motion control of nonlinear mechanical systems. A two-link manipulator with an unactuated first joint, wherein the motion of the system is confined to a horizontal plane, is treated as an example for the nonlinear systems. The control objective is to move the first joint from an initial angle to a target angle. Radial basis function neural networks (RBFNs) are used as the neuro-controller (NC), and particle swarm optimization (PSO), which is one of evolutionary computation techniques like genetic algorithm, is adopted for the learning algorithm. The effectiveness of the proposed approach is verified by a comparison of simulation and experimental results.