Task-Based Optimal Manipulator/Controller Design using Evolutionary Algorithms

A design methodology, based on evolutionary algorithms, for simultaneously opti-mizing the topology and the parameters of a serial robot manipulator along with the required feedback control laws for performing specified tasks is presented. Dynamic performance measures, for the closed-loop system, are optimized by considering a nonlinear PD controller for an end-effector tracking problem. Results of the simul-taneous design optimization involving the topology of a planar manipulator, joint motors, kinematic, inertial and control parameters as design variables illustrate the efficacy of the methodology