Multiresolution wavenet PID control for global regulation of robots

Abstract

A novel global PID control scheme for nonlinear MIMO systems is proposed and synthesized for a robot manipulator. Identification process is used for online tuning of the discrete linear PID feedback gains. Inverse dynamics identification is based on radial basis neural network with daughter RASP1 wavelets activation functions in cascaded with an infinite impulse response (IIR) filter in the output to prune irrelevant signals and nodes. The closed-loop system guarantees global regulation for a class of nonlinear dynamical systems, convenient for instance in plants whose dynamics are rather uncertain or unknown, such as in commercial robots. Real-time experimental study is carried out on a three degrees of freedom robotic haptic interface, the PHANToM Premium 1.0A. Results highlight the performance in global regulation with smooth control effort, without using the mathematical model of the robot, nor cumbersome tuning procedures typical of PID for robots.