Configuration space control of a parallel Delta robot with a neural network based inverse kinematics

Abstract

This paper describes configuration space control of a Delta robot with a neural network based kinematics. Mathematical model of the kinematics for parallel Delta robot used for manipulation purposes in microfactory was validated, and experiments showed that this model is not describing “real” kinematics properly. Therefore a new solution for kinematics mapping had to be investigated. Solution was found in neural network utilization, and it was used to model robot's inverse kinematics. It showed significantly better mapping between task space coordinates and configuration (joint) space coordinates than the mathematical model, for the workspace of interest. Consequently positioning accuracy improvement is expected. Neural network is then used as a part of the control system. Applied control strategy was configuration space acceleration control with disturbance observer.