Artificial Neural Network Based Control Strategy Research and Simulation on Robot Uncalibrated Visual Servoing System

摘要

This paper presents a control strategy and new simulation model for the visual servoing control system with an eye-in-hand configuration from a 6-DOF robot of puma560. Because of the complexity of the calibration method on calculating image Jacobian matrix, an uncalibration method based on neural network is proposed. Firstly, simulation model of image-based robot visual servoing control system in Matlab7.8.0(R2009a) is established. Then, the concept of uncalibration is introduced and BP neural network controller is used as the visual controller instead of the calibration method to calculate image Jacobian matrix. Furthermore a convenient and wide range simulation model of robot uncalibrated visual servoing control system based on BP neural network is designed. The simulation results show that the simulation model is feasible, and can be achieved uncalibrated visual positioning.