Comparative Studies of the Neural and Analytical Model of the PR-02 Arm End Movement

The paper presents selected results of comparative research carried out on a simulation model built in Simulink composed of among others, from the analytical model and the neural model of the PR-02 arm end movement. The neural model was designed and trained as a Perceptron Artificial Neural Network using data from the analytical model. On the basis of the comparative studies carried out in Simulink, high accuracy of the neural model in relation to the analytical model was obtained. Particular emphasis was placed on obtaining high accuracy of the trajectory performed by the end of the PR-02 robot during movement. The Perceptron Artificial Neural Network project was implemented in MATLAB and Simulink using Deep Learning Toolbox.