Performance Evaluation of Feed Forward Neural and Recurrent Neural On Real System Dataset of Robot Execution

Abstract

This article presents approach based on the artificial neural networks (ANN). It is employed to evaluate of performance real date set of real system. The training and testing dataset used in the experiment consists of forces and torques memorized immediately after the real robot failed in assignment execution. Two types of neural networks (NN) are utilized in order to find best performance method - feed forward neural networks (FFNN) and recurrent neural networks (RNN) and an additional evaluation would be to run test sets for each neural network to see how small an error is produced. Moreover, we investigated 24 neural structures implemented in Matlab software. The obtained results confirm that this approach can be successfully applied in this domain.