A Method Proposed for Training An Artificial Neural Network used for Industrial Robot Programming by Demonstration

Robot programming by demonstration has become a central topic in the field of robotics. Artificial neural networks play an important role in this type of robot programming. Artificial neural networks have a great disadvantage: the network must to be trained with a huge number of data in order to achieve good results. In our case (industrial robot programming by demonstration), it is necessary to train the neural network in one single step, when the robot is trained with some data. In this paper we propose a method for artificial neural network training, which works in these conditions. The main idea of this method is to train the artificial neural network with all of the data, before the current training step. At a certain step the network is already trained a huge number of times. A software application was designed for testing the method. This software application implements the training method on a unidirectional multi-layer neural network, using back propagation error algorithm. The results obtained using the software application are also presented.