A learning mechanism for parts recognition in an intelligent assembly system

In this paper, a learning mechanism (LM) for parts recognition in an intelligent assembly system is presented. It differs from the mechanism used in the standard back propagation (SBP) neural network in two ways. First, the searching direction is changed from the negative gradient direction to the variable metric direction. Secondly, the constant learning rate is changed to a variable optimal learning rate. The combination of these two improvements leads to the training process being greatly speeded up, and convergence is assured. Several application examples are presented. Results indicate that the proposed LM is superior to the SBP in learning speed, convergence and stability.