SELECTION OF NETWORK AND LEARNING PARAMETERS FOR AN ADAPTIVE NEURAL ROBOTIC CONTROL SCHEME

This paper presents the results of a study in the design of a neural n etwork based adaptive robotic control scheme. The neural network used here is a two hidden layer feedforward network and the learning scheme is the well-known backpropagation algorithm. The neural network essen tially provides the inverse of the plant and acts in conjunction with a standard PD controller in the feedback loop. The objective of the co ntroller is to accurately control the end position of a single link ma nipulator in the presence of large payload variations, variations in t he link length and also variations in the damping constant. Based on r esults of this study, guidelines are presented in selecting the number of neurons in the hidden layers and also the parameters for the learn ing scheme used for training the network. Results also indicate that i ncreasing the number of neurons in the hidden layer will improve the c onvergence speed of learning scheme up to a certain limit beyond which the addition of neurons will cause oscillations and instability. Guid elines for selecting the proper learning rate, momentum and fast backp ropagation constant that ensure stability and convergence are presente d. Also, a relationship between the r.m.s. error and the number of ite rations used in training the neural network is established.