Learning Control Strategies Based on Neural Network for an Underwater Vehicle

This paper introduces a new approach to underwater vehicle controller design which involves a learning control strategy based on artificial neural networks. Such networks are currently the subject of intensive research in a number of disciplines and we examine here their applicability in feedback control systems. Some preliminary findings are presented to demonstrate how a neural network can learn the essential features of an optimal controller and, following this training period, how the network might behave in various control tasks.