Depth control of an unmanned underwater vehicle using neural networks

Artificial neural networks offer an alternative strategy for the non-linear control of unmanned underwater vehicles (UUVs). This paper presents the results of a simulation study into the development of a neural network controller for depth control of a UUV. Results presented compare the performance of the neural controller based on the multilayered perceptron (MLP) chemotaxis training algorithm with proportional-integral-derivative (PID) controller. Results will show that in the presence of noise and change in mass of the vehicle the neural controller out performed the standard PID controller