Nonlinear model predictive control of an uninhabited surface vehicle

This paper presents a novel nonlinear autopilot design based on a nonlinear model predictive control (NMPC) approach that is compared against a linear quadratic Gaussian control scheme. The autopilot systems are used to control the nonlinear yaw dynamics of an uninhabited surface vehicle named Springer. The yaw dynamics of the vehicle being modelled using a multi-layer perceptron neural network. Simulation results are presented and the performances of the autopilots are evaluated and compared using standard system performance criteria and indices. The autopilot based on the NMPC method is deemed the more apt of the two types examined for Springer in terms of control activity expenditure, power consumption and mission duration length.