Adaptive Neural Tracking Control for an Autonomous Azimuthing Stern Drive (ASD) Tug

Abstract

The ASD tug is considered the modern habour tug due to its versatility in performing a wide variety of tugging/towing operations. With the advent of autonomous systems and desire to improve operations, there are benefits to automating the tug operations. Under different operating environments, there could be uncertainties in the model and the disturbance from the environment significant enough to affect the performance and control of the tug. In this paper, we consider the problem of tracking a desired trajectory for an autonomous ASD tug in the presence of uncertainties and unknown disturbances. The numerical modelling of an ASD tug based on a modified 4-DoF MMG model is first presented. The modified Maneuvering Modeling Group (MMG) model is then used to design a model based backstepping control. Thereafter, a adaptive neural network approximator is introduced which has the capability to account for uncertainties and unknown disturbance. The combination of approximation-based and backstepping design techniques allows us to handle time-varying model uncertainties. Stability analysis is carried out for the control design via Lyapunov analysis. Simulation are carried out for the tracking of maneuvering motion paths to demonstrate the performance of proposed approach.