Current estimation and path following for an autonomous underwater vehicle by using a model-based nonlinear observer

The objective of this thesis is to contribute to Autonomous Underwater Vehicle (AUV) navigation by estimating the prevalent ocean currents and improving the path following guidance system. The Inertial Navigation System (INS) based localisation solution is vulnerable to uncertainties derived from double integration of the inherent errors within the INS acceleration measurements. This can be aided by the velocity measurement using the Doppler Velocity Log (DVL), but DVL aid is unavailable when the distance between the AUV to the seabed is larger than the DVL range. However, the vehicle's velocity can be estimated by a model-aided observer which predicts a dynamic motion response of the vehicle. The benefits from using a more precise AUV model is compelling as the application of underwater vehicles expand into more complicated and harsher environments. In terms of the AUV motion model, however, the effect of current on the vehicle dynamics is often ignored when the AUV motion model is used for control, navigation and estimation. As a solution for this, an AUV dynamic model-based, nonlinear observer design is introduced, which is complemented with the development of an AUV dynamic model in nonuniform and unsteady current and the high-gain observer (HGO). The gain for the HGO is obtained by solving a Linear Matrix Inequality (LMI) representing the estimation error dynamic. The HGO is a robust tool for observer design and well-used in nonlinear feedback control, which has been studied for several decades. Motivated by the design method of the HGO, the current disturbance is considered as the uncertainties of the vehicle dynamic system, and current velocity is estimated in an indirect way. The current velocity is determined by calculating the differences between the vehicle velocities over the ground and the vehicle velocities through the water estimated by the model-based observer. The HGO based on the AUV dynamic model established in this research was validated using experimental data from a set of field manoeuvres using a Gavia class AUV and the performance was compared against other commonly used navigation methods. In terms of the path following problem, an accurate path following guidance system plays an important role for an AUV in oceanic surveys and exploration. The path-following guidance system includes a guidance law, an update law and a proportional and integral controller. This thesis presents a three-dimensional path following guidance logic which ensures the vehicle path converges into the predefined desired path. The desired straight and curved path re represented by using a Serret-Frenet frame which propagates along the curve. The path following guidance system, 'PPNAPG (Pure Proportional Navigation and Pursuit Guidance)' is developed by combining the pure proportional navigation guidance (PPNG) law and pursuit guidance (PG) law, which are widely used in the missile community. The performance of the proposed path-following guidance system, PPNAPG is validated via both simulation and experiment using the MUN (Memorial University of Newfoundland) Explorer AUV. One of the primary benefits of the nonlinear observer based on the AUV dynamic model and the path following guidance system is that it could be employed on any type of AUV plat forms without any elaboration. Furthermore, the proposed techniques require no additional sensors beside the typically available AUV navigation sensors such as global positioning, accelerometers and gyroscopes in the Initial Measurement Unit (IMU). Overall, this thesis suggests that the nonlinear observers based on an AUV dynamic model and the PPNAPG is an effective combination to estimate and compensate for the current and complete a path-following mission. These outcomes and methods enable other researchers and students in the field of AUV control and navigation systems to adapt and extend the methods to other AUV models without using an ADCP to measure the current. The insight gained from this study can also be of assistance to an oceanographic mission by optimising the guidance law to reduce mission completion time.