Intelligent Motion Control and Planning of Autonomous Underwater Vehicles

Autonomous underwater vehicles are vitally important tools with widespread applicationsfor ocean exploration, environmental monitoring, underwater rescue, wreckage recovery,etc. The developments of autonomous underwater vehicles have brought enormouseconomic and scientific benefits for mankind. It emerged as an interdisciplinary area asa synergistic integration of mechanics, electronics, control theory, and other relatedfields of science and engineering. Despite the rapid recent developments, the researchfaces many obstacles. Specifically, the control performance is often compromised due toseveral intrinsic properties of the subjects such as nonlinearity, uncertainty, andnonholonomicity. High-performance motion planning and control methods with the-state-of-the-art integrated technologies are highly demanded for more deployment andapplications of various autonomous underwater vehicles.Originating from engineering process control, an advanced model-based optimal controltechnique, called model predictive control, has been explored in recent years forapplications in areas such as robotics, aerospace and automotive engineering. Because ofits desirable characteristics, model predictive control is widely considered as an attractivemotion control technique in both academia and industries. However, several challengingissues on steering autonomous underwater vehicles remain unaddressed, such as themodel fidelity, real-time optimization efficiency, disturbance attenuation, and robustnessagainst model mismatch and parameter perturbation.In parallel to the research of autonomous underwater vehicles and model predictivecontrol in the past few decades, neural network research has made significant progress,aiming at building brain-like models for modeling complex systems and computingoptimal solutions. It is envisioned that the advances in neural network research will playa more important role in the motion control of autonomous underwater vehicles.In this proposed research, we will develop high-performance motion controlmethodologies for various autonomous underwater vehicles. Our primary objective istargeted to model nonlinearity, uncertainty, and nonholonomicity. The research willconsist of two coherent parts. In the first part, we will develop robust model predictivecontrol and other approaches based on neural computation for steering autonomousunderwater vehicles. Various specific features and issues of autonomous underwatervehicles such as nonlinear dynamics, persistent uncertainties, and underactuation will bethoroughly analyzed. The second part will focus on the applications of the intelligentcontrol methodology to specific motion control tasks. In particular, we will make in-depthinvestigation on the control of underwater gliders which are more practicallyuseful for many oceanographic missions. It is expected that the outcomes of the proposedproject will significantly advance the frontiers of motion control of autonomousunderwater vehicles in both theories and applications.