Dynamics of an Underwater Drifting Instrument: System Modeling and Controller Design for the uFloat

Abstract

This thesis explores optimal depth control for the uFloat, an affordable oceanographic float designed for swarm sensing in coastal waters. The current Proportional-Integral-Derivative (PID) control system combined with a Butterworth filter on the pressure sensor data enables constant depth and constant velocity deployments, but float trajectories could be improved and power use could be reduced. Ship autopilots and autonomous underwater vehicles have made use of the optimal, model-based Linear Quadratic Gaussian (LQG) control method, which combines an Extended Kalman filter (EKF) and a Linear Quadratic Regulator (LQR) to create a stable, optimal feedback control system. Nonlinear and linear models of the uFloat dynamic system were developed and used with an LQG controller that included a nonlinear EKF state estimator and the LQR feedback controller. The EKF estimator delivered smoothed depth and velocity signals without the delay of the Butterworth filter. The simulated LQG controller was able to perform constant depth and constant velocity deployments with more efficient actuation than PID control, but currently has greater overshoot. The dynamic model developed in this thesis and the LQG controller elements could assist with ongoing development of the uFloat, and other buoyancy-driven floats.