Light-emitting diode (LED)-based optical communication is emerging as a promising low-power, low-cost, and high-data-rate alternative to acoustic communication for underwater applications. However, it requires a close-to-line-of-sight (LOS) link between the communicating parties.Achieving and maintaining the LOS is challenging due to the constant movement of underlying mobile platforms caused by propulsion and unwanted disturbances. In this dissertation, a novel, compact LED-based wireless communication system with active alignment control is presented that maintains the LOS despite the movement of the underlying platform. Multiple alignment control algorithms are developed for scenarios that range from a simple one-way two-dimensional (2D) setting to a practical three-dimensional (3D) bi-directional underwater setting. An extended Kalman filter (EKF)-based approach is first proposed to estimate the relative orientation between the heading angle and the LOS direction, which is subsequently used for alignment control. The EKF uses only the measurement of light intensity from a single photo-diode, where successive measurements are obtained via a scanning technique that ensures the full observability of the underlying system. The approach is first examined in a 2D setting, and then extended to the 3D scenario with improvements in both the hardware and the algorithm. The amplitude of the scanning is modulated according to the alignment performance to achieve a sound trade-off between estimation accuracy, signal strength, and energy consumption. The efficacy of the approach is tested and verified via simulation and on an experimental setup involving two robots with relative 3D motion. The EKF approach uses an assumption that the relative motion between the robots is small, and consequently, requires the communicating robots to take the scanning in an alternating fashion for the convergence of the estimator. An alternative approach, first explored in the 2D setting, is developed that allows simultaneous, bi-directional alignment control for both parties. Because of the convex nature of the measured intensity functions, model-free approaches, including both hill-climbing (HC) and extremum-seeking (ES), are explored. The hill-climbing approach is found to be superior to the ES approach in terms of convergence time and computational efficiency. Theoretical analysis is provided for the hill-climbing approach that guarantees finite time convergence to an $O(\delta)$ neighborhood of the LOS, for control step size $\delta$.Finally, a model-free approach for the 3D setting is proposed that maximizes light intensity based on three consecutive intensity measurements from an equilateral triangle configuration. The efficacy of the approach is demonstrated experimentally, first with an underwater robot controlled by a joystick via LED communication and then with two robots performing bi-directional communication and tracking in an underwater setting.
Read
