The capabilities of marine vehicles, such as autonomous underwater vehicles (AUVs) and remotely-operated vehicles (ROVs), have significantly advanced to perform complicated operations such as navigating through unexplored and dynamic environments, inspection of underwater structures, and monitoring of ocean conditions in deep waters. However, the limited energy resources of AUVs and tethered connectivity of ROVs constrain survey lengths and subsequently increase the operational costs. Underwater docking stations that can recharge vehicles and transmit their data offer a potential solution to extend the endurance of vehicles. In this work, we consider the docking station to be powered through a floating wave energy converter (WEC), thereby enabling on-site energy harvesting and power transfer. As a main contribution, we present a navigation framework that couples flow state estimation with model predictive control (MPC) to perform autonomous underwater docking with a WEC under various ocean conditions.
Unknown background flow can cause perturbations to an underwater vehicle's motion, leading to an unsuccessful attempt at docking. Typically, acoustic Doppler current profilers (ADCPs) are used to estimate flow velocity but they are expensive. In this paper, we adapt the approach presented in , which considers the coupling between vehicle motion and ocean currents to facilitate flow state estimation, for reliable docking with fewer failures.
Uniform complete observability of the flow state, along the vehicle trajectory, is maintained by obtaining optimal control input that minimizes the maximum variance of the flow state estimation. To facilitate this, F is defined as the information metric, that corresponds to the observability and performance of the flow state estimation, and can be described as
F = λmin(Gcg (-∞, k + 1)) (1)
where k is the discrete timestep, λmin(.) denotes the smallest eigenvalue of (.) and Gcg represents the constructability gramian of the linearly time-varying ocean flow system.
Extending our previous work , this article presents a navigation framework that incorporates flow state estimation into the MPC formulation for a more reliable and robust docking approach. This is accomplished by modifying the objective function to include the information metric, defined in (1).
The key significance of the proposed approach is that it enables the vehicle to counter the influence of unknown flow disturbances while simultaneously moving closer to the docking station, in an optimal manner.
Initial results demonstrate the vehicle successfully approaching the docking station while countering the influence of ocean currents along its path. Additionally, we reformulated the problem in a sequential manner and validated its performance under various ocean conditions. Furthermore, we are also explicitly modeling the dynamics of different types of WECs. Future work includes conducting field trials on a real robot and extending this concept to support wave-current interactions.