A novel localization-free approach to system identification for underwater vehicles using a Universal Adaptive Stabilizer

In this work, we propose a new approach to the otherwise lengthy and involved process of system identification for underwater vehicles. The algorithm is based on Universal Adaptive Stabilization (UAS). The proposed methodology does not require localization equipment as it is based on the more easily attainable linear and angular velocity measurements. It also does not rely on accurate initial estimates to converge. The algorithm is proven to be stable, and its capability to yield the correct parameters of the vehicle is established. In simulation, multiple classes of vehicles are tested to verify the ability of the algorithm to identify varying arrangements of parameters depending on the class and shape of a given underwater vehicle. The algorithm is validated using simulation and experimental data. The simulation tests show very promising identification results. The algorithm is additionally validated on a publicly available Autonomous Underwater Vehicle (AUV) dataset. Experimental results establish the capacity of the algorithm to realize accurate estimates using nothing but generally available state measurements using sensors available on commercial underwater vehicles. In both testing environments, the algorithm produced parameter estimates with accuracy upwards of 95% in a Normalized Mean Absolute Error (NMAE) sense.