Optical underwater distance estimation

Data communication with high data rate and precise underwater positioning with an accuracy of several centimeters is an unsolved problem. Precise positioning is important for autonomous vehicles. State-of-the-art acoustic communication and distance estimation faces difficulties underwater, e.g. multipath fading or variation of propagation speed. In this work, we propose optical distance estimation, which is the basis for positioning in future work. We combine the Beer-Lambert law and the inverse-square-law to model the underwater channel of the medium. We investigate different wavelengths and employ curve fitting based on the Levenberg-Marquardt algorithm to determine the unknown coefficients of the model, e.g. absorption. Our evaluation shows promising results and distance estimation of up to 25 m in pool water is possible. In stream water we determined a mean error for the optical distance estimation of up to 0.02 m with a wavelength of 470 nm.