Depth Sensor-based Precise 3D Correction Method for USBL Position Estimate

This study presents a method for precise 3D correction of USBL (Ultra-Short Baseline) position estimates through depth sensor measurements. Since USBL’s vertical and horizontal position estimates are correlated, correcting the vertical estimate through depth sensor measurements also improves horizontal accuracy. However, due to the nonlinear characteristics of the USBL positioning model, conventional fusion methods require linearization or sampling, potentially introducing errors and distortions. To address this issue, the proposed method corrects the model inputs, such as measurement data and system parameters, rather than the USBL position estimate, which represents model output. The proposed method was validated through simulations using USBL and depth sensor models based on related studies. The results demonstrate that the proposed approach improves both the accuracy and precision of vertical and horizontal positions and achieves superior performance, particularly under low-elevation conditions.