A Factor Graph-Based Multiple Sensors Fusion Method for Enhanced Underwater Navigation of Remotely Operated Vehicles

The requirement for precision in underwater navigation has driven the development of advanced remotely operated vehicle (ROV) systems, necessitating the integration of multiple sensors to achieve robust performance. This paper addresses the critical need for enhanced accuracy and reliability of ROV's navigation by proposing a factor graph-based fusion framework that synergistically combines multiple navigation sensors. Employing the sum-product algorithm, the framework adeptly fuses the data of multiple sensors, optimizing the estimation of ROV's state variables within a probabilistic graphical model. The methodology transforms the complex task of multiple sensors integration into a tractable problem of maximum a posteriori estimation, leveraging the inherent strengths of each sensor to mitigate individual weaknesses. Through simulation experiments, the proposed method demonstrates superior accuracy over traditional approaches, as evidenced by reduced root mean squared error (RMSE) in position, attitude, and velocity estimations. The findings underscore the efficacy of the factor graph model in providing a dependable navigation solution, setting a foundation for advancements in autonomous underwater systems and fault-tolerant navigation algorithms.