Improved MHE-Based Tightly Coupled SINS/USBL Method for Underwater Navigation

In the field of underwater integrated navigation, multi-sensors are required to obtain accurate navigation information. Traditional sensor data fusion algorithms include Kalman filter (KF), extended Kalman filter (EKF), etc. However, these filtering algorithms do not take into account the constraint problems in practical systems, such as the limitation of effective measurement range of sensors, underwater dynamic environmental noise, physical boundaries of state variables, etc. This may lead to inaccurate navigation information. Moving horizon estimation (MHE) can explicitly handle the state estimation problem for constrained systems. It has better applicability and accuracy for state estimation in nonlinear systems with constraints and uncertainties. At the same time, the prior covariance matrix in the traditional MHE is computed using EKF. However, in order to simplify the calculation, EKF brings a problem of precision decline. In response to the above issues, proposing an Improved MHE-Based Tightly Coupled SINS/USBL Method for Underwater Navigation. Using the improved adaptive unscented Kalman filter to compute the prior covariance matrix in the cost function and employing the pigeon swarm algorithm to solve the problem. Simulation and experimental results verify that this method is more applicable and accurate than conventional methods for nonlinear systems.