Acoustic-optic assisted multisensor navigation for autonomous underwater vehicles

Underwater navigation is a challenging problem in the field of Autonomous Underwater Vehicles (AUVs) due to significant electromagnetic wave attenuation and uncertainties in the underwater environment. In general, the AUV's navigation system uses a number of different sensors, including Doppler Velocity Log (DVL) or transponders, etc., to perform the positioning task. Recently, the use of Visual Odometry (VO) based on underwater optical data or Acoustic Odometry (AO) based on high-frequency sonar data, to assist AUV navigation has become a potential research field. However, due to the different turbidity of seawater, insufficient light, and other factors, there are loopholes in the method based on a single sensor, which needs to be applied simultaneously to improve navigation quality and robustness. The use of multi-sensor navigation techniques can effectively achieve high navigation effects, but will also lead to an increase in the computational burden, which needs to be balanced. Therefore, this paper adopts a sensor fusion framework based on a parallel federated Unscented Kalman filter (UKF) to reduce the computation of the filter and employs the trapezoid formula to further improve the accuracy. The experimental data collected from the WEBOTS platform has been used to verify and compare different navigation methods.