Abstract:
Objective To address the problem of ultra-short baseline (USBL) relative navigation and positioning during the dynamic docking of an autonomous underwater vehicle (AUV) in shallow water, a compensated unscented particle filter (CUPF) algorithm based on a disturbance observer is proposed.
Methods The CUPF algorithm compensates for missing values in the USBL data based on the dead reckoning model, estimates the unknown disturbances in the dynamic docking of the AUV using observer technology and filters outliers in the USBL data combined with the unscented particle filter to realize the estimation of the AUV's state.
Results The lake trial data shows that the proposed CUPF algorithm can effectively remove outliers and fill in missing values for USBL positioning with a speed estimation error of less than 15% and computation time reduced by 57% compared with traditional methods.
Conclusions The CUPF algorithm can improve the positioning accuracy of USBL while smoothing the AUV motion trajectory and state estimation by fully utilizing relative measurement information for AUV docking.