A Kalman filter for the navigation of remotely operated vehicles

A Kalman based asynchronous data fusion algorithm for the navigation of a tethered remotely operated underwater vehicle is presented. Using a non-linear dynamic simulation of the tethered ROV system, the performance of the Kalman filter is measured for various motion sensor combinations. The sensor suite tested includes a Doppler velocity log, fiber-optic gyrocompass, depth sensor and an ultra-short baseline position system. Provided the gyrocompass functions properly, the study shows that an extended Kalman filter which uses a complete model of the ROV, including, drag, tether and thruster effects, does outperform a constant velocity model in instances of sensor drop out. The positioning error is reduced by 20% in these instances. It is found that the ultra-short baseline system is the driving factor in the smoothness of the results.