Autonomous underwater vehicle localization using ocean tomography sensor nodes

Abstract

The work is a preliminary study of autonomous underwater vehicle (AUV) localization using ocean acoustic tomography (OAT) sensor nodes. OAT is a method to measure the current velocity and temperature fields using the time-of-flight of acoustic signals. These acoustic sensor nodes broadcast signals periodically in the area of interest. Each ocean tomography sensor node consists of a surface buoy, an acoustic transceiver, GPS module and the processing unit. An acoustic tomography sensor was installed on an AUV, acting as a moving node in addition to the seafloor moored nodes. Since the GPS signal is not always available on AUV, the chip scale atomic clock is operated as the system clock to reduce the clock drift. With the scheduled transmission times, the one-way travel time is measured between the moored nodes and the AUV. The high accuracy of travel time between moored nodes and AUV is used to improve the AUV localization. Based on dead-reckoning and travel-time measurement, Extended Kalman Filter is employed to improve the vehicle localization. An experiment was conducted using an AUV and one moored node deployed in a shallow water environment southwest of Taiwan. The AUV was operated near the surface to obtain the GPS position as the ground truth for the performance evaluation. Preliminary results show the RMS position error between EKF prediction and the ground truth is about 80 m by incorporating the tomographic signals into a low-cost AUV navigation system.