Map uncertainties for unmanned underwater vehicle navigation using sidescan sonar

Abstract

Navigation of unmanned underwater vehicle (UUVs) in unstructured environments without long baseline transponders is a challenging problem due to the lack of predefined sea floor landmarks. This study proposes a real time sea floor mapping method using sidescan sonar on UUVs. The mapping system combines the motion information and the observation model of the sidescan sonar. The observation process firstly detects environmental features and then localizes the feature positions relative to the UUV. The landmark localization procedure considers the working principle of sidescan sonar and the acoustic velocity in sea water to localize the landmark and calculate the sidescan sonar footprint region. An occupancy grid mapping algorithm is then used to map the sea floor combining the uncertainties of UUV's motion and measurement errors from the observations. This procedure was implemented on an UUV to verify its landmark detection capability. Experimental results conducted on sandy seabed with concrete targets are demonstrated.