Submap based Normal Distribution Transform Scan Matching by using Heterogeneous Sonars for AUV Navigation

In this study, we propose two difference sonars based scan matching method for AUV application. One sonar that we used is an acoustic lens-bead multibeam sonar (ALMS) called DIDSON, which scans short-range horizontal information, and another one is a mechanically scanning imaging sonar (MSIS), which scans medium-range vertical profile of the front area of AUV. The data measured in different viewpoints are combined and represented using a probabilistic approach named normal distribution transform (NDT). In addition, there exists a time difference between the two data from the sonars with different ranges, which is utilized to reduce the drift errors. By matching the data with less-accumulated errors and the data with more-accumulated errors, drift errors are reduced, and consistent map is generated. The proposed method and experimental results are presented.