Abstract
Many underwater applications that involve the use of autonomous
underwater vehicles require accurate navigation systems. Image
registration from acoustic images is a technique that can be used to
achieve this task by comparing two consecutive sonar images and estimate
the motion of the vechicle. The use of deep learning (DL) techniques for
motion estimation can significantly reduce the processing complexity and
achieve high-accuracy position estimates. In this paper we investigate
the performance improvement when using two sonar sensors compared to
using a single sensor. The DL network is trained using images generated
by a sonar simulator. The results show an improvement in the estimation
accuracy when using two sensors.