The goal of the project is to develop an acoustical navigation system for an AUV (Autonomous Underwater Vehicle). The AUV is to be used for environmental studies and for inspections of installations on the seabottom. In the project, methods for detection and identification of objects on basis of sonar signals are developed.
The information from the developed system is, among other things, to be used to avoid unknown objects under water. The sonar used has a very high resoloution and therefore it produces a very large amount of data which is to be processed in a comparatively short time to ensure realtime detection of obstacles for the AUV. It is tried to combine methods from adjacent research areas: low resoloution sonars, medical ultrasound techniques, digital image processing, methods for object tracking, e.g. kalman filter techniques, and clasification techniques.
Shown below is an unprocessed sonar image together with the processed version of the same image. Note that the objects are isolated quite well although the image is of very poor condition. Especially it is concluded that the very noisy area of the image does not contain any obstructing objects.
The entire Martin concept is developed by the following institutes, companies and organizations. The persons listed are the contact persons:
Department of Automation (IAU): Sonar image processing, Object avoidance, Pathplanning, Mission Management
Morten Lind, Professor: Coordinator & administrator
Lars Henriksen, M.sc.EE., Ph.D. student: Processing of sonar images.
Mikael Bliksted Larsen, Research assistant, M.sc.EE.: Pipescan project.