New intelligent sonar system technology for autonomous underwater vehicles

With the increase of demands for autonomy for underwater related operations new challenges lay ahead. One of the biggest challenges fro successful autonomous operations is sensing. The new generation of AUVs will have to cope with difficult and constantly changing environment as well as coping with unplanned situations. Therefore autonomous vehicles need intelligent sensors and in particular intelligent sonar systems to improve their situation awareness and deal effectively with dynamic scenarios.

In the mine counter measure context image based automatic target recognition is confronted to mainly two limitations: in a cluttered and heavily cluttered environment recognition algorithms observe a drastic increase in the PFa (probability of false alarm) making the output results unpractical at the best. The second main limitation is unknown threats such as IEDs (Improvised explosive devices) or simply unknown types of mines.

Traditionally sonar systems have evolved in the last decades toward imagery systems. The inner reason behind this choice is that we human are more comfortable with images that we are with 1D signal such as sounds. Imagery sonars have come a long way and have improved drastically in term of noise and resolution. Some sonar systems now can offer centimetric pixel resolution. However the core process of creating an image (match-filtering, envelop detection, down-sampling) from a coherent system such as sonars suppress the information contained in the phase of the signal. Image based ATR has been an active field of inter-disciplinary research for the last decade. And so far all the proposed algorithms fail in difficult and cluttered environments and with unknown targets.

In order to access the information contained in the phase we took a step back from the imagery approach to come back to the raw acoustic signal. One of the main issue with sonar images is the speckle noise inherent to coherent systems. A lot of demonising algorithms have been developed over the years to smooth sonar images and make them "look better". However speckle is not a noise strictly speaking. Speckle is indeed the consequence of interference patterns due to different acoustic contributions within the target echoes. Wideband sonar systems such as BioSonar allow a deeper understanding of the echo structure. By tracking the phase within the echo we can unwrap the different echo contributions and then have access to more information than traditional sonar images such as structure, composition… In this paper we present examples of how the wideband sonar BioSonar can identify contents of similar objects, distinguish between manmade and natural objects or automatically segment seabed type by recovering their physical properties.