Autonomous surface vehicle localization using sidescan sonar detections of Acoustic IDentification tags

Acoustic Identification tags more commonly referred to as AID tags have been developed by Satish et al. to serve as underwater acoustic landmarks that are designed to reflect unique backscattering signatures when ensonified by Sound Navigation and Ranging (SONAR) devices. If the location and the backscattering signature of a deployed AID tag is known a priori, then future detections of the AID tag by a SONAR-equipped autonomous vehicle can be used to localize the vehicle in relation to the known position of the tag. This capability allows autonomous systems to execute missions in underwater environments where accurate GPS may be challenging to obtain, and where optical sensing is limited owing to water turbidity or limited range of visibility. A series of experiments were conducted at Georgia Tech Research Institute (GTRI) where multiple AID tags were deployed in a representative body of water and a custom designed ROV equipped with a high bandwidth Commercial-off-the-Shelf (COTS) SONAR system was used to detect and localize itself in relation to the tags. Tag detection algorithms were implemented on composited and filtered sidescan imagery, and localization algorithms were used to estimate vehicle position and error based on detections. A discussion on the full acoustic signal processing pipeline, localization accuracy, system design and limitations, and future steps is presented.