Shark Tracking with AUVs

LAIR is collaborating with biologists from CSU Long Beach to develop a multi-AUV robot system that is capable of tracking tagged sharks for long periods. The system consists of both hardware implementations, as well as novel signal processing and path planning algorithms.

 

 
 Late night hardware debugging on the dock at USC Wrigley Marine Science Center

Late night hardware debugging on the dock at USC Wrigley Marine Science Center

 

The autonomous underwater vehicles (AUVs) currently used in this project are Iver 2 manufactured by L3 OceanServer. The AUVs are equipped with stereo hydrophone systems that detect acoustic signals transmitted by acoustic tags. In addition, the team has integrated acoustic modems with the AUVs, so they are in constant communication with each other, sharing their position and sensor measurements.

 
 
 OceanServer Iver 2 equipped with two hydrophones to listen for shark tag signals.

OceanServer Iver 2 equipped with two hydrophones to listen for shark tag signals.

 

Using acoustic measurements obtained from the hydrophones, a particle filter determines the position, velocity, and orientation of the tagged shark. The estimated location of the shark is used by the AUV controller to circle around the shark so as to maintain a "stand off" distance to prevent affecting the shark's behaviour.

 A schematic demonstrating uncorrected TDOA data as a result of clock drifts

A schematic demonstrating uncorrected TDOA data as a result of clock drifts

The team has also worked on ways to estimate tagged sharks’ position using unsynchronized hydrophone systems, for situations where building a custom ADC/data recorder plus hydrophones are unfeasible.
The team has developed algorithms that compensate non-linear clock drifts caused by rapid temperature changes, as well as using mixed-integer linear program to estimate TDOA between hydrophones.

 
tracks_unsynchronized_hydrophone.png
 
 Results of shark-tracking trials

Results of shark-tracking trials

These innovations have produced many hours of autonomous tracking and following of tagged sharks using the prototype cooperative multi-AUV system.