ROV-based tracking of a shallow water nocturnal squid

This paper describes the use of a Remotely Operated Vehicle (ROV) equipped with a monocular vision system to find and track the squid Euprymna scolopes, so that motion behaviors of the squid could be characterized through the use of off-line image processing and state estimation. The ROV was deployed for several nights at several nearshore locations off Oahu, resulting in 10 hours of squid footage. Using blob-tracking image processing techniques and a Particle Filter state estimator, the squid can be detected and tracked. The position, velocity, and acceleration of the squid relative to the stationary ROV can be determined. Experiment results from tracking a simulated squid at known positions in a swimming pool and tracking of live squid in the ocean validate the performance of the tracking system. Results show the 3-D trajectory of the squid in a test feeding video. To the best of the authors' knowledge, this is the first observation and tracking of this species of squid in its natural environment.