Scribd
Upload
31 views

Diver Detection System Requirements

The document reports on progress towards developing algorithms to detect and track divers in 2D imaging sonar data for underwater human-robot interaction scenarios. It outlines assumptions for the diver detection system, including that the diver and sonar platform can be stationary or mobile, and the diver's shape is deformable. It proposes detecting divers based on their trajectories or single-frame feature detections. The report also details initial plans to implement a template-based frame-level detector and collaborate with other researchers to collect additional sonar data of divers.

Uploaded by

SyllogismRXS
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
31 views

Diver Detection System Requirements

The document reports on progress towards developing algorithms to detect and track divers in 2D imaging sonar data for underwater human-robot interaction scenarios. It outlines assumptions for the diver detection system, including that the diver and sonar platform can be stationary or mobile, and the diver's shape is deformable. It proposes detecting divers based on their trajectories or single-frame feature detections. The report also details initial plans to implement a template-based frame-level detector and collaborate with other researchers to collect additional sonar data of divers.

Uploaded by

SyllogismRXS
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Detection, Tracking, and Trailing of a Diver for Underwater Human-Robot Interaction Scenarios

Progress Report
February 23, 2015
Kevin DeMarco

Diver Detection System Requirements


The primary purpose of the future work is to develop algorithms to detect and track a diver in 2D imaging
sonar data. Since underwater optical cameras are unreliable at long-range, the optical camera data will only
be used to verify the existence of the diver at close-range.
The following is a list of assumptions for the diver detection system:
The diver can be either stationary or mobile.
The sonars platform (robot) can be either stationary or mobile.
The frame-level diver detector needs to be scale-invariant and rotation-invariant.
The divers shape is deformable.
The divers features (legs, arms, head, etc.) can be occluded from frame-to-frame.
When a human observes 2D imaging sonar data for the purpose of detecting a diver, the human is most
likely to detect the divers presence in the sonar by the divers trajectory or by single frames in which the
divers features are present. Thus, the diver detection system should take advantage of classifying objects in
the sonar based on their trajectories and by single-frame feature detections with high-confidence values. In
this case, a false negative report would be preferable to a false positive report.
Finally, it may also be interesting to try to relate objects to their acoustic shadows. In some frames, the
acoustic shadow provides a higher resolution feature of a divers arm or leg than the actual object that first
reflected the acoustic wave.
To establish a baseline for the frame-level detector, I will implement a brute-force template-based approach. First, between three and five template images will be selected from the sonar data that has already
been collected. Second, a number of scale and rotation transforms will be applied to the templates. Finally,
the sliding-window template algorithm will search for diver objects using the library of scaled and rotated
template images.

Other News
Dr. Noel du Toit at the Naval Postgraduate School (NPS) said that he will grant me access to any diverrelated sonar data that he collects since he does not have time to work on the diver detection problem. Also,
I am going to figure out the logistics for me to travel to Key Largo in July to participate in Dr. Noel du
Toits underwater HRI experiments with NASA.
Anthony told me that Dr. Schmidt is planning on traveling to Savannah on a regular basis this spring
and that divers will be in the water with their vehicle. I e-mailed Dr. Schmidt and asked her if I can tag
along and passively record data of the divers in the water.

You might also like