Researchers, led by Mark Patterson, associate professor of marine science, at the Virginia Institute of Marine Science at the College of William and Mary have developed an artificial neural network for use with an autonomous underwater vehicle (AUV) named Fetch.
Characteristics of different fish species were compiled using the side scan sonar data. This information was then grouped into test sets used for training artificial neural networks (ANNs). The team combined the use of enhancement algorithms and image processing, in conjunction with the ANNs, to "teach" the computer to recognize the characteristics of various fish species.
As reported in the latest edition of New Scientist (vol. 177 issue 2382 - 15 February 2003, page 16) the training was successful; scientists were able to have Fetch2 recognize two marine fish species - jacks and sharks. Fishes that weren't either species couldn't fool the classifier. Patterson noted, "It's amazing how well this particular type of neural network works with noisy data. In the future, we hope to expand the classifier's library to include dozens of species."
"We have only scratched the surface of this technology," said Zia-ur Rahman, research associate professor of Applied Science at the College of William and Mary. "The computer could be trained to recognize anything -- a person swimming, a submarine, a missile or a mine, anything." "This technology has countless applications," he added. Ultimately, the scientists hope to have Fetch2 autonomously follow the objects it detects.
Once trained to recognize these underwater dangers Fetch2 could be used to patrol coastlines, harbors and moored naval assets like ships and subs becoming an important tool in the war on terror and the battle to keep our shores safe.
The above post is reprinted from materials provided by College Of William And Mary. Note: Materials may be edited for content and length.
Cite This Page: