Moths inspire explosives-hunting alorithim
Scent-tracking behavior provides a model for uncertain robot sniffers; Infotaxis algorithim developed by U.S. and U.K. researchers helps the robot develop and react to a “probability map”; exploration and exploitation tendencies are coordinated
Dogs and robots (and, as we now know, humans) are good at tracking by scent, but in all cases the effort is hampered by the tendency of the track to fade over time and distance. This presents the tracker with a quandry: to continue on in the same direction after the track evaporates, or to strike out adventurously in a different direction. Fortunately, a new algorithim call Infotaxis, developed by scientists at the Pasteur Institute, the University of Marseilles, and UC Santa Barbara, provides a solution. Infotaxis, the scientists explain, operates like the flight of moths by balancing the two tendancies in order to maximize information gain.
Why moths? It seems that female moths discharge a certain scent during breeding times, but that these are received as occasional plumes by male moths, who dart wildly back and forth attempting to locate their potential mates in a behavior known as chemotaxis. Infotaxis attempts the same thing, but it is intended for use in software in sniffer robots, which by definition know nothing about the source of potential explosives and consider any direction as posessing a fifty-fifty likelihood of success. By employing various exploratory strategies such as moving in a circular motion, it updates its own “probability map” as it accumulated more information in suspicious piffs of air. “You singled-mindedly go to the place where you think you’ll gain more information,” explained professor Boris Shraiman. “If you make a step and you find the source, boy, you’ve really localized it! That’s a very informative step.”
-read more in Nikhil Swaminathan’s Scientific American report
