Wyoming ranchers have a new source of expert advice in responding to grasshopper infestations, thanks to the Case-based Range Management Advisor (CARMA) built with Allegro CL for Windows.
"This is an application designed to model the expertise of professional pest managers, like USDA extension agents, who advise ranchers about how to deal with grasshopper infestations," says Karl Branting, Ph.D., Associate Professor at the University of Wyoming. "In particular, it models the expertise of University of Wyoming entomologist Jeff Lockwood and other entomologists like him."
According to Lockwood, grasshopper infestations on average consume 21 to 23 percent of available grassland forage each year. Often, the cost of pesticides is greater than the value of the forage consumed, and pesticides may be detrimental to beneficial insects and cause pollution. "Therefore, it is important for ranchers to have the best possible advice about the most economically and environmentally sound options for responding to grasshopper infestations," says Lockwood.
Branting and Lockwood settled on a case-based approach to developing CARMA after a careful study of entomological problem solving. "We found that experts give advice by comparing each new case to prototypical past cases. The recommendations associated with the most similar prototypical case are then adapted to fit the unique facts of the new case," says Branting. Based on this observation, the development team collected a library of typical Wyoming infestation cases from seven experienced entomologists. "CARMA contains cases from experts with 140 years of combined experience," says Branting. "CARMA also uses numerical models of grassland population dynamics, rules for inferring the species and density of grasshoppers and for selecting appropriate biological or chemical control agents, and a Markov-chain model of reinfestation probability."
Branting and Lockwood chose Allegro CL's Common Lisp for Windows as their development tool for several reasons. "Our customers are county extension agents, weed and pest district offices, and ranchers," says Branting. "These users generally have PCs. We needed to build a PC-based application that had an intuitive user interface that could display bitmaps of various species and developmental stages of grasshoppers and GIS [Geographical Information System] maps showing information such as the infestation history across the state." Even more important that the need for displaying bitmaps and GIS information through an intuitive user-interface, Branting and Lockwood also needed symbolic reasoning ability.
Branting explains: "CARMA involves several different AI techniques, including case-based, rule-based, and model-based reasoning. To implement all these techniques in C would have added considerably to the development time. And expert system shells are just too rigid to model the kind of flexibility that humans exhibit. In Allegro CL, we could write our own inference engines and our own control strategy to guide their interaction."
CARMA was funded by public grants and is deployed free of charge in every county throughout Wyoming.
For more information about CARMA, you may visit the University of Wyoming's web site at http://meru.cs.uwyo.edu/~hastings/carma.html.
|Copyright © 2013 Franz Inc., All Rights Reserved | Privacy Statement||