Evidential ReasoningforGeographicEvaluation for Helicopter RoutePlanning

In order toplanoperations whereknowledge ofsignificant elements isimprecise anduncertain, ameansofcharacterizing thesit- uation intermsofthevarious factors thatmayinfluence those opera- tions mustbeprovided. Inthis paperwediscuss anapproach tothat characterization thatusesevidential reasoning tohandle theuncer- tainty, imprecision, andincompleteness typical ofsources ofreal-world information andknowledge, tosupport planning routesformilitary helicopters. Evidential reasoning isamaturing collection ofinference techniques forreasoning withuncertain information. Basedon theShafer- Dempster theory ofevidence, evidential reasoning usesanon-Bayesian updating schemetocombineevidence provided bymultiple diverse knowledge sources. Knowledge sources inanevidential reasoning sys- temarenotrequired toattribute their belief toauniverse ofdiscourse comprised solely ofmutually exclusive, exhaustive, singleton events, as required byaclassical probability approach. Rather, they mayexpress levels ofignorance explicitly byallocating belief todisjunctions of propositions, thereby leading directly toaninterval measureofbelief; ignorance isexpressed bythewidthofthis interval. Evidential reasoning evolved fromconsideration ofappropriate models forreasoning aboutinformation acquired fromsensors, and therefore seemsnatural fordrawing conclusions fromsensordataand prestored mapsregarding thedegree towhichaselected geographic areawill support certain activities. Here, wediscuss evidential reason- ingandillustrate theutility ofthetechnology forclassifying geo- graphic areasbydescribing ourcurrentmap-and-sensor-based re- search inwhichweestimate theutility oflandareasforconcealing helicopter operations.