Geographical route planning based on uncertain knowledge

Reports on the development of a geographical route planning system using uncertainty reasoning. This system searches for the route from a start point to a goal point indicated on a map. This system uses the digitized terrain map available from the Japanese Geographical Survey Institute. The target area of the route planning is subdivided into fine meshes of 100 meters square. This system is composed of two reasoning modules: a mobility cost evaluation module and a route planning module. The former module evaluates mobility costs for every 300/spl times/300 meshes included in the designated area. Domain experts often use ambiguous data interpretation knowledge for evaluating terrain circumstances and deciding the mobility cost in a mesh. We introduce two uncertainty reasoning mechanisms to represent such a data interpretation process: one is fuzzy reasoning, and the other is Dempster-Shafer theory. The route planning module uses the F* optimization algorithm. The geographical route planning system also offers knowledge editing facilities for describing the mobility cost evaluation knowledge, such as a dataflow diagram editor for designing the data integration process and a membership function editor for designing data abstraction methods. These knowledge editors facilitate the development and modification of a mobility cost evaluation knowledge base.