OBJECT-ORIENTED CONSTRAINT SATISFACTION PLANNING FOR WHOLE-FARM MANAGEMENT

Over the past three years, a whole-farm planning system (CROPS) has be en developed and tested on Virginia farms. The CROPS system solves a c ritical scheduling problem for farmers who want to implement environme ntally sound agricultural practices based on crop rotation. The system 's inputs include the farmer's personal goals, farm production goals, economic requirements, and a farm inventory consisting mainly of infor mation about the land: field boundaries, slopes, topology, and soil ty pes. The system then uses heuristics and constraint-satisfaction metho ds to find a multi-year farm plan that meets the farmer's needs within environmental constraints. Recent work has focused on making the cons traint satisfaction scheduling algorithm more efficient and more gener alized. The current implementation is object-oriented and employs part ial arc-consistency algorithms, variable ordering, and constraint rela xation. This paper describes the constraint-based scheduler (CBS), its representation, and how it handles constraint relaxation. It also dis cusses the potential application of the CBS to agricultural and natura l resource management in general.