Whole-farm planning is a very useful concept for considering total resource conservation. However, generating such a kind of plan is very complicated because the search space from which a "good" plan can be selected is very large. It even becomes much more complex to plan for a vegetable farm than for a crop farm, for the former, a plan has to be dynamic to satisfy the dynamic market demand of many kinds of vegetables. To help farmers produce a good plan that achieves diverse goals while complying with production, environmental and other constraints, a planning tool called CROPS (Crop Rotation Planning System) was developed by N. D. Stone and a multidisciplinary team at Virginia Tech by using artificial intelligence planning and scheduling technology. The CROPS system has been successfully used to generate whole-farm plans for some crop/livestock farms. However, because this system was developed for crop/livestock farms which usually have few crops and need a static plan, it does not have the ability to generate a plan for other specific types of farms, such as a vegetable farm which has many more crops and needs a dynamic plan to adapt dynamic market demand. In order to produce different plans for a various farms, the CROPS system need more flexibility to generate acceptable and practical plans for different farms while keeping specificity for a specific farm. The purpose of the proposed study is to modify CROPS and make it useful in generating flexible plans for different farm types. To achieve this goal the following specific objectives will be done: 1. Develop a knowledge base by hierarchically representing general and specific farm crops, production practices and other information in different levels of hierarchy. 2. Develop a hierarchical planning engine that can generate plans at different levels of specificity. The existing CROPS planning engine will be used to generate farm plans for vegetable production system. 3. Test and evaluate the new system on a small scale vegetable farm Hierarchical representation of knowledge is arrangement of knowledge in hierarchies whose higher levels hold more general information such as crops, and whose lower levers hold more specific information such as corn and wheat. All attributes and behaviors defined for objects in the higher levels can be inherited by the objects in the lower level of the hierarchy. The hierarchical planning engine should produce abstract plans which can be specified at various levels. This, together with the hierarchical representation of knowledge, will ensure the planning system has enough flexibility to generate a suitable plan for different type of farms. Constraints satisfaction planning and other existing CROPS planning methods will be tested as a specific planning algorithm for vegetable farm. This new system will be used to generate plan for a small vegetable farm to test and evaluate its acceptance and practicality.