Planning based on Time-Oriented Skeletal Plans: Methods and Tools for Design, Visualization, and Basic Execution of Time-Orientd Skeletal Plans.

The main goal of this research proposal is to develop a planner based on time-oriented skeletal plans. Skeletal plans are plan schemata at various levels of detail that capture the essence of procedures, but leave room for execution- time flexibility in the achievement of particular goals. They are a powerful way to reuse existing domain-specific procedural knowledge in different contexts. We are focusing on domains that are time-oriented with respect to both external states and plan actions, and that might require intermittent execution and adaptation of plans over multiple (possibly disjoint) periods of time, and ill-covered requirement in classical planning models. We will develop task-specific problem-solving methods and toois to design, to visualize, and to execute skeletal plans. During this project the following subgoals will be achieved: developing higher-level temporal abstraction methods needed in the representation of skeletal plans, creating an appropriate temporal representation of skeletal plans, acquiring time-oriented knowledge, developing cooperative agents performing time-based concurrent plan observation (verification and validation) and basic execution, and visualizing both the different components of the skeletal plans and the reasoning processes based on the skeletal plans in task-specific way. Our application domain will be medicine. Especially, we are motivated by the demands and the needs for knowledge-based support in modern intensive care units (ICUs). Medical scenarios will serve as resource for interesting research areas and for identifying the demands and needs for user-oriented problem-solving methods. We will proof the validity and usefulness of the proposed approach within real-world environments of ICUs. The starting points of the proposed project are the results of our previous work in the problem area, in particular the outcomes of the VIE-VENT project and the Asgaard project. The proposed project will result in a planner based on time-oriented skeletal plans, which support both designers and executors of time-oriented skeletal plans.