Planning in Complex Worlds Via Mixed-Initiative Interaction.
Final rept. Feb 91-Jan 96,
ROCHESTER UNIV NY DEPT OF COMPUTER SCIENCE
Pagination or Media Count:
This final report presents an overview of the research performed at the University of Rochester under Rome Lab contract F30602-91-C-0010. The project addressed problems in developing large-scale plans in complex worlds. The results can be divided into four general areas. First, we addressed some fundamental representational issues by developing new representations of actions and plans that increase the expressiveness of plan representations. Second, we addressed efficiency issues by developing a set of temporal reasoning algorithms for the efficient handling of very large-scale temporal databases. Third, we developed a new set of heuristic search methods that greatly improve the efficiency of well-founded planning systems such as UCPOP. Several of these techniques have been incorporated into the latest version of UCPOP distributed by the University of Washington. Fourth and finally, we addressed the problem of developing plans in the real world by constructing and testing a new model of mixed-initiative planning. The model of mixed-initiative planning may be the result that has the most lasting impact on the field. By viewing the human as an essential part of the planning process, we dramatically change the problems that are important for the ultimate successful application of planning technology.
- Operations Research
- Logistics, Military Facilities and Supplies