Mixed-Initiative Development of Plans With Expressive Temporal Constraints
Abstract
The overall objective of this project was die development of methods for efficiently finding solutions to problems that consist of sets of expressive constraints including both overconstrained and underconstrained situations. In either case, constraint management is necessary: in an overconstrained situation, it is necessary to determine how best to relax existing constraints so that a solution can be computed, while in an underconstrained situation, it is necessary to select good (or, if time permits. optimal) solutions from amongst the alternatives. We developed novel and highly efficient solutions using two classes of techniques: fully automatic techniques that manage constraint sets given explicit preference functions, and mixed initiative techniques that allow interactive control by a human user to manage constraint sets, and we evaluated these on both synthetic and real-world data sets.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 14, 2007
- Accession Number
- ADA474565
Entities
People
- Martha E. Pollack
Organizations
- University of Michigan