An Adaptive Planner for Real-Time Uncertain Environments
Abstract
The accomplishments under this contract were: (1) the researchers built an adaptive planning architecture for a complex, real-time task environment and a testbed for its principled analysis, (2) developed a model- based methodological approach and used it to analyze numerous aspects of the Phoenix agent architecture, (3) development of a procedure called failure recovery analysis (FRA), for analyzing execution traces of failure recovery to discover when and how the planner's actions may be causing failures, (4) extending the previous work with envelopes with the development of a simple one- parameter decision rule called a slack time envelope, (5) taking several steps toward a formalizing of the problem of plan execution monitoring, (6) building causal models of AI program behavior using path analysis and (7) expanding the scope of the methodological approach and authoring a textbook on empirical methods for Artificial Intelligence.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 29, 1992
- Accession Number
- ADA268147
Entities
People
- Paul Cohen
Organizations
- University of Massachusetts Amherst