AAAI Publications, Twenty-Sixth AAAI Conference on Artificial Intelligence

Font Size: 
Generalizing and Executing Plans
Christian James Muise

Last modified: 2012-07-14

Abstract


In a dynamic environment, an intelligent agent must consider unexpected changes to the world and plan for them. We aim to address this key issue by building more robust artificial agents through the generalization of plan representations. Our research focuses on the process of generalizing various plan forms and the development of a compact representation which embodies a generalized plan as a policy. Our techniques allow an agent to execute efficiently in an online setting. We have, to date, demonstrated our approach for sequential and partial order plans and are pursuing similar techniques for representations such as Hierarchical Task Networks and GOLOG programs

Full Text: PDF