My thesis topic is plan generation in temporal, parallel, probabilistic domains with oversubscribed goals. I have defined a framework that includes two novel extensions. First, the plans can include parallel steps that serve the same goal and increase the probability of success in addition to parallel steps that serve different goals and decrease execution time. Second, already executing plan steps can be terminated if doing so saves resources to achieve more goals. My algorithm called CPOAO* (Concurrent, Probabilistic, Oversubscription AO*) can deal with these extensions. In this paper, I summarize the design and implementation of CPOAO* and its associated heuristics, and propose a plan of research.
Subjects: 15. Problem Solving; 15.7 Search
Submitted: Apr 8, 2008