Temporal Planning with Preferences and Probabilities

Morris, R., Morris, P., Khatib, L., and Yorke-Smith, N. Temporal Planning with Preferences and Probabilities. Proceedings of ICAPS'05 Workshop on Constraint Programming for Planning and Scheduling, Monterey, CA, June 2005.

Abstract: In an uncertain world, a rational planning agent must simultaneously reason with uncertainty about expected outcomes of actions and preferences for those outcomes. This work focuses on systematically exploring the interactions between preferences for the durations of events, and uncertainty, expressed as probability distributions about when certain events will occur. We expand previous work by introducing a means for representing events and durations that are not under the control of the planner, as well as quantitative beliefs about when those events are likely to occur. Two reasoning problems are introduced and methods for solving them proposed. First, given a desired overall preference level, compute the likelihood that a plan exists that meets or exceeds the specified degree of preference. Second, given an initial set of beliefs about durations of events, and preferences for times, infer a revised set of preferences that reflect those beliefs.

A slightly improved version of this paper appeared at the IJCAI'05 Multidisciplinary Workshop on Advances in Preference Handling.

PDF (154K) | Bibtex entry

Workshop homepage