Project "Probabilistic Planning with Realistic Preference Models"
(scroll down for publications)

Probabilistic search methods typically either maximize the probability or minimize the expected cost of achieving a given goal. However, human decision makers often have more complex preference models, and planning systems need to adopt these objective functions to be helpful for their users. For example, human decision makers are often risk-averse in high-stake one-shot decision situations, such as planning for crisis situations (for example, oil spills) or space applications.

While AI researchers had concentrated on how to search efficiently for more realistic (and thus larger) world models, my collaborators and I concentrated on how to exploit the structure of probabilistic planning problems to search efficiently for preference models that allow one to model the preferences of human decision makers better than the traditional ones. For example, we developed fast dynamic programming methods that maximize the expected utility for Markov decision problems with non-linear utility functions, such as an exact backward induction method for one-switch utility functions and several variants of approximate functional value iteration for arbitrary utility functions. These dynamic programming methods manipulate concisely represented functions from wealth levels to values. Overall, we developed probabilistic search methods for sensor planning, planning with risk-attitudes for high-stake one-shot decision situations, and planning for deadlines and other real-valued scarce resources, such as energy. I received an NSF CAREER award on this topic.

Currently, we are working on speeding up deterministic bi- and multi-objective search since many applications need to tradeoff between two (or more) competing resources, such as travel time and energy. Examples include solving transportation problems, planning power-transmission lines, scheduling satellites and routing packets in computer networks.

Representative Overview Publications

Representative Publications on High-Stake Planning

Representative Publications on Multi-Attribute Planning

Representative Publications on Hard or Soft Deadlines


Some of this material is based upon work supported by the National Science Foundation under Grant No. 9984827 and an IBM Faculty Fellowship. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or the other sponsoring organizations.

Many publishers do not want authors to make their papers available electronically after the papers have been published. Please use the electronic versions provided here only if hardcopies are not yet available. If you have comments on any of these papers, please send me an email! Also, please send me your papers if we have common interests.

This page was automatically created by a bibliography maintenance system that was developed as part of an undergraduate research project, advised by Sven Koenig.

Home Page of Sven Koenig