Many desirable application domains for planning are inherently dynamic: representation of the domain is ill-specified or incomplete, the existence or verity of information changes with time, and situations arise during plan execution that were not taken into account at plan generation time. Unfortunately, the traditional approach to planning and problem solving has concentrated most heavily on generating plans, with little attention given to problems associated with plan execution. As a result, these planners have not been generally successful when applied to many physical domains.