Contingency Analysis in the Design to Criteria Scheduler

The Design to Criteria scheduler is a domain independent system that schedules complex AI problem solving tasks to meet real time performance goals In this paper we further extend the scheduler to more e ectively deal with uncertainty present in a schedule which can be critical in hard deadline or hard cost situations This is based on an analysis of available schedules that can be used to recover from a situation in which partially executed schedules cannot be completed successfully In addition to evaluating schedules e ectively from the uncertainty perspective we also implement method reordering techniques to minimize uncertainty