Probabilistic knowledge of external events in planning
Jim Blythe · National Conference on Artificial Intelligence · 1994
My research tries to improve the robustness of plans by using limited knowledge about external events. These are events that are not directly caused by the planning agent. I use a discrete-time model and assume that the probability of occurrence for a particular type of event in a given situation is known, but the specific occurrence of such an event cannot be predicted with certainty. For example, when a bicycle is left outside a building, there is some probability p that it will be stolen at each time point. The probability that the bicycle is still outside the building after n time units is then (1 p)“, neglecting the effects of other possible events.