Computationally Efficient Algorithms for Near-Optimal Decision-Making in the Presence of Uncertainty

Aviation is a very “uncertain” business in which critical decisions have to be made well in advance of execution. The traditional approach to decision-making in the presence of uncertainty is to assume at the time the decision must be made that the available information is certain, determine the optimum (or at least best possible) course of action, and then change your actions (often in a sub-optimal way) if the operating environment is different than expected. I will present a set of computationally efficient algorithms for near-optimal decision-making in the presence of uncertainty. The common feature of all the algorithms is a stochastic branch and bound-based implementation where the possible sequences of events are enumerated on a decision tree and the branches to explore are determined via Monte Carlo simulation-based sampling