Modeling information in freight transportation

It is common in operations research to focus on modeling the flows of physical assets (trucks, trains, containers, freight).  For many questions, and especially in the world of operational models, it is very important to explicitly model the organization and flow of information and decisions. We discuss the subtle but important notational conventions required to model information, and introduce the four classes of information and demonstrate their use in different applications within freight transportation. We show how approximate dynamic programming techniques can be used to capture uncertain future information, and present techniques for capturing expert knowledge.

Warren Powell is the director of CASTLE Laboratory which specializes in the development and implementation of dynamic optimization models for freight transportation. His research focuses on models and algorithms for large-scale, stochastic dynamic resource allocation problems, and related problems in demand forecasting and pricing.