A Stochastic Formulation of the Dynamic Assignment Problem, with an Application to Truckload Motor Carriers

Abstract

The dynamic assignment problem arises in a number of application areas in transportation and logistics. Taxi drivers have to be assigned to pick up passengers, police have to be assigned to emergencies, and truck drivers have to pick up and carry loads of freight. All of these problems are characterized by demands that arrive continuously and randomly throughout the day, and require a dispatcher to assign a driver to handle a specific demand. We use as our motivating application the load matching problem that arises in long-haul truckload trucking, where we have to assign drivers to loads on a real-time basis. A hybrid model is presented that handles the detailed assignment of drivers to loads, as well as handling forecasts of future loads. Numerical experiments demonstrate that our stochastic, dynamic model outperforms standard myopic models that are widely used in practice.