Our research is motivated by the need for decision support systemsin the areas of logistics, transportation, and distribution.In our models we take into account the dynamic and stochasticnature of the environment in which logistics operations typicallytake place. We started by studying the problem of optimally accepting loads for LTL transportation, and of dispatching vehicles. The problem involves loads with random sizes and rewards, arriving stochastically in time, and with a waiting cost for holding unserviced accepted loads, with and without deadlines for dispatching vehicles. This problem has many other applications, and is called the Dynamic and Stochastic Knapsack Problem.
Next we considered an extension of this problem, where vehicles as well as loads arrive stochastically in time. This is typical of the situation faced in a network of distribution terminals, where loads arrive from customers and vehicles arrive from other terminals, to be dispatched in time to other terminals in the network.
We are currently working on approximation methods to find good policies for managing freight distribution operations over large networks of end-of-line and breakbulk terminals.