Topaloglu and Powell (2005), Topaloglu and Powell (2006),
Topaloglu and Powell (2007) are some of the recent examples of
fleet management problems modeled as Approximate Dynamic Programming
(ADP). Fig. 6 gives an illustration of the essence of ADP
problems. These studies offer a nice framework for modeling and
solving a variety of real world problems, especially those including
time windows or labor restrictions, by addressing them at the level
of the local subproblem. Topaloglu and Powell (2006) extend previous
works on ADP models for fleet management to include heterogeneous
resources and substitution among them. Their method
uses a hybrid of linear and piecewise-linear approximations of
the value function. Topaloglu and Powell (2007) develop sensitivity
analysis methods for a stochastic dynamic fleet management
model to compute the change in the objective value in response
to changes in fleet size and load availability. Topaloglu and Powell
(2005) extend the dynamic resource allocation problem to a distributed
decision-making case. They use non-linear functional
approximation to model the coordination of actions of different
agents.