Topic: More Efficient Learning in Traffic Grids Viewed as Complex Adaptive Systems Using Agent Based Modeling
Presenter: Chuck Lane
Equation based modeling (EBM) is the most common form of scientific modeling. However, the creation of an appropriate EBM for a large system is likely to be very complicated and computationally expensive. In contrast, consider the result if we treat even a large system as a complex adaptive system (CAS), apply the basic tenets of a CAS, and model it using the concepts of agent based modeling (ABM). ABM requires only the definition of the model environment, the identification of key agents, and a minimum number of key behaviors of those agents. It is my contention that a CAS/ABM model when compared to a corresponding EBM model would be easier to design, implement, execute, and extend and that the results would still be predictive and useful.
My proposal is to create an ABM of the Uptown Charlotte, North Carolina traffic grid. Using traffic volumes based on actual Charlotte traffic counts, I will attempt to demonstrate that the grid with autonomous traffic signals will operate as a CAS. Finally, I will attempt to show that my CAS/ABM model will be more efficient to design and operate than a comparable EBM.