Distributed evolutionary estimation of dynamic traffic origin/destination

This paper focuses on updating time varying demand matrices using real-time information. An Artificial Intelligence technique based on Distributed Evolutionary Algorithms (DEA), which is capable to exploit the use of grid computing, is developed. This EA-based demand estimation framework is implemented into a model that we call DynODE (Dynyamic O/D Estimator). DynODE provides a direct way of fusing information of varying types, with different levels of accuracy and from different sensors/sources. DynODE is integrated with an existing Dynamic Traffic Assignment platform (i.e. Dynasmart-P) and is evaluated on a medium size network for various search space sizes and for different quality of the apriori matrix. The obtained results, in terms of replicating observed vehicle counts and the closeness to the real demand, are promising and point to the robustness of the gradient-free framework and its high performance irrespective of the quality of the apriori travel information. The use of Distributed EA is also shown to provide good results within fast computing speeds.