Limitations of current traffic models and strategies to address them

Abstract

Traffic flow modeling has evolved over a few decades with numerous models at the macroscopic and microscopic levels being proposed. It appears that traffic flow modeling has been adequately addressed and it is natural to ask if there is still any gap to fill or any room to improve. After examining existing models and comparing them against ideal outcomes, we identified four limitations in both traffic flow theory and simulation, namely (1) the lack of model consistency among macroscopic models and between microscopic and macroscopic models, (2) the lack of model flexibility to admit driver heterogeneity, (3) the lack of model capability to look ahead into the near future, and (4) the lack of model expandability beyond one dimensional traffic. Strategies to address these limitations are proposed with emphasis on general approaches rather than proposing specific models.