Evaluation of Professional Taxi Drivers’ Car-Following Behaviour on Capacity of Urban Highway Segments

In the field of traffic operation analysis, one of the most important topics is that of estimating highway capacity, which determines the reliability of capacity-related traffic optimization strategy. As proposed by the Highway Capacity Manual, various factors are considered to improve the accuracy of capacity estimation under prevailing conditions. In particular, the driver population factor, which accounts for the heterogeneity of driving behaviour, is determined by “arbitrarily” judging whether highway facilities serve drivers travelling for commuting (or otherwise-accustomed drivers). However, since a significant proportion of professional taxi drivers plays important role in urban transportation system, major difference between driver populations can origin from the heterogeneity of driving experience associated with drivers’ professionalism. In this paper, an adjustment factor is developed, aiming to precisely evaluate the impacts of professional taxi drivers’ car-following characteristics on capacity of urban highway segments. Based on the car-following data measured by our vehicle-mounted system, it is found that professional taxi drivers have statistically significant difference in spacing as compared to non-professional drivers. Then regression models are developed to relate speeds to spacing by means of the equilibrium relation in the Intelligent Driver Model, which achieve high performance of explaining car-following data as well as significant interpretability on a macroscopic traffic flow level. Being compatible with the paradigm of capacity estimation in the HCM, a quantitative adjustment factor was proposed for capacity determination by identifying the impacts of professional taxi drivers’ car-following characteristics. It is found that the presence of professional drivers can lead to at most 31% capacity increment. Moreover, this study compared the estimated and the measured capacities. The results suggested that the estimation accuracy of capacity is significantly improved by taking into account the proposed adjustment factor.