Mixed traffic capacity estimation of autonomous vehicles impact based on empirical data

This study refined the composite headway distribution model to estimate the mixed traffic capacity by utilizing extensive empirical mixed traffic data. The enhanced model integrated an automatic iterative algorithm into the stochastic process to capture the headway distributions of autonomous and human-driven vehicles in five types of car-following pair combinations. The proposed iterative algorithm automatically fitted the free-flow headway distribution and dynamically determined the threshold for distinguishing the car-following and free-flow states of autonomous and human-driven vehicles based on extensive trajectory data. Quantitative analysis was carried out to assess the impact of time period and vehicle type on the headway distributions of following vehicles in the five car-following pairs. The mixed traffic capacity was estimated using the overall composite headway distribution model, considering the effects of penetration rate, platoon intensity, and maximum platoon size of the autonomous vehicle. The findings indicated that autonomous vehicles showed more cautious and sensitive car-following behaviors compared to human-driven vehicles. Compared to the daytime, autonomous vehicles showed more cautious car-following behaviors at night. When following heavy vehicles, autonomous vehicles typically adapt conservative and stable car-following behaviors. In addition, the results also showed that the penetration rate, platoon intensity, and maximum platoon size significantly affected the capacity of mixed traffic flow. These results of this study can assist traffic management authorities in formulating rules or policies that take into consideration the growing prevalence of autonomous vehicles.