Freeway capacity modeling and analysis for traffic mixed with human-driven and connected automated vehicles considering driver behavior characteristics

Abstract

Future freeway traffic flow will include both Human-driven Vehicles (HV) and various sizes of Connected Automated Vehicles (CAV) platoons. When following CAV platoons, HV drivers may change the car-following distance for psychological reasons, which could further affect the road capacity under mixed traffic flow. In this paper, we constructed a capacity model that considered the spatial distribution characteristics of mixed traffic flow and the behavioral traits of HV drivers when they follow CAV platoons. We surveyed 331 HV drivers using a questionnaire. 51.1% of drivers opted to increase the car-following distance when following a CAV platoon, and the amount of change in distance increased in direct proportion to the size of the platoon. A Structural Equation Model (SEM) based on survey data was developed to examine the connection between psychological variables and HV drivers’ driving behaviors. These results of the questionnaire were introduced into the capacity model construction, and the trends of capacity with CAV penetration, maximum platoon size, platoon intensity, and time headways were revealed. The results show that previous studies overestimated the capacity under mixed traffic flow, especially when the maximum platoon size is equal to or greater than 6 or the CAV penetration is between [0.6, 0.8], the overestimation of the capacity is higher. CAV penetration, maximum platoon size, and platoon intensity are positively correlated with capacity, while time headway is negatively correlated with capacity. The capacity growth is not significant when CAV penetration is between (0, 0.4]. Six is the ideal size of CAV platoons under mixed traffic flow when considering the impact of CAV platoons on HV drivers’ driving behaviors. By taking measures to promote CAV aggregation and making platoon intensity greater than 0, the capacity will be improved more significantly. For a specific car-following type, a different value of time headway will lead to a different capacity, but the impact trends will be similar.