Modeling and analysis of human-machine mixed traffic flow considering the influence of the trust level toward autonomous vehicles

Abstract

The mixed driving of human-driven vehicles (HVs) and autonomous vehicles (AVs) is an inevitable stage of future traffic developments. As HV drivers have different trust levels toward AVs, interactions between these two vehicle types will lead to discrepant characteristics in HV driving behaviors, which will impact the traffic flow state of expressways. However, few studies have considered this. Based on questionnaire data, this paper analyzes the changing features of the trust level and its influence on driving behaviors. The quantitative model for the trust level is constructed using the fuzzy logic approach. On this basis, classical cellular automaton models are improved to reflect the features of human-machine mixed traffic flow. Finally, correlations between the trust level and the influencing factors are analyzed, and the impact of the trust level on traffic flow operations is described in terms of both efficiency and safety. The questionnaire results reveal that the influence of the trust level on driving behaviors is universal. The trust level varies with personal and contextual attributes. The simulation results show that traffic density and road congestion greatly influence the trust level. However, the trust level is not sensitive to changes in the penetration rate of AVs. Interactions between these two vehicle types are stronger when the penetration rate approaches 50%. The efficiency and safety of traffic flow operations under each condition decrease by different magnitudes from the influence of the trust level.