Bidirectional Control Characteristics of General Motors and Optimal Velocity Car-Following Models Implications for Coordinated Driving in a Connected Vehicle Environment

In natural traffic flow, the information from preceding vehicles predominantly determines driver behavior. With connected vehicle technologies, drivers can receive information from both preceding and following vehicles. This information creates new opportunities for vehicle coordination and control at the microscopic level on the basis of bidirectional information. Although bidirectional car-following models have been studied since the 1960s, most existing car-following models, especially those used by adaptive cruise control technologies, are still forward-only car-following models. This paper serves as a first step toward the use of bidirectional car-following models for microscopic vehicle coordination and control. The focus is on the study of the models' general control characteristics and impact on traffic flow stability. A general bidirectional control framework is proposed to convert any car-following model into its bidirectional form. Four representative General Motors and optimal velocity car-following models are reformulated and calibrated against field vehicle trajectory data collected in the next-generation simulation program (NGSIM). The bidirectional control characteristics of the selected models were evaluated by tuning of the percentage of backward information considered in the final car-following decision. The evaluation uses forward versus backward acceleration diagrams and a ring road stability analysis of equilibrium states obtained from NGSIM data. The results indicate that the increase in the contribution of backward information may help alleviate traffic congestion and stabilize traffic flow. An operating range of the backward information contribution of between 5% and 20% is recommended to ensure that the resulting models are still physical and realistic for both free-flow and congestion situations.