Risk-Quantification Based Longitudinal Planning Model for Connected and Autonomous Vehicles

Current car-following models, mostly developed under the premise of simulating human driver behavior, struggle to fully meet the requirements for efficiency, safety, and stability of Connected and Autonomous Vehicles (CAVs). Therefore, this study introduces a risk-quantification based longitudinal planning model (RQM) hinged upon a novel omnidirectional risk indicator. Depending on the preceding vehicle's state, the RQM exhibits distinct driving modes, with different parameters influencing its behavior and macroscopic characteristics. Besides, based on the linearized approximation, theoretical analysis confirms that the RQM can ensure overdamped properties and local string stability while maintaining comparable or even higher traffic capacity than existing models. Furthermore, various numerical experiments are designed to test the properties of the RQM and validate the theoretical analysis. Braking tests demonstrate that RQM consistently enables safe emergency braking under diverse velocity conditions while maintaining both overdamped behavior and string stability. Comparative experiments with the similarly traffic-capable Intelligent Driver Model (IDM) and Adaptive Cruise Control (ACC) further prove that vehicle platoon controlled by RQM can swiftly and efficiently dissipate oscillations originating downstream in both time and spatial dimensions. RQM proposed in this study offers a significant enhancement in stability and capacity over traditional car-following models while ensuring safety. It provides a highly stable model-driven longitudinal planning method for CAVs while preserving their traffic efficiency.