Research on heterogeneous traffic flow considering conditionally automated driving vehicles in car-following state

This study develops a traffic-condition-based framework for modeling takeovers in conditionally automated driving vehicles (CADVs) and evaluating heterogeneous traffic stability and theoretical capacity. Rather than prescribing fixed takeover speed limits, a speed-density suitability function incorporating traffic speed, density, and vehicle composition identifies an applicable automated driving speed range of 5.86–21.34 m/s. A cellular automaton uses these boundaries and a 7 s takeover time to simulate control state transitions and generate trajectories for calibrating an equivalent CADV car-following model. Human-driven vehicle (HDV) parameters are calibrated using Waymo trajectories, while fully automated driving vehicles (FADVs) use cooperative adaptive cruise control (CACC). Results show that increasing total automated vehicle penetration improves stability and theoretical capacity, even when all automated vehicles are CADVs. At 60% penetration, replacing all CADVs with FADVs increases the stability recovery rate from 7.80% to 31.40% and theoretical capacity from 2434.10 to 2963.00 veh/h.