Connected and automated vehicles (CAVs) are expected to improve both traffic safety and efficiency by reducing the human driver errors. Recently, many researchers have focused on the simulation-based studies in order to evaluate the benefits of CAVs due to the lack of real-world data. However, none of the previous studies have attempted to differentiate the benefits of CAVs over automated vehicles (AVs) by incorporating multiple preceding vehicle information (i.e., acceleration, position, etc.). This paper aims to fill the existing gap by utilizing separate car-following models for both CAVs and AVs in order to approximate their driving behavior in the Aimsun Next simulation platform. Additionally, a different car-following model is used for the connected vehicles (CVs) without automation by addressing the human driver compliance factor. This study also utilizes mixed penetration of CAV and CV with no automation. A well calibrated and validated simulation testbed is developed for SR417 in Orlando, Florida which is the base scenario in this study. To this end, the impact of CAVs, AVs, and CVs are evaluated based on both traffic efficiency (i.e., travel time) and safety (i.e., traffic conflicts) under various market penetration rates (MPRs). The traffic efficiency results show that travel time is significantly reduced for any MPRs of CAVs, AVs, and the mixture of CVs and CAVs compared to the base scenario. A generalized estimating equation (GEE) model is developed to quantify the travel time improvement for CAVs, AVs, and the mixture of CVs and CAVs. The results suggest at least 20% penetration is required for CAVs to get travel time improvement while 40% penetration is needed for AVs. Also, CAV significantly outperforms AV for the same MPRs. For the safety assessment, traffic conflicts are estimated by using different surrogate measures i.e., time-to-collision (TTC) and time exposed time-to-collision (TET). The results imply that crash risk is significantly reduced for CAVs, AVs, and the mixture of CVs and CAVs scenarios compared to the base condition. A Bayesian zero-inflated negative binomial model is developed in order to model the number of traffic conflicts as a function of MPRs of CAVs, AVs, mixture of CVs and CAVs, and traffic parameters. The results confirm that CAVs are more efficient in reducing crash risk compared to AVs for the same MPR. The mix penetration rate of CV (60%) and CAV (20%) shows almost similar reduction of crash risk with the 80% MPR of AV. Also, crash risk analysis based on different vehicle types shows that CAVs driving behavior is safer compared to the AVs. Finally, the results of this study indicate a significant improvement of both traffic efficiency and safety by implementing CAV with multivehicle communication system on the freeway segments.

联网和自动驾驶汽车（CAV）有望通过减少人为驾驶错误来提高交通安全性和效率。近来，由于缺乏实际数据，许多研究人员专注于基于仿真的研究，以评估CAV的优势。但是，以前的研究都没有尝试通过合并多个先前的车辆信息（即加速度，位置等）来区分CAV相对于自动车辆（AV）的好处。本文旨在通过针对CAV和AV使用单独的汽车跟踪模型来填补现有差距，以便在Aimsun Next仿真平台中近似其驾驶行为。此外，通过解决人类驾驶员合规性因素，无需自动化即可将不同的汽车跟随模型用于联网汽车（CV）。这项研究还利用了CAV和CV的混合渗透，没有自动化。在佛罗里达州奥兰多市为SR417开发了一个经过良好校准和验证的模拟测试台，这是本研究的基本方案。为此，基于各种市场渗透率（MPR）下的交通效率（即旅行时间）和安全性（即交通冲突），评估了CAV，AV和CV的影响。交通效率结果显示，与基本方案相比，CAV，AV的任何MPR以及CV和CAV混合的任何MPR都可大大减少行驶时间。建立了广义估计方程（GEE）模型，以量化CAV，AV以及CV和CAV混合行驶时间的改进。结果表明，CAV至少需要20％的渗透率才能改善旅行时间，而AV则需要40％的渗透率。同样，对于相同的MPR，CAV明显优于AV。为了进行安全评估，可以通过使用不同的替代措施（即碰撞时间（TTC）和暴露时间的碰撞时间（TET））来估计交通冲突。结果表明，与基本条件相比，CAV，AV以及CV和CAV混合场景的崩溃风险显着降低。建立贝叶斯零膨胀负二项式模型，以根据CAV，AV，CV和CAV的混合MPR和交通参数对交通冲突的数量进行建模。结果证实，与相同MPR的AV相比，CAV在降低撞车风险方面更为有效。CV（60％）和CAV（20％）的混合渗透率显示出与MPR的80％MPR几乎相似的碰撞风险降低。此外，基于不同车辆类型的碰撞风险分析表明，相比自动驾驶汽车，CAV的驾驶行为更安全。最后，这项研究的结果表明，通过在高速公路路段上使用多车通信系统实施CAV，可以显着提高交通效率和安全性。