Abstract
With emerging technologies of Connected and Automated Vehicles (CAV), various applications are being developed to help drivers with different driving tasks. Due to the ignorable reaction time and accurate control of CAVs, most of the transportation issues, such as traffic safety, congestion, and energy economy, are expected to be improved. In the meantime, there are public concerns regarding vehicle automation including system liability, deliberate interference, and interaction with human drivers. Therefore, this dissertation aims to assess the impacts of potential risks in CAVs and evaluate the mixed traffic performance.
However, the traditional method of evaluating the mixed traffic uses strong assumptions by simply setting different parameters, such as desired headways, for CAVs from human drivers. Therefore, an integrated simulation platform is developed for realistic modeling of CAVs and mixed traffic. With the simulation platform, this research aims to: (1) provide a realistic platform for CAVs and mixed traffic simulation; (2) quantify the safety impact of cyber-attacks to Cooperative Adaptive Cruise Control (CACC) platoon, which is the prevailing and mature CAV application; (3) improve the traditional CACC algorithm to overcome the potential risks; and (4) provide the impacts of CACC platoons in the mixed traffic flow.
Particularly, the proposed platform explicitly simulates CAVs by considering vehicle dynamics, realistic sensors, communications, and controllers, where CACC is adopted in this research. Extreme cases, i.e., different degree of cyber-attacks that are most likely to happen, are simulated. The crashes are reconstructed to quantify the injury severity using a dedicated mathematical method. Cyber-attacks do not always result in crashes, but they create large oscillations. Even with Emergency Braking System (EBS) being implemented, some crashes caused by radar attack are avoided, since EBS heavily relies on radar. Besides, mode switches between CACC and EBS cause many control jerks and violent speed oscillations.
Considering the potential risks in the traditional CACC, a robust CACC algorithm against cyber-attack is developed. The proposed algorithm combines the advantages of all-predecessor following (APF) and predecessor-leader following (PLF) control methods to improve the stability and robustness. String stability of the proposed CACC algorithm is theoretically proven, and the performance is validated with simulation. Except for very extreme braking, which rarely happens, the proposed algorithm is capable of ensuring the robustness in various cyber-attacks without control jerk and outperforms the traditional CACC algorithm.
CACC platoons keep a short gap, i.e., 0.6s, to allow CAVs to bind together for efficient movements. However, the long and dense platoons may impede lane changing of human drivers, especially on non-basic freeway segments. Therefore, the mixed traffic on a weaving segment is simulated to explore the impact of platoons. The results reveal that the speed in mixed traffic drops more than in normal traffic (up to 27 km/h), especially with shorter desired gap, lower speed, or higher market penetration rate. Therefore, it is suggested to activate CACC mode at high speed or increase gaps when approaching weaving segments.
This research is expected to provide a powerful and useful tool for mixed traffic simulations. The pros and cons of CAVs in traffic flow-wide can be explored to help the researchers improve the algorithms for traffic system before further implementation. The platform can be enriched by adding various vehicle dynamics models and CAV controllers to be compatible with comprehensive mixed traffic simulation in the future.
