Modelling the impact, configuration and behaviour of connected and automated vehicle

• Main Author: Zhou, Jiazu
• Other Authors: Zhu Feng
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Civil engineering::Transportation
• Online Access: https://hdl.handle.net/10356/155737
• Institution: Nanyang Technological University

Connected and automated vehicles (CAVs) are designed to have potential to provide significant social and economic benefits: improved safety, increased road capacity, energy consumption and emissions reduction and congestion elimination. It is projected that CAVs will come into our life and change the way we travel in the near future. Human-driven vehicles (HVs) will share the road use with CAVs before full penetration of CAVs, which results in the mixed traffic consisting of CAVs and HVs. It is necessary to carefully model and study the effects of CAVs before the implementation of CAVs such that benefits provided by them can be full reaped. This report mainly focuses on the modeling of CAVs-involved traffic especially the mixed traffic including CAVs and HVs. Fundamental diagram (FD) is viewed as the basic of traffic flow theory, which is widely used in transportation planning, traffic control and management and traffic simulation. The first part of this report is devoted to building a FD for CAV traffic. A deterministic FD is derived based on the HV car-following model and platoon control policy. A stochastic FD is established based on the headway distribution. Simulations are completed to verify the proposed model and show the effect of various parameters such as CAV penetration, platooning intensity on the scattering of FD. Results show that increasing CAV penetration can increase traffic flow and reduce scattering in FD. However, larger platooning intensity might cause more scattering. The first study does not include the platoon size constraint in the modeling. In order to model the CAV traffic in a more realistic manner, the second study investigates the effect of CAV platoon size to provide guidance for a more effective and practical platoon configuration. More specifically, the road capacity and energy saving are formulated considering the maximum allowable platoon size constraint. Based on the developed formulation, the effect of platoon size is discussed. Simulations are performed to validate the models. Results show that increasing maximum platoon size can increase road capacity and energy saving. However, the improvements of capacity and energy saving become negligible when platoon size is large. Therefore, a moderate platoon size is recommended considering the trade-off between various stakeholders. The first two studies only focus more on a macroscopic modelling of the CAVs’ impact and configuration. The more in-depth details, e.g., the microscopic behaviours of CAVs, are not included. In order to have more details of the CAVs, microscopic modelling, e.g., car-following modelling are required. We then realised that there is a research gap in the current literature on accurate car following (CF) models used to capture the CAV’s behaviour at the microscopic scale. The accurate CF models is critical to understanding of the CAV’s longitudinal behaviours. In the third work, we move more in-depth into the behaviors of AVs. Specifically, the focus of the third work is adaptive cruise control (ACC), which is the one of the most popular applications of CAVs. Most existing studies modelled Adaptive Cruise Control (ACC) car-following (CF) behavior using conventional CF models which were originally built for human-driving vehicles (HVs) and calibrated with HV data. In this study, firstly a learnable CF model is proposed by resorting to Long Short-Term Memory (LSTM) for ACC systems, which utilizes the ACC data for model construction and offers extraordinary adaptability and accuracy compared to conventional CF models. Nevertheless, the applicability of the LSTM CF model is hindered by the scarce ACC data problem, as training the model requires a large amount of data. To address the ACC data scarcity problem, a transfer learning strategy for the LSTM model is further developed, leveraging on the large-scale open-source HV data and the similar driving patterns hidden in HV and ACC-equipped vehicles. In the transfer learning-based LSTM model, a unified framework incorporating an alignment layer is developed to transfer the useful knowledge from HV data and meanwhile calibrating the CF model with ACC data. Comparison results show that the proposed model outperforms other CF models which are built with only ACC data or using simple transfer learning methods. Further, microscopic simulations are performed to verify the applicability of the transfer learning-based LSTM CF model.