Civil Engineering Departmental Seminar - Smart Transportation and Car-following Modelling

Smart Transportation and Car-following Modelling

Car-following behaviour is a common phenomenon in road transport. The modelling of ca-following behaviour contributes significantly to the understanding of traffic flow and urban congestion. The rise of Adaptive Cruise Control (ACC) in smart transportation posts new challenges for the modelling of car-following. Most existing studies modelled the car-following (CF) behaviour of ACC vehicles using conventional CF models which were originally built for human-driving vehicles (HVs) and calibrated with HV data. To address the research gap, this study proposes a learnable CF model 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 further address the ACC data scarcity problem, a transfer learning strategy for the LSTM model is 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. Further, microscopic simulations are performed to verify the applicability of the transfer learning based LSTM CF model. In all, this study develops and verifies a data-driven learning-based model for the car-following behaviour of ACC vehicles. The outcome of the project contributes to a more solid understanding of the car-following behaviour and the mobility impact of ACC vehicles.