T. Sultana, Y. Hassan
The development and deployment of connected vehicles (CVs) is expected to increase mobility, safety, and environmental benefits. CV technology can be utilized to disseminate safety-related messages and develop a speed advisory system to control speeds and minimize idling time and fuel consumption. However, CV speed behavior related to road geometry is not currently quantified. This paper uses data collected from the Safety Pilot Model Deployment (SPMD) Project to model CV speeds on horizontal curves. Random effects were incorporated in the model to account for the variability of the data with multiple trips conducted by the same driver. Multiple linear mixed effect models (LME) were developed using the instantaneous speed measures representing CV speeds at different points of horizontal curves. Machine learning, an advanced modelling technique, was also used to develop artificial neural network (ANN) models which can be utilized with continuous updating in real-time monitoring and prediction of CV speeds in a specific region. The findings of the study related to CV driving behaviour are compatible with most of the studies available in the literature in non-CV (NCV) environment. The results indicated that CVs may travel on horizontal curves at higher speeds than NCVs, which may increase speed disparity for fleets of mixed vehicle technologies.
Keywords: operating speed; connected vehicles; linear mixed effects models; machine learning; artificial neural networks