Traffic Flow Variables Estimation by Artificial Intelligence: Theoretical Aspects and Case Studies

Macroscopic traffic variables are of fundamental importance in traffic engineering. This chapter summarizes the results obtained using an automatic road traffic data acquisition procedure called MOM-DL. This procedure is primarily based on Wardrop’s well-known moving observer method (MOM) and requires deep learning algorithms. The proposed technique facilitates the estimation of the vehicular flow rate (q), the space mean speed (vs) and the vehicle density (k) on roads with uninterrupted and stationary flow conditions. The method was applied experimentally on the SS624 highway in Italy. The results demonstrate that the MOM-DL technique is reliable in determining the macroscopic traffic variables of the flow and can be increasingly used in the future in various traffic and transportation engineering applications.