Real time delay estimation for signalized intersection using transit vehicle positioning data

The paper proposed a practical method to accurately estimate the vehicle delay at signalized intersections. Firstly we derived an analytical delay formula based on the observation of the queue forming/discharging process. According to uniform arrivals assumption, the formula was then simplified to a linear relation between the travel time and the observed vehicle location. However, the factors in the formula are associated with the specific intersections´ queue characteristics. So we collected massive historical trajectory data using GPS instrumented transit vehicles, and analyzed it with DBSCAN and Least Square Fit algorithms to determine the time-dependent factors at each intersection. In this way the formula can calculate precise intersection delay only base on vehicle positioning data, without the need to know signal or traffic information. Field experiments are conducted and the results verify the effectiveness of the proposed method.