Studying the Dynamics and Spatiotemporal Interdependencies of Traffic Flow in Heterogeneous Conditions
The main goal of this thesis was to study the dynamics and spatiotemporal Interdependencies of traffic flow in heterogeneous conditions and their effects on section performance. To accomplish that, trajectories of vehicles were extracted using image processing techniques and these data were then analyzed for calculating the desired characteristics of the traffic flow in a developing country.
DATA
The traffic flow observations were carried out for two major highways in Tehran. The data were gathered in such a way that ensured enough data were gathered for all traffic conditions (saturated, unsaturated and the transition phase). The process of extracting vehicles’ trajectories from the videos was conducted in MATLAB using OpenCV (Open Source Computer Vision Library). the above figure shows a sample frame, before and after the image processing operation.The output of the image processing was in pixel coordinates and it had to be converted into the actual coordinates (in meters). This conversion was conducted using a transformation matrix. Using this matrix, the perspectives in the pictures are also removed. After modifying the raw outputs of the image processing, the results were validated using field data.
Lateral position of vehicles in the section
In order to investigate frequency distribution of lateral position (lateral distribution) of vehicles in the roadway section, the section was divided into imaginarily 25-centimeter strips and then the frequency distribution of vehicles in each strip was calculated. The picture at the left shows the frequency distribution of lateral positions of vehicles in one of the sections in saturated and unsaturated traffic conditions. As can be seen, in saturated traffic conditions, vehicles reduce their lateral distance and do not follow the lane discipline. In this situation, on a six-lane highway, we may observe up to six rows of vehicles.
In order to observe how vehicles tend to reduce their lateral distance and form new rows on highways, the frequency distribution of vehicles studied in different vehicles density. As can be observed, in a six-lane highway, at densities around 200 vehicles per kilometer per section, vehicles start to form new rows. This is the point that the section capacity occurs (maximum flow).
Lane utilization
Lane utilization can have a significant impact on sections capacity. Using the trajectory database, lane utilization factors were calculated for each section at different ranges of densities (values were aggregated at 30 seconds intervals). Afterward, a simple curve was fitted to the scatterplot. The results were then compared with those in developed countries. Here, we only present one of the figures to compare lane utilization between Iran and the USA.
One interesting difference is that in the USA, the fitted curves converge earlier (at k/kj=0.5), while the curves for Iran converge later nearly at the critical density (k/kj=1). This can have negative impacts on the performance of the section in terms of throughput (capacity) and safety (due to the higher conflicts between vehicles trying to change their lane to use the adjacent free lane near the capacity situation).