MSc Thesis Proposal

Optimal off-ramp metering

Mentor: B. De Schutter

Keywords: traffic control, optimization

One of the possible causes of traffic jams on highways is that sometimes the traffic cannot leave an off-ramp of the highway fast enough. The growing waiting queue spills back to the highway and causes a congestion there. An example of such a situation can be found in Delft (see figure below) on Saturdays, when the IKEA attracts a lot of visitors. The visitors coming from the south (Rotterdam) want to turn right after the off-ramp at intersection I, but there is a conflicting stream of vehicles coming from Delft that enters the highway via the on-ramp heading to the north (The Hague). Currently, there is no traffic signal at this intersection and the stream leaving Delft has the right of way. This limits the flow leaving the off-ramp which sometimes results in a congestion on the highway A13. Changing the geometry of the intersection by giving the right of way to the off-ramp traffic is no option because at the beginning of the peak-hours a lot of people want to leave Delft and the waiting queue at the intersection could spill back to the street network of the city, which is undesirable.

Layout of the
intersection near IKEA
If the visitors of IKEA cannot leave the off-ramp fast enough the congestion spills back to the highway A13. A conflicting stream comes from the vehicles that want to leave Delft in the direction of The Hague.
We want to prevent congestion spill-back to the highway by controlling the traffic signals at intersection I. The goal is to design an intelligent/adaptive/predictive traffic signal control algorithm that optimizes the traffic flows on the intersection and the highway (and thus prevents from blocking back both to the highway and the Delft street network). The controller should change the traffic signal settings as a function (of the prediction) of the flows that want to cross the intersection.

If you are interested in selecting this project as your MSc project, please come along or send us an email for more information.

This page is maintained by Bart De Schutter.