Vejle Municipality

Since December 2019, IntelliGo has tested a new type of traffic control in a traffic signal system at Bredstenvej / Skibetvej west of Vejle Municipality.

The IntelliFlow control algorithm reduces the total waiting time at the intersection from all arriving vehicles.

The control works with free phase selection, as notified phases can be skipped even if they have been notified. If, for example, only one vehicle has been notified, it is taken into account that it is very “expensive” in time to engage the phase. Therefore, congested traffic flows are generally prioritized over weak traffic flows. The algorithm does not operate with a fixed maximum green time and turnaround time, but boundary value conditions can be set to control the waiting time so that it does not become unreasonably long.

After setup, the system takes over control via phase change commands, while the necessary safety and control functions are maintained in the control unit. An essential element is to utilize traffic data for day-to-day operations by being able to quickly troubleshoot, change control and evaluate the effect.

Flexible control with local adaptation:

As long as the system has an overview of all incoming traffic, the green time is distributed so that a minimum waiting time is achieved in the individual intersection. There is thus no need for traffic technical follow-ups or special manual adjustments of the signaling over time, even if the traffic increases or there is a short-term change in traffic load in connection with eg road works, incidents or events.

The principle of signal control:

1. AT-Flow processes signal and detector input, and a virtual image of location and speed is created for all road users.

2. The vehicles are divided into lanes and types and weights are assigned based onthe choice of the road authority.

3. AT-Flow uses machine learning to review 1-3000 combinations withfree phase selection 20-25 seconds forward and calculates 1 / second the total weighted travel time for each combination.

4. Every second a decision is made on retention or change of phase, whichexecuted by the control unit. 

The possibility of using traffic data to a much greater extent to analyze the nature and extent of the problem combined with the possibility of adjusting the control provides a much more flexible approach to traffic operations in particular. Here, data can be used to reduce or completely replace the need for manual sampling observations at the intersection to determine settlement problems, while adjusting the priority allows to test the effect in the plant immediately. 

For the intersection, a pre- and post-calculation of travel times has been made for all traffic flows with Floating Car Data. That is, car journeys that have actually been driven through the intersection. Point data has been used in the evaluation, and on the basis of these, a total of approx. 20,000 individual car journeys distributed over all traffic flows.

The car journeys have been analyzed in order to map the driving time through the signaling system for each of the 12 traffic flows, and a calculation has been made of the change in average driving speeds for each traffic flow, as shown in Figure 1. The driving time is calculated at an approx. 500 meters stretch before the intersection and to just after the stop line. The stretch must be long enough to include the normal maximum queuing.

Results and experiences

An independent evaluation shows an overall reduction of the waiting time through the intersection of 10-20%.

There has also been a reduction in the number of stops. In general, it is demonstrated that the large straight-ahead traffic flows are prioritized and the smaller swing currents are downgraded.

The control algorithm works exclusively by optimizing single crosses. At coordinated facilities, simulations have shown good results for travel time compared to traditional coordination. 

Overall, the average travel time change per. trip through the intersection improved quite markedly compared to the original control of the traffic signal system:

• Morning (07-09): -2.85 sec.

• Afternoon (15-17): -4.11 sec.

• Outside rush hour: -1.65 sec.

• All days (05-20): -2.49 sec. 

The improvements correspond to a 10-20% reduced travel time for the main flows, while the smaller traffic flows that are in conflict with the main flows experience a significantly increased travel time. Overall, travel time is reduced due to the high traffic volumes of the main streams.