Page Nav


Classic Header




Driving the traffic, not the car - smart intersections

The biggest problem with city traffic is intersections. The intersections cause gridlocks, delays, and are the main frustrations with dri...

The biggest problem with city traffic is intersections. The intersections cause gridlocks, delays, and are the main frustrations with driving in the city.

Of course, it is necessary to regulate traffic. However, today's intersection regulation by traffic lights is mainly only to avoid accidents instead of controlling traffic. Usually, the traffic lights are programmed with mostly one or maybe a few settings according to what part of the day, however, these settings are not flexible to change the traffic's demands. Sometimes, a linked intersection uses synchronized traffic light to provide a green-wave; however, this feature is not sensitive to the traffic; it is only time synchronization between intersections.

There is some example to make the traffic lights driven intersections smarter by sensors built in the streets. This type of intersections are more sensitive for the traffic, but its sensitivity is very short-ranged, cannot see far enough, or very expensive to set up and easy to malfunction.

A smart intersection is sensitive to the traffic, it sees the speed of the cars far enough to respond flexibly with less uphold of the traffic, can recognize the change of the traffic situations, can automatically adapt to it, even capable of optimizing its operation by self-learning.

Now the technology is ready to change this situation, ready to make intersection control to be smart. Using the sensors similar to which are built into smart cars is capable not only to drive a car automatically but driving the traffic at an intersection too if it used for that purpose.

The cameras, laser scanners and other sensors used in the smart car technology is capable to be used to set up smart intersections, capable to see the traffic in its movement, recognize the speed of the cars, recognize the pedestrians, even recognize the emergency vehicles far enough to get the smoothest flow of the traffic through the intersection. An intersection does not need many sensors. A sensor, which is capable of seeing in 360 degrees, is capable of seeing the traffic from all directions. These types of sensors are available, sometimes used in smart car technology.

To make an intersection smart, we need a capable software besides the sensors. The software controls the traffic light based on the inputs from the sensors. The primary purpose of it to maximize the average speed of the cars crossing the intersection. Because the logic to control the traffic does not need to change, the software does not even need self-learning capabilities. However, to make the intersection control more flexible, the try-till-success type self-learning software would make the intersection more independent of the maintenance, does not need rule updates for more optimized operation.

The smart intersections do not even need to communicate with each other directly sending status signals to the nearby smart intersections. The smart intersections are self-organizing systems. The nearby intersections are communicating with each other indirectly, through the traffic. This indirect way makes it not necessary to set up direct communication between intersections. The self-organizing feature is a theoretically interesting complementary property of this kind of system. It would be interesting to observe how the independently existing units become a self-organizing system without any centralized control.

In summary, the technology is ready for smart intersections. The hardware is available and cheaper and cheaper, the software can be tested and simulated in a computer environment, the smart intersection system can be easily applied to the existing traffic lighted intersections, and can be implemented gradually, one by one. Besides these, we must think about the economic, environmental and lifestyle benefits to make a city more livable. The traffic is waiting for smart intersections.

This kind of passive smart intersections is easier to set up, maintain, control and operate than the other kind, the active approach to creating smart intersections, where the vehicles inside the traffic communicate directly with the traffic lights, as some research in this field is directed. The active approach is more complicated, expensive, and vulnerable than the passive, described above.

1 comment

  1. Finally happening:
    Fully-Autonomous, Vision-based Traffic Signal Control: from Simulation to Reality