Adaptive traffic control in Wroclaw
Adaptive traffic lights signalling was only possible with inductive loops until recently and did not achieve a widespread adaptation as the data they provide are very limited while maintaining them not only disrupts the traffic but is also costly. On top of that having just rough car passage data without information on the other traffic participants and awareness of the overall situation in the intersection can get the efficiency of adaptive traffic control only so far. However, with the advances in computer vision and AI, adaptive traffic control with Smart cameras or other on the edge processing devices like our TrafficXRoads became not only possible but also quite affordable.
Thank to AI video analytics it is possible to detect the exact number of vehicles in each lane, determine their category and monitor their wait time. All while also detecting pedestrians and bicycles and other traffic participants and monitoring their numbers and wait times and then combining this data with the vehicle data to create a tailor-made traffic lights signalization plan for the specific situation. All this is done in real-time with minimal communication delays allowing for truly adaptive traffic control.
In places where costly redesign and reconstruction of the junction are out of the question, adaptive traffic control with video analytics can create the much-needed improvement of traffic flow sometimes even doubling the throughput. On top of gathering real-time data, the system is capable of gathering long-term statistics which was one of the areas city of Wroclaw was interested in. Seeing the clear benefits of AI video analytics in perspective of the constantly rising traffic that could not be ignored, the city of Wroclaw has decided to take the first steps to test out and see what this system is capable of firsthand. Therefore they have decided to install the TrafficXroads unit with the help of our local partner in early 2022.
TrafficXRoads - adaptive traffic control setup
TrafficXroads is very easy to deploy to existing crossroads and allows to upgrade them with adaptive traffic control capabilities. The TrafficXRoads unit was conveniently installed inside the case of the traffic controller. As it needs video as input to extract traffic insights, the cameras were installed, one for each direction of the T junction and one for monitoring the public transport stop. The cameras were 3rd party cameras as decided by the project partner and were set in such a way so each of them could monitor multiple lanes, therefore, requiring only 4 in total for this project. This way incoming cars can also be detected from a distance to trigger green light to allow free passage without braking if the junction is clear of traffic.
List of the 4 added cameras (one analytics per camera)
After the HW was installed, the multipurpose sensor system was configured in the FLOW Insights desktop application. This could have also been done on the spot but thanks to the internet connectivity of the traffic controller a remote configuration was done instead. All of the 4 camera streams were added in FLOW insights after connecting to the unit. Then the detection zones were set up that act as “virtual inductive loops” in the footage in accordance with the project design. Special zones were also set up to be able to adaptively give priority to public transport vehicles. This data was communicated via data sinks to the traffic controller. Apart from the various data sinks FLOW devices are also compatible with I/O relays that can work with any type of traffic controller and also support the SDLC protocol.
Data analytics configuration
After the setup of adaptive traffic control zones for real-time data the collection of of long-term traffic statistics was also done including the counting and visualisation of the category distribution of traffic flow (cars, vans, heavy vehicles and buses) including active mobility categories such as pedestrians and cyclists, turning the unit into a strategic traffic monitoring sensor.
To better visualize and interpret the long-term data, dashboard was created in FLOW that was analyzing multiple values and performance measures such as:
FILL IN BASED ON PRINTSCREEENSS
- Number of vehicles that have stopped / total number of vehicles
- Average stationary duration in the crossroad
- Traffic intensity – categorization into 7 classes
- Average vehicle speed
While the configuration was being done it was possible to see the real-time output of the analysis helping to understand what data is being gathered with ease.
FLOW interface – virtual induction loops setup – configured in a few minutes
The deployment of the system was successful and it is running well without any disruptions. The dynamic control of traffic is helping with traffic flow and also to keep public transportation on schedule by giving it priority. The long-term data that is being gathered can also be helpful to the city in regards to understanding traffic compositions and traffic flow paterns in the city to help with further projects and city development. Lastly, the improvement of traffic flow also had a positive impact on the local noise and emission levels.
Want to read more? Go back to the BLOG.
- Full coverage – all directions
- Long distance detection
- Real-time detection and classification of all traffic participants
- Full data integration with traffic controller
- Comprehensive adaptive traffic control accounting for the full picture of the current situation
- 2 in 1 solution – extraction of real-time adaptive control data and long-term traffic statistics in one system
- 24/7 real-time traffic monitoring with camera surveillance
- Non-intrusive installation of the system
- On-site processing – fully autonomous adaptive traffic control system
- Strategic traffic sensor gathering long-term statistics and real-time traffic insights