Detecting traffic jams in Ulm
Traffic flow improvement is very important to cities as the number of road users keeps growing. To be able to get more out of current road capacity, municipalities are turning to smart digital technologies like traffic detection and traffic control to improve the situation. The city of ULM wanted to become a smart city and utilize non-intrusive cutting edge technologies for traffic monitoring based on AI so they have decided to implement an innovative solution and CROSS delivered it together with DataFromSky and INVIPO.
The city of Ulm (DE) decided to monitor the formation of columns at the entrance to a busy roundabout and to understand how they form and why. To get the full picture of how the column forms, it was necessary to know at what times of the day the column emerges, what is the difference between working days and weekends and how the tourist season or holidays affect it. From the traffic monitoring perspective, the most challenging part of this project was to get full coverage of a 400 m long segment in a cost-efficient way.
The system had to provide very detailed information from each monitored section such as the unique functionality for measuring level of traffic and the length of congestion in meters if it was present. All video processing needed to be done on the edge due to GDPR compliance and there also was a need for having only one communication link with LTE and low data bandwidth available so a modem was used to minimize the costs of data transfers. The city wanted to find and implement an innovative non-intrusive telematic solution for which FLOW was the ideal candidate.
Ideal camera placement designed by DataFromSky.
DataFromSky solution - 4x TrafficEmbedded with FLOW inside
A set of 8 collaborative RTSP cameras is installed on light poles to cover the 400m driveway just before the roundabout. We designed the camera and unit placement location and height to achieve optimal coverage. There are 8 cameras connected to 4 TrafficEmbedded units to process the video feed into traffic data. WiFi connection is established between the units for data sharing and the master TrafficEmbedded Micro unit has LTE to communicate the traffic insights to the INVIPO smart city using REST API every 5 minutes sending aggregated data. The operating expenses of this system were significantly reduced thanks to the solution having only one LTE modem and communicating this data at longer intervals instead of streaming it continuously. Edge processing not only makes the units GDPR compliant but also reduces data communication needs drastically.
In the image you can see one of the spatial filter zones that monitors the traffic, measures passage time through the zone and based on the speeds of cars in the different zones the system determines the length of the traffic jam.
The evaluation of exact vehicle traces extracted from the video streams is done by FLOW framework running on the TrafficEmbedded units. The units were remotely set up to measure traffic intensity, calculate the level of service, and provide the information about the length of traffic jams. All of this in real-time to allow immediate responses to the current traffic situation.
FLOW insights offers a range of data visualisation options to choose from including simple values, statistical values, time and distribution based graphs, OD matrices, heatmaps and gridmaps.
For getting the correct level of service the FLOW filtered out all single track vehicles using the category filter. Level of service is then calculated from the objects that have already passed through the zone. Using the time block settings the data output is the average of the last 5 minutes. FLOW scripting feature was used for the optimization of measuring the length of the column by a traffic expert. Thanks to this unique function FLOW provided the optimal solution for pinpointing the length of the columns as well as evaluating the changes of traffic flow in the monitored road segment.
Powering the smart city platform INVIPO
The traffic data and events detected by FLOW framework are automatically sent to a smart-city platform INVIPO via a secured REST API. This platform stores all data and offers detailed historical statistics and exports, which the operator can then use to improve the traffic situation in the city. The platform provides a virtual twin of the intersection from bird’s-eye view for easy comprehension. The most recent data from the scene is clearly visualized with an overview of each of the road sections and data on traffic flow in each of the traffic lanes. Each of the visualized zones has color-coded level of service and traffic flow data for 5 minutes and 1 hour. There is also a graph showing the length of the traffic congestion throughout the day which is enabled by the FLOW expression scripting feature.
INVIPO customized dashboard – visualization of the current traffic situation in a 400m long road section in real time. It shows the current level of service, length of the congestion, and traffic statistics per segment in a defined time-window.
We have been able to deliver better value, functionality and easier installation compared to the original proposed solution of using induction loops. We have achieved continuous 24/7 monitoring and data gathering of detailed traffic statistics for each road lane. Cost-effective video-analytic traffic monitoring system was deployed thanks to optimal camera coverage, smart data usage and multi-purpose nature of the FLOW framework. The detection and reporting of the level of service and possible congestion is automatic and all 8 video streams are processed on the fly on locally installed video-analytics units TrafficEmbedded. The whole system has been running for months without any issues.
The solution is a joint project of the Czech companies #CrossZlín, #Incinity, and #DataFromSky. Where CROSS Zlín is a project coordinator, INCINITY delivered the smart city platform INVIPO to aggregate, gather and visualize the data, and DataFromSky delivered the video-analytic traffic monitoring solution based on the FLOW framework including HW.
- Real-time monitoring of a 400 meter long section of high speed city road with multiple junctions – use of non-intrusive technology
- Measuring the length of traffic congestion and the level of service
- Connection with the INVIPO smart city platform for visualisation of the current situation as well as gathering of historical data
- 4x TrafficEmbedded with 8x RTSP cameras installed on light poles based on ideal design in terms of positioning and height for maximum coverage
- Wireless communication – WiFi between the units and one master unit with LTE for external communications
- FLOW framework for making the cameras collaborative
- FLOW LoS operator combined with FLOW scripting feature for congestion detection
- FLOW REST API for communication with INVIPO
- 24/7 real-time monitoring of traffic flow and level of service
- Unique function of measuring the length of the traffic jam
- Non-intrusive nature and fast deployment of the solution
- Remote and on-demand addition of new analytics features / traffic metrics thanks to FLOW framework
- Smooth integration with the INVIPO smart city platform