In the Smart Junction project the goal is apply new sensor technologies to improve the situation awareness for various purposes including evaluation of the traffic performance, improvement of traffic management and signal control as well as for connected and automated vehicles.
The basic approach is to combine sensor data from various sources together with the general knowledge about the site and traffic flow properties, in order to create one consistent picture of the present traffic situation. This picture (called real-time digital twin), is supposed to reflect the actual traffic situation in high spatial and temporal accuracy and with very low latency.
The Smart Junction project is carried out in co-operation between Aalto University, City of Helsinki, Port of Helsinki and Conveqs Oy. Several other companies are involved in the project, too. City of Helsinki is providing the test area called Jätkäsaari Mobility Lab. Aalto University is carrying out reseach by using the new sensor data. Conveqs Oy is developing the open data platform.
New type of traffic sensors are being installed to the six critical junctions from the harbour to the Länsiväylä motorway. Each junction is equipped with multiple sensor stations consisting of camera, radar and edge computing unit. Each sensor station is pointing upstream of one approach of the junction. Laser scanning is applied in one junction to detect pedestrian and bicycles.
The data from the sensor stations is collected by an open data platform that connects the data to an open interface (API). Data is also collected from the signal controllers (state of traffic signals and/or detectors) and from the public transport API (real-time positioning of trams and other public transport).
The data from various sources is combined with so called real-time digital twin that represents the overall traffic situation in a machine-readable form. The digital twin is also used to fill in the gaps between various sensors and to reduce the noise of the sensor data. The digital twin is also internally consistent model thus resolving the potential conflicts between various sensor data.
A multitude of measures of performance (MoE) can be computed from the digital twin including aspects of traffic performance, traffic safety and environmental impact. The refined data can be shared through API:s or web-based dashboard. The detailed information of the traffic situation can also be used for optimizing the traffic signal control.
Several types of new sensors are installed into the Smart Junction test area within the Jätkäsaari Mobility Lab of Helsinki. The area to be covered consists of six consecutive intersections that are the main connection between the harbour and the motorway.
Sensors and installations
A sensor station consist of edge computing unit and the sensors like camera and radar. Each unit is installed to point from the intersection towards the approaching vehicles. within the intersection area an additional lidar unit can be used.
Additional data is obtained from the signal controllers, namely the status of traffic signals and detectors. The position of public transport like trams is received in real-time through and open data API.
Live video stream is obtained from the area through several cameras. The camera data provides a ground truth against which the other sensor data can be verified. From the video stream the vehicle types can be recognized better than from radar for example, but speed and position data is less accurate. Lighting conditions may also have an effect.
Radar data is used to obtain accurate speed and position of the vehicles. Radar is less prone to weather and lighting conditions than a camera. Radar is not so good in detecting slow moving and small object like pedestrians and bicycles.
Laser scanning (lidar) is a potential new method for collecting real-time traffic data. The range is somewhat smaller than than that of radar, but lidar can be used to detect the vehicle/road user type, too. In the Smart Junction project lidar is used to detect the traffic within the intersection area including the pedestrian and bicycle traffic.
All the real-time traffic data is collected by using an open data platform of Conveqs Oy. The platform collect the data from the sensors several time per second and process it in a server. The sensor data is provided to other users through an open API.