An innovative solution for the adaptive control of traffic lights


An innovative solution for the adaptive control of traffic lights

Adaptive regulation allows you to get the maximum possible mobility without rebuilding the existing transport infrastructure, that is, with a minimum investment. The basis for such traffic optimization is the regularly updated data on the road traffic situation. Based on the information received, the following measures can be applied:

  • organization of reverse movement;
  • adaptive regulation of intersections;
  • coordinated management of the intersection network;
  • setting the recommended driving speed.

The obstacle to the widespread introduction of adaptive regulation today is the laboriousness of the regular collection of data on traffic flows and the high cost of maintaining a fleet of transport sensors. The problem with the content of the sensor is that when anomalies are detected in its readings, it is difficult to establish whether it is correct and displays an objective picture.

To resolve this issue, it is necessary to duplicate sensors, travel to the location, and install video cameras. As a result, projects on adaptive regulation of intersections often degenerate into scenario regulation, which actually comes down to changing the RRP and does not give the desired effect.

But this problem is becoming a thing of the past: computer vision and machine learning technologies have made adaptive regulation a reality. The only complete solution in Russia today is TrafficData software. At the same time, for the implementation of the adaptive regulation system on its basis, no additional hardware is required, the available outdoor video surveillance cameras are sufficient.

Let us consider the problem in more detail using the example of a standard cruciform intersection. To do this, let’s take one of the crossroads of the city of Ufa, equipped with a video surveillance camera (Oktyabrya Avenue – 50th Anniversary of the USSR Street).


We will formulate the problem of adaptive regulation as follows: creating an intersection capable of automatically selecting the optimal signal plan, responding to changes in the traffic situation (Fig. 1).

Adaptive Intersection Control consists of 4 main steps:

  1. Collecting data on the traffic situation. This is done using computer vision software TrafficData Land. TrafficData Land turns an ordinary CCTV camera into a wide-range video sensor. The detection of vehicles and pedestrians is carried out, the trajectories of traffic participants are built. In this case, for example, for a standard cross-road, one camera is enough installed with the correct angle. Data is collected in streaming mode using the TrafficData Live module from road network surveillance cameras. To do this, it is enough to provide the program with access to the ip-camera via the RTSP protocol.
  2. Traffic flow analysis using video analytics tools built into TrafficData Land. The main initial data that are required to optimize the duration of the phases of traffic light objects are the intensity of vehicles and pedestrians in the directions of movement, as well as the determination of the composition of the traffic flow to bring the intensity to a passenger car. In the absence of congestion situations, that is, when all the cars in the direction have time to pass the permitting signal of the traffic light, this data is sufficient. Otherwise, additional data is required:
    a) determination of occupied lanes;
    b) determination of the lengths of queues on the way to the regulated object;
    c) determination of the delay in the movement of the vehicle in the queue and when passing the intersection at the permitting signal of the traffic light.
  3. Calculation of the optimal durations of the phases of traffic light objects based on the transport model of the intersection. Traffic data updated every 5 minutes, processed by the video analytics module, is transferred to a previously prepared mathematical model of the intersection. Next, the optimal durations of the phases of the traffic light cycles of the intersection are calculated. The optimum criterion is a minimum delay in the movement of vehicles and pedestrians.
  4. Updating traffic light cycles. The optimal traffic controller program is calculated and downloaded every 5 minutes, based on the current initial traffic data. The program can be loaded automatically or you can ask for confirmation from the manager. The described cycle of 4 steps is repeated every 5 minutes. This allows you to quickly update the operating mode of the traffic light object, responding to changes in the traffic situation.
Figure: 1. Statement of the problem of adaptive regulation
Figure: 1. Statement of the problem of adaptive regulation
Рис. 2. Анализ транспортного потока
Figure: 2. Analysis of traffic flow
Determining queue lengths
Determining queue lengths



To monitor the quality of adaptive regulation of intersections, a dispatching system has been developed that performs the following functions:

  • diagnostics of anomalies – automatic assessment of the performance of a video surveillance camera, checking signal reception, focusing, lens contamination / obstruction, viewing direction, etc.;
  • when anomalies are detected using video analytics, the system sends a notification to the dispatcher about the need to configure a video surveillance camera;
  • display of the traffic situation in real time, supplemented by video analytics information;
  • display of active signal plans selected by the system;
  • system operation management; the ability to switch between automatic and manual assignment of signal plans;
  • displaying geodata of a traffic light object; graphic display of the regulated object on the map for ease of navigation in the presence of a network of objects.
  • Dispatching the regulation process
    Dispatching the regulation process


It is possible to expand the number of automatically controlled intersections and combine them into a single network. This scalability allows you to solve the following tasks:

  • creation of an automated traffic monitoring system UDS;
  • regular data updates to update the city’s transport model;
  • automation of the management of plans for the coordination of a set of intersections of the road network;
  • creation of a technical base for ITS.


The speed of construction of highways often does not keep pace with the increasing traffic load on the road network. Therefore, traffic optimization has long attracted engineers as an economical and efficient way to solve traffic problems. Unfortunately, previously there was a lack of technologies that would make it possible to fully implement the optimal use of the UTS. Today the situation has changed. Now we can make a more balanced decision about new construction and resort to it after we have achieved the maximum efficiency of the road traffic system by optimizing traffic flows.

(Source: Magazine Roads. Innovations in construction # 88. September, 2020. Download in PDF)