Yandex Traffic Jam Technology Overview

How Yandex.Traffic Works

Yandex.Traffic shows the picture of current traffic conditions in a city. It gathers information from different sources, analyses this data and maps results on the city’s map on a web-based mapping service, Yandex.Maps. For the larger cities, where traffic jams are a serious problem rather than a small inconvenience, Yandex.Traffic also calculates average levels of congestion on a scale from 0 to 10. To put a current and accurate picture of traffic conditions in a city on the map, Yandex.Traffic goes a long way, sourcing information, among other places, from those who generate traffic. In other words, Yandex’s technology helps drivers help each other.

Sourcing data

Imagine having a small traffic accident – no victims, just a couple of scratches. Your bad luck blocks two out of three lanes on a major city street. Drivers in these two lanes now drive around your scratched vehicle, while drivers in the third lane let these cars move into their lane. Some of the third-lane drivers use the Yandex.Maps application on their mobile devices. The app sends information about the movements of their car to Yandex.Traffic. The users of the Yandex.Traffic app slow down as they approach the spot of your accident, while their mobile devices send signals of a potential traffic jam to the Yandex.Traffic service.
The only thing a city motorist needs to join the common effort of gathering traffic information is an internet-enabled mobile device – smartphone or tablet – with a GPS function, built-in or using external receiver, and the Yandex.Maps app or the Yandex.Navigator app. After activating the app’s “Send traffic information” option, the user starts sending every few seconds their geographic coordinates, direction and speed to the automated analytical system of the Yandex.Traffic service. All information sent from each mobile device is non-personal – there is nothing that could possibly betray any specific information about the user or the car. Yandex.Traffic’s automated analyzer then integrates the speed and coordinate information from all participating vehicles driving along the same route into unified traffic patterns – tracks. In addition to contributions from private users, Yandex.Traffic also receives information from companies that have fleets of vehicles operating in the city’s streets on a regular basis.
Other than sending their coordinates to Yandex.Traffic, drivers can also inform the service about traffic accidents, road works or other events that can cause congestion. Having spotted your accident, for instance, a conscious driver can mark it on their Yandex.Maps map.

Tracks Processing Technology

To make a track for a moving car, Yandex.Traffic uses a number of geographic coordinates, which are delivered by the driver’s GPS device and sent to the service through the Yandex.Maps app. The GPS accuracy, however, has the error margin from one to ten meters in all directions, which may result in positioning a car on a sidewalk or rooftop of the nearest building. To solve this problem, GPS coordinates are mapped to the digital map of the city, which accurately displays roads and streets with all the markings, buildings, parks, and other urban facilities. This detailed mapping allows the system correct the course of a car based on the real physical layout even if the GPS coordinates say that the car is on the wrong side of the road or has cut through a building instead of following road markings and turning around the corner.
Another important issue is to understand how useful the speed information received from the driver is, as it may or may not truthfully reflect the real situation on the road. When a driver who sends information about their movements via the Yandex.Maps app slows down or stops, they may do that because they don’t know if they need to turn, or because they just want to grab some milk from a corner shop. If all other cars sending information to Yandex.Traffic proceed as normal on the same route, the system ignores the rogue track and this data isn’t considered for the general traffic evaluation. This is exactly why the number of Yandex.Traffic users matters. The more drivers send information to Yandex.Traffic via Yandex.Maps, the more accurate the picture of the real-time traffic situation is.
The Yandex.Traffic system uses several reliable tracks to colours specific road segments on the map “green”, “yellow” or “red” according to the current traffic density.

Merging Data

The next step is to bring all available information together. Every two minutes, Yandex.Maps aggregates, like a jigsaw puzzle, all information from all users of the Yandex.Maps app and maps the results on the Traffic Jams layer, both in the mobile application and on the desktop service.

Point scale

For Moscow, Saint Petersburg and other large cities in Russia, Ukraine and Turkey, where traffic jams have almost turned into a natural disaster, Yandex.Traffic offers a ten-point scale for levels of traffic congestion, with zero points for free flowing traffic and 10 points for a ‘complete standstill’. This scale helps drivers to instantly estimate, roughly, how much time they are likely to waste in a traffic jam. An average seven points in Kiev, for instance, means that the travel is likely to take approximately twice as long as when the roads are free.
For each of the cities, the scale is localised – what is ‘slightly sluggish’ in Moscow is a big problem somewhere else. Congestion level of six points in Saint Petersburg would make a local driver waste as much time as in Moscow at a traffic congestion level of five points.
The Yandex.Traffic congestion scale is based on reference time – the amount of time an average driver would spend driving by the rules at a zero traffic through a standard route, which covers all major roads and avenues. The service estimates the city’s general traffic congestion at a given moment and calculates the difference between the reference time and the time calculated from the speed information it receives from all participating cars. Using the time difference for every car, the program then calculates the mean average, which translates into traffic congestion points for the whole city.