Сognitive Navigation

The main articles are:

• What is the Internet of Things (IoT)
• Video analytics (terms, applications, technologies)
• Computer vision (machine vision)

2024: Cognitive Navigation Performance

Russia has developed its own navigation system for smart trams. Cognitive Pilot (Cognitive Robotics) announced this on July 4, 2024.

Cognitive Pilot solution is being implemented in St. Petersburg.

One of the key problems in organizing autonomous traffic in urban conditions is determining its exact location. The traditional approach to solving it is to use a GPS signal. However, in the conditions of dense development of high-rise buildings (canyon effect), park areas, bridges, tunnels, defense and other organizations where the GPS signal is "jammed" by external devices, it is not possible to constantly use GPS navigation.

To effectively solve it and create a guaranteed mechanism for determining the coordinates of a city tram, Cognitive Pilot specialists have developed their own navigation system - Cognitive Navigation technology.

The technology we have developed allows, according to experts, to increase traffic safety up to 20-25%. The market for such solutions has not yet been formed, but analysts believe that its volume in the next 5 - 7 years may amount to about 1 - 1.5 billion dollars, "said Olga Uskova, CEO of Cognitive Pilot.

The main elements of Cognitive Navigation are an inertial module (a gyroscopic device responsible for the orientation of the tram in space), an odometric sensor (a means of determining the distance traveled by the tram based on the measurement of the angle of rotation of the wheels) and a map of the entire rail network of the city (including all possible routes of movement, transport depots, interchanges, etc.).

Cognitive Pilot

In the autonomous tram control system, knowledge of the exact coordinates is necessary to ensure the operation of a number of key subsystems. Firstly, the system for controlling the passage of intersections at traffic lights. This is required when recognizing traffic lights, matching them with a map and determining the possibility of moving along an intersection. Secondly, for the operation of the collision prevention system. This is necessary when performing braking in order to calculate its necessary parameters to ensure the maximum smoothness mode for passengers. And thirdly, for the speed profile control system (autopilot does not allow the tram driver to exceed the specified speed in various sections of the route). To activate (turn off) this module, it is necessary to clearly know at which points it is necessary to do so.

Knowing the coordinates of the tram at every moment of time is the most important tool for ensuring traffic safety, "said Gennady Savitsky, leading developer of Cognitive Pilot. - We can calculate them with an accuracy of 1-2 cm, although in most cases it is enough to have an accuracy of up to half a meter.

The principle of Cognitive Navigation is as follows. Before operating the autopilot, in situations where all components of the technology work and all sensors: GPS, odometry and inertial unit, a specially designed mathematical apparatus allows you to calculate the error of each of the systems, which will later be taken into account if one of the modules does not work. For example, if a GPS signal is missing, navigation is carried out based on data from an odometric sensor, an inertial block, as well as a path map, which allows you to maintain centimeter accuracy.

The mechanism for determining errors is simply unique. It allows you to form the history of their changes for each tram, determine the serviceability of sensors, observe the dynamics of changes, as for example, when pulling wheels - reducing their radius during operation, - said Gennady Savitsky.

Like all Cognitive Pilot autopilot systems, Cognitive Navigation also uses an anthropomorphic model of artificial intelligence when determining the exact coordinates of a tram. As a person, finding himself in a familiar place, begins to remember the nearest landmarks (road infrastructure objects) from memory and compare them with what he actually sees, Cognitive Navigation compares the data received from the computer vision system with the data of the terrain map and in case of their discrepancy makes the necessary correction of coordinates, giving preference to vision, just like in a person.

It is important to note that Cognitive Navigation does not require any additional infrastructure for its operation (wify beacons, RFID tags, etc.).