Cognitive smart situation capture Technology of understanding of a road situation

Cognitive smart situation capture is technology of understanding of a road situation from Cognitive Technologies.

2018: Announcement

The Cognitive Technologies company, the Russian AI developer for pilotless vehicles, in January, 2018 provided technology of understanding of a road situation of Cognitive smart situation capture.

Cognitive smart situation capture allows not only to detect objects of a road scene, but also to do the exact forecast of development of a road situation for all participants of a road scene, to create scenarios of the safe movement including in critical road situations — the Head of Department of development of pilotless vehicles Yury Minkin told. — It is obligatory control in order that driving in the pilotless mode was safe.

The case when in lack of visible interferences for the movement passing vehicles from the right or left side begin to reduce speed can be an example of work of Cognitive smart situation capture. A system at the same time understands that emergence on the road of a dangerous object can be the cause of reduction of speed (for example, the old woman passing the route in the unlawful place). Therefore a system needs to reduce speed, in addition to browse the necessary sector to be convinced, really someone crosses the road or not, etc. And if really the person crosses the road, then a system undertakes safe braking, explained in Cognitive Technologies.

In the whole Cognitive smart situation capture allows to manage safely in unforeseen situations which are not solved "bluntly", Yury Minkin added:

We learned to define areas to which this or that object of a road scene can (or cannot) move and with what probability and also dangerous and safe areas for maneuver.

These zones, according to him, are calculated also proceeding from the current physical characteristics of an object (a trajectory of the movement, linear and angular speeds, acceleration, etc.). At the same time natural restrictions of physical laws and opportunities of an object (for example, the person will not be able to run with a speed of 60 km/h, and the car cannot go "sideways", etc.) and also other signs, indirectly confirmatory intentions of this or that participant of a road situation (included in a signal of turn of the car the pedestrian's hand raised up, the switched-on alarm signaling, etc. are considered).

On the basis of such forecasts specific scenarios of behavior of the unmanned vehicle can be under construction. According to experts, the solution will allow to increase traffic safety to 30%.^[1]

Notes