21 Jan '22
21 Jan '22
“We have created our own set of algorithms, an approach to object recognition and classification and routing. Our control system generates 2048 trajectories forward 6.4 seconds every 0.05 seconds. At the same time, it takes into account static and dynamic obstacles surrounding the vehicle in the range of 360 degrees at a distance of 220 m. Thus, it can be argued that, unlike a human, such a heavy-duty vehicle sees everything and plans the movement, taking into account all factors, ”explains the head Laboratories of autonomous transport systems Salimzhan Gafurov.During scientific research, IT university specialists developed mathematical models of the main modules of the system, formed the appearance and design of a quick-detachable module with aerial reconnaissance. Experts have found that off-road navigation can be carried out without the help of accurate maps, for this you can only use the front lidar and work on maps provided by the UAV.
“Within the framework of this project, the employees of our laboratory have developed and manufactured an aerial reconnaissance platform. The unmanned aerial vehicle can take off and land vertically on this platform, and it is stored and charged there. The platform is located in the body and allows the UAV to operate in a temperature range from -40 to +60 degrees. The drone conducts aerial reconnaissance and reports the traffic situation to the unmanned vehicle, on the basis of this information Kamaz and builds trajectories, ”says the head of the Laboratory of Autonomous Transport Systems.Scientists conducted experiments and obtained results that fully provide conformity of a production truck with an automatic transmission to the third level of the ADAS driver assistance system classification. This correspondence means conditional automation, when the driver does not need an instant reaction to situations arising during the movement: he must be ready to intervene within a limited time. According to the developers, there is no clear requirement for this time of human intervention, it depends on the scenario, because driving in courtyards requires more care than on the motorway.
Salimzhan Gafurov: “That is why we are not planning to test Kamaz on highways yet, especially since this is not a new thing - testing at long distances was shown by Mercedes back in 1994, and later by other companies. The most difficult task for heavy vehicles and at the same time the least studied is navigation in the city. And we are focused on this task, now the car has been put into trial operation and is being tested on the roads of Innopolis and has already passed 3 thousand kilometers. ”Before driving on real roads, the developers test the system on their own simulator. Innopolis Simulator works out various scenarios on the road, simulates the movement of traffic and pedestrians and their detection, simulates all the necessary sensors and sensors - radar, lidar, cameras, GPS, IMU.
“The software package helps to avoid critical design errors, fix them at early stages, conduct more prototype tests and save on field tests,” explained Sergey Kopylov, research engineer at the Autonomous Vehicles Laboratory.Experts at Innopolis University believe that despite the fact that the self-driving car is ready for use in real life, and the technology is at a high level of readiness and has proven itself well, there are legal constraints.
“The answers to key questions in the field of liability, insurance, implementation in big cities, infrastructure have not yet been determined. Until we solve them, it is difficult to say when we will see this car in operation, ”explained the head of the Laboratory of Autonomous Transport Systems Salimzhan Gafurov.