MAC monitoring

2022: Implementation of the hardware and software complex

In Tyumen, artificial intelligence (AI) began to be used to examine the quality of roads. This was reported in the press center of the regional traffic police at the end of October 2022.

We are talking about the hardware and software complex "MAC Monitoring," which uses AI in order to identify potholes and cracks on the roads, as well as deviations in markings and signs.

In the Tyumen region, a hardware and software complex was introduced to find holes on the roads

According to the Main Directorate of Construction of the Tyumen Region, a video camera with stabilization and a smart unit receiving signals from it are installed in the control car. Thanks to the video capture, the microcomputer at 60 frames per second recognizes deviations on the road network in the video. Further, the photos with deviations fall on the hard disk, after which the GPS unit assigns coordinates to the defects, and the LTE module sends the corresponding cards to the dispatcher, who is already in contact with the contractors to eliminate the defects. The latter, having worked out the problem, takes a new photo and closes the task.

It took developers about 2 years to train the neural network. Now AI runs each frame through about 50 thousand similar images. Some elements, like communal hatches, which the neural network could previously recognize as a pothole, had to be "retrained." There was an interesting experience with road signs, instead of which artificial intelligence accepted some advertising banners in the distance.

In the future, it is possible that artificial intelligence will teach and calculate the cost of work on the restoration of the road surface and reporting. At the same time, next winter neural networks will have to learn how to find deviations on the barrier fence and determine ice and snowdrifts.^[1]

Notes

Шаблон:Remarks