RoadBotics uses neural networks to monitor the state of urban roads.

RoadBotics uses neural networks to monitor the state of urban roads.

January 28, 2019 0 By autotimesnews

Startup RoadBotics monitors road conditions using machine learning and image recognition algorithms. This allows you to monitor the state of the roadway almost non-stop.

As the engineers explain, at present, road maintenance costs the local authorities a pretty penny. In addition to the fact that weather conditions and vibrations from the movement of cars naturally destroy asphalt, salts, fuel and water, penetrating into the cracks formed, over time, lead to even more serious damage that is more difficult, longer and more expensive to repair.

“If repairs are made 5 or even 10 years before the damage becomes very serious, the life of the road will increase significantly,” emphasizes Benjamin Schmidt, technical director of RoadBotics.

To assess the condition of the track, RoadBotics sends a car with a smartphone attached to the windshield that captures the condition of the canvas, fixes the location and sends this data to the cloud, where the algorithms analyze each shot frame by pixel. Engineers claim that neural networks are capable of recognizing not only deep holes, but also small cracks, irregularities and depressions.

The results of the analysis are displayed in the form of a detailed virtual map, on which parts of the road with a length of 3 meters are marked with different colors. Green color means a normal condition of a paving, red – critical. Based on this information, officials can subsequently make an informed decision about the need for maintenance of certain road sections and the allocation of funds for their repair.

It is reported that since its inception in December 2016, a startup has evaluated the road surface for 90 cities, towns and districts throughout the United States.

“The Roadbotics solution saves precious time and money for local authorities. So, it took three years for the authorities to assess the condition of all 1,100 km of the roadway in the city, in Savannah, Georgia, authorities. This procedure cost an impressive $ 130,000, while the results were unreliable and fragmented (each driver who participated in the study, in its own way, assessed the condition of the road).

RoadBotics provided the city with an objective rating of pavement in just three months and a third of the cost, ”writes IEEE Spectrum.

Finally, we note that in the future RoadBotics intends to use its work to assess the condition of other urban assets, in particular, power lines, advertising banners, green areas and road lighting.