It may sound weird to you, but government doesn‘t know about all the potholes in its cities. You may be cursing driving over the same pothole every morning, asking yourself, why no one has fixed it yet, but chances are no one knows about it. But now researchers at the University of Waterloo developed an AI system, which can automatically analyse images from vehicle-mounted cameras and detect potholes, cracks and other defects in the road surface.
Governments definitely don’t know everything, but they still have to make decisions. They have to see which streets have to be fixed first, what has to be done to them, how quickly they need attention. They also have to detect new potholes before they start causing significant damage to the vehicles, because people will be complaining and maybe even suing government for compensation. AI system could make this job easier and could even help reducing road taxes and make streets safer. AI is just a much superior way to current methods.
Currently governments around the world use to methods to detect defects in the road surface – direct observation, when workers are simply driving around visually inspecting the road, and cameras fitted to trucks. In any case, after something is detected, the area has to be inspected manually. And so a human error can be easily introduced – less experienced workers may disregard some cracks that already need to be filled. AI system should be as accurate (if not more accurate), allowing more timely repairs and consistent results. Scientists started testing this system with Google Street View images and later applied it to images from other sources, such as vehicle-mounted cameras. Scientists say that AI would work perfectly with the current method of workers driving around – it would simply analyse the images.
Cameras could be mounted in various vehicles that are already driving around, such as buses, garbage trucks, utility vehicles and so on. Scientists are already thinking about other applications as well. For example, AI system could analyse drone images to inspect the condition of buildings, bridges, infrastructural objects. It could also monitor the progress of construction projects. John Zelek, an engineering professor, one of the creators of this AI system, said: “If the parts of a new building aren’t coming together properly, it would obviously be very beneficial to flag problems before the next 30 storeys have been put up”.
If it works as promised, results would be fantastic. Roads would be fixed before they get too bad, making repair simpler. It would also reduce the cost of road maintenance, which could also result in lower taxes.
Source: University of Waterloo