This AI system takes on potholes and other road defects

Researchers with the University of Waterloo have developed a new system that uses artificial intelligence to deal with potholes and other road issues. Though robots aren't out fixing these damaged spots (yet), machine intelligence is helping humans identify the defects using images gathered by cameras mounted on vehicles. Rather than manually scanning roads for issues, cities can use the system to automatically find and flag these spots.

Potholes and cracks are two very common defects on roads and bridges that require repair, particularly following a winter season with freeze and thaw cycles. Though large issues are commonly found and fixed readily due to complaints from drivers, many smaller issues may be overlooked until they grow in size, increasing the cost to fix them.

In small cities, these road defects are typically found by sending manned crews out to drive around and search for issues. Larger cities typically send out vehicles with cameras to record roads, passing the data off to human analysts who manually sort through it, flagging problems. Waterloo's automated system could replace the human team in the latter example, doing the hard work at a more rapid pace.

That's good news for cities that may save money and more efficiently repair small issues before they grow in size. Doing so will reduce costs, therefore reducing the amount of taxes needed to cover the repairs. Researchers say their system is, at minimum, comparable in accuracy to human analysts while being capable of more frequent monitoring.

The system doesn't require any special cameras; in fact, the university says that cities could mount cellphones on city vehicles and use the data gathered during routine operations to feed the AI. The system may also be suitable for evaluating bridges using images gathered by drones.

SOURCE: University of Waterloo