Electric scooters also engage in “visual recognition”, and a Swedish company uses AI to prevent “scrolling” phenomenon

For a long time, electric scooters have received a large number of complaints due to irregular sidewalk riding and parking chaos. The use of artificial intelligence technology can detect when electric scooters leave the road and drive on the sidewalk, and whether the vehicle is parked correctly, which allows AI companies Saw the opportunity.

According to foreign media reports, the Swedish shared electric scooter company Voi Technology is launching the world’s first large-scale electric scooter computer vision pilot in Northampton, UK.

Voi worked with an Irish technology company called Luna Systems to use real-time lane segmentation (similar to technology in high-end cars) to determine whether the rider was driving in compliance with local traffic regulations.

The project has been going on for 6 months. Voi has integrated a smart camera system and computer vision algorithms trained on several hours of Northampton video clips. In addition to keeping riders away from the sidewalk and parking properly, the Voi/Luna collaboration will also convey real-time data on how vehicles are used, using a series of AI tools to identify problem areas and adjust them.

The pilot program is carried out in two phases: In the first phase, a controlled team will test computer vision technology and collect more shots for the algorithm to be trained. The technology is designed to correctly detect the riding surface and emit when driving on the sidewalk. Audible alarm; the second phase will install approximately 100 cameras on the public fleet of Voi electric scooters in Northampton.

Voi also plans to test the automatic deceleration mechanism. If the rider is detected to be driving on an incorrect surface or in a densely packed sidewalk area, the electric scooter will slow down on its own.

Voi reports that these data mainly record the riding method and location of electric scooters, and will not disclose the privacy of the rider’s name, contact information, personal identity, etc. And, these data will be shared with Northampton City Council.

“It is undeniable that the trial version does look a bit cumbersome,” but Luna Systems stated that they hope to integrate the device directly into the handlebars of electric scooters by 2022.

“With the help of computer vision, electric scooters can be trained to see and recognize dangerous situations,” said Fredrik Hjelm, CEO of Voi Technology. This will set new safety standards for the new mode of transportation of electric scooters.

“After helping riders ride more than 60 million times in Europe, we have a deep understanding of the issues involved in the safety of electric scooters and are always looking for ways to do better.”

Leave a Reply