Bird vehicles can now accurately detect sidewalk riding and slow riders to a stop

Shared micromobility operator Bird is getting on the scooter ADAS bandwagon. After three years of research, the company has finally launched new technology that can detect when a rider is scooting on the sidewalk and slowly bring them to a stop.

The new tech is currently being piloted in hundreds of scooters in Milwaukee and San Diego, and is expected to reach Madrid early next year, as well as other cities around the world in the coming months. Scott Rushforth, chief vehicle officer at Bird, said all new vehicles are being built with the tech, and that we can expect “tens or hundreds of thousands” of sidewalk detection-equipped vehicles to come off the assembly line between now and the beginning of next year.

The new feature, designed to address the most annoying and dangerous shared micromobility concern of cities everywhere, is made possible through a partnership with u-blox, a Swiss company that produces wireless semiconductors and high-precision positioning modules. Together, the two companies co-developed a unique version of u-blox’s ZED-F9R module, tailored specifically to meet the needs of the shared micromobility industry, according to Bird.

“It takes input from the GPS sensors, and it uses a dual-band GPS sensor, which is the best of the best in terms of GPS,” Rushforth told TechCrunch. “What we’ve added on top of that is a system called RTK, which stands for real-time kinematics. And then we’ve added on top of that a system that uses sensor fusion that takes all this data together as well as data from the vehicle itself, like how far the wheel has gone or the angle the scooter is leaning and infuses it with the GPS location so that it actually gets an extremely accurate view of where the vehicle is, even in times where the GPS signal does not work very well.”

Riders on the sidewalk will receive audio and visual alerts of their transgressions via the mobile application and a new 16-bit color display before the scooter removes its throttle and brings itself to a smooth stop.

Bird says it’s been exploring different sidewalk detection technology since 2019, and it’s not at all alone. In the world of micromobility rider-assistance systems, there appears to be two camps: The first camp is populated by companies that rely on super precise positioning and sensor fusion to detect poor riding behavior and correct it in real time. Superpedestrian recently acquired Navmatic to put the company’s positioning software to use in a similar manner, which will be built into every new LINK scooter by December.

In the second camp, we have the likes of Spin, Voi and, more recently, Helbiz, that are working with startups like Drover AI and Luna to attach cameras that can detect things like sidewalks, bike lanes and pedestrians.

“It’s a very good thing that there is now a micromobility race to the top on safety,” a Superpedestrian spokesperson told TechCrunch. “Bird has now reached the same conclusion we did: Vision has its limitations both in terms of the unsafe behavior it can detect, and how scalable it is.”

Computer vision companies have often argued that location-based scooter ADAS is useless in areas without GPS, like urban canyons or underground parking garages, but Rushforth disagrees. He says u-blox’s module has advanced dead reckoning, which is basically a process of calculating the current position of a moving object by using a previously determined position. GPS would likely be the starting point, so if the vehicle loses satellite reception completely, it can still rely on other sensors to determine how far in any direction it’s gone.

The module processes data like wheel speed, acceleration, spatial orientation and kinematics and fuses it together to come up with a “scary accurate” location and centimeter-level mapping, according to Bird. That information is passed via the vehicle’s circuit board to Bird OS, the company’s proprietary operating system, which then makes the decision of what to do with the data.

This not only helps Bird vehicles stay precisely within city-determined geofenced areas, but generally having high levels of position accuracy help with a range of adjacent features.

“If you’re in an area where the GPS by itself, without all this added technology, is kind of just OK, you might be outside the parking corral and the vehicle might think it’s inside, so this higher precision will absolutely improve the experience for parking,” said Rushforth.

Operations, which accounts for one of the largest costs of micromobility companies, could also see huge benefits from knowing exactly where vehicles are by knocking down time spent searching for a vehicle that might be hidden behind a trashcan or actually on the southwest corner when the GPS says it’s on the northwest.

“I would say our whole business is going to get better in measurable ways across the board, because so much of it relies on knowing exactly where the vehicles are,” said Rushforth.

Bird said it has explored several other solutions, from camera-based to ultra-wideband, which is a form of radio technology that uses short-range, high-bandwidth communications across a radio spectrum, but found this to be the best option for scaling.

“Now we have a solution that we don’t have to train very much,” said Rushforth. “All we have to do is go in our back end and show where the sidewalks are, and then we can scale this very quickly, and we’re only spending probably about $10 or $12 more per vehicle.”

Camera-based solutions, on the other hand, could cost operators around $200 per vehicle for the hardware and services to start, which could scale down to $80 per vehicle over time, but that’s still a pretty penny for a business that hasn’t yet figured out how to be profitable.

“If I had a crystal ball, I would say that everyone will copy us with this over the next 24 months because this is the only way to do it in a cost-effective manner,” said Rushforth.

This article has been updated to include more information and commentary from Superpedestrian.