Riding an electric scooter on the sidewalk is dangerous, but scooter-sharing companies have struggled to prevent their customers from engaging in risky behavior. Bird thinks it has found a solution: annoying beeps.

Using technology provided by a company called u-blox, Bird is rolling out an end-to-end GPS system “designed to deliver centimeter-level accuracy specifically for the micromobility industry.” Riders who stray onto the sidewalk will be bombarded with audio alerts from the scooter, as well as notifications on their smartphones, warning them to get back on the street. Failure to follow directions results in a de-powered scooter: the vehicle will slow down and eventually come to a stop.

The problem Bird is trying to solve is a seemingly intractable one. Riders have a difficult time tracking down available scooters when they want one. And scooters sometimes block sidewalks, obstructing the path for people in wheelchairs and other pedestrians with mobility concerns. They all end up cluttered in a handful of places rather than spread evenly around a city. And cities have complained about the companies failing to place enough scooters in low-income and minority communities to ensure equal distribution across economic lines.

Lime says it uses “a sophisticated statistical model” with 95 percent accuracy to determine when a scooter has gone up onto a sidewalk. Other scooter companies have invested in remote-operation technology, including cameras and other sensors, to improve fleet management. These front- and rear-facing built-in cameras enable remote operators to move the scooters when they happen to be blocking sidewalks or street traffic. They can also make it possible for an e-scooter to travel several blocks to riders.

But Bird claims that its GPS-based solution is easier and cheaper to roll out than expensive camera-based systems. It also says that u-blox’s ZED-F9R module (great name) will make it easier for customers to find nearby scooters, even in cities with tall buildings that typically obstruct satellite signals. This is how it works:

The ZED-F9R is a dual-band multi-constellation GNSS receiver that supports up to 8 times more satellite signal types and 4 times more constellations (GPS, Galileo, GLONASS, and BeiDou) than standard solutions. The module processes real-time vehicle data including wheel speed, IMU sensor data including acceleration and spatial orientation, and real-time kinematic data that corrects for ionospheric interference. The technology is also optimized for e-scooters by applying dynamic models matching the movements of the vehicle.

Bird is also using what it calls “centimeter-level sidewalk mapping” to determine when a scooter has left the road and onto a sidewalk. This allows the company to create a geofence that deactivates the scooter’s throttle when it enters a prohibited zone. Centimeter level, you say? How?

Step 1: It starts with a geofence outline constructed from satellite imagery or city GIS data.

Step 2: From here, we use surveying equipment to measure the location of three city landmarks. Only a few measurements are needed for each city.

Step 3: Once the landmarks have been identified, we compare their location to the satellite imagery to determine offsets and rotations.

Step 4: We then use these offsets and rotation values to shift and transform each of our original geofence outlines.

Step 5: Finally, after our geofence outlines have been updated, they are pre-loaded onto our vehicles to eliminate latency.

Bird says it will be testing the new system in scooters deployed in Milwaukee and San Diego, with Madrid to soon follow. If all goes well, the company plans to roll it out to the rest of the global fleet “soon.”