Could Autonomous E-Scooters Become a Reality?

Peter, Ben and Richard Parker spend a lot of time in the University of Maryland’s E.A. Fernandez IDEA Factory, a hub for interdisciplinary innovation, but sometimes you’ll catch them cruising around campus.

The Parkers are neither siblings nor students. They’re self-driving electric scooters built in the Maryland Robotics Center that could change the face of urban mobility.

The Parkers’ surname is a nod to their ability to autonomously park themselves back at a hub. (Peter and Ben Parker are “Spider-Man” characters while Richard is from “Life of Pi.”) But whimsical monikers notwithstanding, their technology is no joke.

Should the UMD models go to market, scooters cluttering sidewalks could become a thing of the past: Passengers could drop off these e-scooters anywhere without blocking pathways or amassing financial penalties from rideshare companies. Hurried travelers could summon them to their homes, eliminating schleps to parking docks. For train commuters, imagine a scooter puttering below tracks to arrive at preferred platforms. The prototypes move less than five mph in self-driving mode but otherwise reach 12 mph speeds.

Designed for rideshare companies and funded by a University of Maryland Grand Challenges Grant, the scooters have been tested during several cross-campus missions, the results of which were recently featured in IEEE Transactions on Intelligent Vehicles. Though the technology is proven, the challenge for commercialization is their high cost: The prototypes are valued at $2,000. Vandalism is also a concern. Still, the researchers have engaged with private vendors.

“Self-driving isn’t just for cars; it can be for scooters, too,” said Derek Paley, the Willis H. Young Jr. Professor of Aerospace Engineering Education and director of the Robotics Center who built the autonomous contraptions with his students. “Imagine an e-scooter that’s always available where you need it, when you need it.”

The off-the-rack scooters are customized with autonomous software and sensors. Each unit employs a GPS module, steering-column motor and Robot Operating System computer that directs it to travel preplanned paths. A depth-sensing stereo camera allows it to identify short-distance targets like bike racks and park safely.

An expert in autonomy, Paley works closely with the Department of Defense to support military robotic systems, from drones to mechanical dogs. He’s been an e-vehicle devotee for more than a decade, having graduated from big-wheel transporters like early Segways to electric unicycles. During one commute, he wondered if autonomy were possible.

Generally, the average scooter trip in the United States is one mile, completed in about 10 minutes, said Paley. For an e-scooter rideshare company to be profitable, each unit in its fleet must be used three or four times per day. By introducing autonomous scooters, companies could add one to two trips because customers could collect them more quickly, Paley estimated.

There are other benefits. During off-peak hours, self-driving scooters could motor to designated spots for bulk collection by service providers prior to charging and redistribution. Shorter collection routes would decrease carbon emissions and lower logistics costs.

In addition, company-imposed parking restrictions could be eliminated. “Sometimes the scooter won’t let you park, and you have to drive around a bunch of buildings to find a dock,” said Paley. “A self-driving scooter would have none of those issues.”

[Autonomous Cars Don’t Understand How Blind People Move Around. A Research Team Is Trying to Boost Safety.]

After earning a National Science Foundation Grant, he and his students interviewed more than 100 professionals, from rideshare operators to elected officials tasked with fielding pedestrian complaints of scooter blockades. They began building the Parker models shortly before the pandemic.

Though e-scooter use is on the rise and gaining support from city governments, some officials question the near-term feasibility of self-driving models, citing safety concerns.

“The development of such autonomous systems is a difficult problem,” acknowledged Dinesh Manocha, Distinguished University Professor; Paul Chrisman Iribe Professor of Computer Science and Electrical & Computer Engineering at UMD, who has worked extensively on autonomous robot platforms. Still, he said Paley’s team “has been doing amazing work.”

For his part, Paley said autonomous scooters are a lot more practical than autonomous cars in urban environments.

“It’s a lot safer to have a 50-pound self-driving vehicle going 5 mph than a 2,000-pound self-driving vehicle going 50 mph,” he said.