LTU Student Researchers Make E-scooters Safer

GEKOT President Dan Champoux

E-scooters are everywhere these days, including places where it's not exactly safe for them to zip along at top speed--like sidewalks. Many municipalities have banned riding them on sidewalks, due to the risk of accidents with pedestrians.

So how do you design a scooter that can sense when it's traveling on a sidewalk--triggering an alarm to the rider, and a reduction in the scooter's top speed?

Two Lawrence Technological University students, working with a scooter company that originated as an elementary school project, have figured out a way to do it that's cost-effective.

The "sidewalk detection" research project was multidisciplinary, originating in the Department of Electrical and Computer Engineering within the LTU College of Engineering, and involving the Department of Mathematics and Computer Science from the LTU College of Arts and Sciences. The student researchers were N. Maxine Kennebrew, who earned a Master of Science in Computer Science from LTU in August 2023, and Jarin Kurashige, a senior electrical and computer engineering student.

"The nice thing is, we created a project that's not only multidisciplinary and multi-department, but multi-level, involving a graduate student and an undergraduate student," said George Pappas, assistant professor of electrical and computer engineering and director of LTU's Master of Science in Artificial Intelligence program.

Maxine Kennebrew, a 2023 Master of Science in Computer Science graduate

The researchers worked with GEKOT Robotics Inc., a Bloomfield Hills company that grew out of a group of students at East Hills Elementary School in the Bloomfield Hills district and their design for a collision avoidance system for e-scooters, originally developed for a robotics competition in 2019. Kennebrew said she applied for an internship at GEKOT—an acronym for Great Engineering Kids of Tomorrow—and she and Pappas developed the idea for the sidewalk detection project that could be commercialized into another scooter product for GEKOT to sell.

Kennebrew said cameras have been used as sidewalk detection systems, but they're prone to vandalism--and they're expensive, especially when it comes to replacing cameras lost to that vandalism.

Instead, Kennebrew and Kurashige reasoned that sidewalks create different vibrations in a scooter than asphalt or concrete roadways and bike paths. So, they could use an inexpensive sensor—an accelerometer and gyroscope that cost about $4—to detect differences in vibration picked up by sensors on the scooter. Then, Kennebrew used her computer science skills to develop an artificial intelligence algorithm that could detect whether the scooter was on a sidewalk or not.

"We're picking up 200 readings a second, and we're putting them through the model and getting an output of sidewalk or not-sidewalk," Kennebrew said.

The two began working on the project in January 2023, and research continued through the summer, including one heck of a lot of riding on a scooter to gather data on vibration differences. Kennebrew said she personally rode about 30 hours on a sensor-equipped e-scooter over the summer, and brought one home and had teenagers in her neighborhood ride it to gather more data. "We wound up with about 130 hours of data at 200 readings a second," she said.

Students Maxine Kennebrew, a 2023 Master of Science in Computer Science graduate, and Jarin Kurashige, a senior electrical and computer engineering student

The computer involved is called an ESP32, a low-cost, low-power system with sufficient processing speed and memory to handle the incoming data and process it through the algorithm.

Kennebrew and Kurashige say they believe this low-cost system can be scaled and built into e-scooters to help their manufacturers comply with local ordinances restricting sidewalk travel, as well as enforce rider compliance by slowing down the scooter while it is traveling on sidewalks and warning a rider that they're violating local ordinances.

"This will be able to be embedded in the deck of the scooter, below the rider's feet," Kennebrew said. "We hope to have a demonstration vehicle by the end of January."

Kurashige was responsible for designing and building the firmware and hardware for the sidewalk detector, while Kennebrew was responsible for the algorithm.

The project also led to a job for Kennebrew. In September, she became GEKOT's lead engineer, one of seven employees at the company. In this role, she also supervises four interns.

Maxine Kennebrew, a 2023 Master of Science in Computer Science graduate

GEKOT and Lawrence Tech's Centrepolis Accelerator funded the sidewalk detection project with approximately $30,000. The prototype sidewalk detection unit runs on an Arduino microcomputer system.

Founded in 2020, GEKOT has attracted $480,000 in startup funding from the Michigan Economic Development Corp., Centrepolis Accelerator, private investors, and the scooter manufacturer Razr.

In a note sent to Pappas during the project, Kennebrew described it as "one of the hardest school experiences I've had because I started from a place of not knowing anything... I had to learn about hardware and embedded software and circuitry and many other engineering principles unfamiliar to me. I am most grateful that I was given both guidance and autonomy to explore solutions, and the space to fail and regroup when things didn't quite work out the way I thought they would. I gained so much confidence... that I was able to continue my work, be presented with new challenges, and not at all be intimidated... Most importantly, I have been able to start mentoring students in various stages of their academic careers by passing along what I learned from this experience, which is that we learn the most from doing, and we can't be scared to tackle a problem."

And Kurashige, a native of Hawaii, who heard about LTU through its varsity bowling program, said: “The most interesting thing was, this was one of the first times that I put all my book learning, all my class work, into a real world solution that a company will use. That is the coolest part of what I was creating.”

GEKOT technology, now patented, uses several automotive sensors and applies them to the micro-mobility market, changing the way that e-scooter and e-bike operators' ride. With the guidance of GEKOT President Dan Champoux, an automotive engineer, GEKOT continues to develop safety technology, seeking ways to reduce collisions, prevent injuries, and create safer streets.

"I got to meet the GEKOT kids," Kennebrew said. "They're like 12 or 13 years old now, and they're great."