Smart Bike Predicts Cars' Trajectories, Honks To Warn of Impending Crash

Onboard sensors and sophisticated algorithms could alert drivers, but what about cost?

3 min read
Image of a new smart bike prototype that can protect itself against collisions.
E-bike loaded with sensors mashes through the streets lot like a boss.
University of Minnesota

Cycling can be fun, but it can also be a dangerous sport. A record 846 bicyclists were killed in motor vehicle crashes in the U.S. in 2019—the second highest number in three decades. And a further 49,000 were injured, according to the latest figures from the National Highway Traffic Safety Administration.

To reverse this trend, a slew of safety measures have sprung up in recent years: more bike lanes and apps to help you find themsmart lights and helmets, devices that send out your coordinates in a crash, just to name a few.

Now, there might be one more to add to that list. A team of engineers from the University of Minnesota have created a new smart bike they claim can protect itself from collisions. The invention was described in a recent publication of IEEE Control Systems.

The prototype pictured above is a specially outfitted e-bike, tracks nearby vehicles using a suite of sensors: a low-density lidar at the front can scan for cars at an upcoming intersection, while two lasers (one at the rear, the other on its left side) keep tabs of cars behind and beside it.

The sensors send data to an onboard microprocessor, which then uses complex algorithms to predict a vehicle’s trajectory. “It calculates the relative velocity of the car to the bicycle,” explains Rajesh Rajamani, a mechanical engineering professor who led the research efforts. If the system determines a crash is imminent, the bike sounds a horn in warning.

The aim is to get errant drivers—rather than the cyclist—to alter their course, says Rajamani. “People who have installed sensors on bicycles typically only try to warn the bicycle rider...but nobody has tried to warn the car driver and alert them to the presence of the bicycle.”

The other novel aspect of the work, he says, is just how sophisticated the bike’s tracking system is. It’s capable of predicting the most common types of bicycle-vehicle collisions—rear-end, rear-sideswipe, right-turn, and left-turn—from up to 30 meters away.

“We’re probably the first research group to track vehicles in a lot of different types of scenarios,” says Rajamani, whose team successfully conducted a field test involving three riders using the bikes on their daily commutes about Minneapolis (you may watch the videos here).

“That’s hard for a couple of reasons,” says Hanumant Singh, a professor of electrical and computer engineering at Northeastern University. “Mostly because it’s vehicles [rather than bikes] that are running these sensors and they’re much more stable. So the researchers have solved some nice problems here.”

It’s also refreshing to see research exploring how sensors can be deployed on bikes, rather than just vehicles, says Frauke Behrendt, an assistant professor at Eindhoven University of Technology in the Netherlands, who has been studying smart cycling for the past decade.

“People often think of cycling as a traditional, offline form of transport,” she says. “But it’s really important that bicycles can also be part of these new innovations around smart technology and the Internet of Things.”

Still, Singh cautions that “there are a number of issues that need to be addressed before the bike goes into the real world.”

For one, will the driver have enough time to react? “With a car and lidar, the advantage of 100 meters is that you have a couple of seconds to react. With a bicycle, that’s less than two-tenths of a second,” says Singh. “Most reaction times of drivers are of the order of two seconds.”

Then there’s the weather to worry about. Most sensors perform poorly in bad weather because the beams they send out “get reflected back and you just get shadows of what you’re expecting to see,” says Singh. “A lidar sucks in rain and snow.”

The new bike hasn’t been tested in adverse weather, admits Rajamani, although all its components are waterproof.

The other big issue is cost. “Most car manufacturers are shuddering at the cost of a lidar because it’s so expensive,” says Singh. “Putting a lidar on a bike...I don’t know about that.”

However, Rajamani says the sensors on his new bike are inexpensive, lightweight ones, with the whole system costing roughly $500. He’s now looking for a licensing partner to take the technology to market, so cyclists can purchase the system to make their bikes safer. He also hopes to adapt his invention for other vulnerable road users, such as e-scooter riders and pedestrians.

“Still, we must think about the affordability and inclusivity of cycling, especially because it’s a really affordable type of mobility,” says Eindhoven’s Behrendt. “Are these technologies developed for wide-ranging types of users who want to, and should, feel safe?”

Plus going high-tech isn’t always the best way to protect cyclists, says Singh. “I’m a sensor guy... but if I were to protect the cyclist, the first thing I would do is just paint the living daylights out of the bike and make it much more visible.”

Behrendt adds: “We also need policy, infrastructure, culture, and the legal context to be very much in favor of cycling.”

Still, she’s encouraged by the new developments. Anything that helps with the “shift towards more active and lighter forms of transportation is going to be helpful for sustainability and making cities more livable,” says Behrendt.

Enhancing cyclist safety is one of those things, she says, because “feeling safe is very important, obviously nobody wants to fear injury or death when getting around.”

The Conversation (0)