It’s one common scenario in car crashes: a pedestrian approaches an intersection, hidden from an oncoming driver by a building or trees, and steps off the curb. The driver can’t stop in time. It’s a problem that new sensors on vehicles can’t avoid, since someone behind a building isn’t any more visible to a sensor than to the person behind the wheel.
New technology now in development aims to use mobile phones to help. The tech uses new AI—designed to be embedded in phones and cars—to predict an accident before it happens. “We have a mathematical model that resides in the mobile phone that has the ability to predict human behavior several seconds ahead,” says Gabi Ofir, CEO of Visiblezone, the Israel-based startup building the technology. If the pedestrian is in danger, the phone immediately starts to transmit a beacon to the surrounding area; an approaching car receives the signal, uses more artificial intelligence to determine risk, and if an accident is likely, the car pings the driver with an alert.
“Our technology utilizes the fact that most pedestrians are carrying a mobile phone,” Ofir says. Ironically, phones can directly cause accidents; in one survey, nearly a third of adult drivers admitting to texting and emailing while driving, and younger drivers were even more likely to text. But the ubiquity of phones could make them a useful tool in a warning system. The company’s vision isn’t to offer an app, but to donate the software to phone manufacturers to embed in their firmware so it’s available to everyone. Car manufacturers will pay to install the corresponding software in the operating system in cars. The technology doesn’t require any changes in hardware, so it could be rolled out quickly. It also works without a cellular connection—by directly beaming a signal between the phone and the car—a critical feature in areas without reception or with overloaded networks that cause delays.
Some cities are beginning to install cameras at particularly dangerous intersections that can perform the same function of warning drivers if cars have corresponding technology inside. But these exist only in a handful of locations, and a system that works with phones could work anywhere, from rural roads to streets in towns with no budget for new infrastructure. The warning system also works whenever visibility is low, such as at night or in fog or snow. It’s only useful if drivers heed the warnings, so the developers carefully worked to build algorithms that only send an alert if it’s actually necessary. “We don’t want to annoy drivers with an alarm that’s not relevant for them,” says Ofir.
It’s only one piece of a solution for eliminating accidents, since pedestrians are also more likely to be hit by cars when streets are badly designed. But it could be helpful as the rate of pedestrian deaths continues to climb. In the U.S., an estimated 6,627 pedestrians were killed by drivers last year—roughly 35% more than a decade ago. Even in many cities with “Vision Zero” goals of eliminating traffic deaths, the number of accidents continues to grow. Ofir believes that the tech can fill one gap in safety. “We are tackling maybe the most complicated issue in the automotive domain: human behavior near roads,” he says. “This is a really big enigma.” The startup has a working prototype now and is working on its next proof of concept.