Radio waves could help driverless cars see around corners

Get the Popular Science daily newsletter💡

Breakthroughs, discoveries, and DIY tips sent six days a week.

In late January, an Alphabet-owned Waymo self-driving car was cruising near an elementary school in Santa Monica, California, when a young child suddenly darted into the street. Waymo’s LiDAR sensors detected the student, who had just emerged from behind a parked SUV, but it was too late. Despite slamming on the brakes and slowing from 17 to six mph, the driverless car struck the child, knocking them to the pavement. Luckily, reports show that the child only suffered minor injuries, but that’s likely little comfort to parents whose children live in the growing number of cities where driverless cars operate.

In this case, the Waymo detected the child once they came into view—but what if it could have “seen” them from around the corner? That is the general idea behind new research emerging out of the University of Pennsylvania, where a team of engineers have developed a sensor system that uses radio waves to help robots detect objects (or people) hidden behind walls.

In testing, the HoloRadar system was mounted on small mobile robots and successfully identified people through walls while roaming around campus. Though the research is still in its early stages, the team is optimistic that its sensor system could be integrated into robots to help them navigate their surroundings more effectively. For self-driving cars, it just might make the difference in preventing a future collision.
“Robots and autonomous vehicles need to see beyond what’s directly in front of them,” study senior author and Computer and Information Science professor Mingmin Zhao said in a University of Pennsylvania blog post. “This is an important step toward giving robots a more complete understanding of their surroundings,” Zhao added.

Robots That Can See Around Corners Using Radio Signals and AI

Using radio to ‘see’ behind walls

Though there are some notable exceptions (Tesla chief among them), most emerging self-driving vehicles rely on a combination of cameras, radar, and LiDAR sensors to “see” the world around them. LiDAR systems emit millions of laser pulses in all directions and measure how quickly those pulses bounce back. The systems then use that data to almost instantaneously generate a highly detailed 3D map of the vehicle’s surroundings. 

But LiDAR has limitations. Large buildings or other obstacles can block its laser pulses, preventing it from detecting what is hidden around a corner. To try and solve that problem, Zhao and his colleagues decided to explore an approach largely abandoned by autonomous vehicle (AV) companies: radio waves. Researchers had previously avoided radio signals because they produce much longer wavelengths than visible light, resulting in lower resolution and less clarity when detecting objects. Image clarity, in that case, is crucial to figuring out if that blob in the street is a floating plastic bag or a scurrying pet. The team from Penn eventually realized that this apparent limitation could actually be an advantage when it comes to LiDAR seeing around corners.

Radio waves emitted by a robot that strike a wall or other surface are much larger than the small surface variations on that wall. As a result, only part of the radio signal continues past the barrier, while a significant portion is reflected back toward its source. The researchers explain that this effect essentially causes surfaces to behave like mirrors, reflecting radio signals in predictable patterns.

“It’s similar to how human drivers sometimes rely on mirrors stationed at blind intersections,” study co-author and Pennsylvania University doctoral student Haowen Lai said in a statement. “Because HoloRadar uses radio waves, the environment itself becomes full of mirrors, without actually having to change the environment.”

To sift through all that data, the researchers developed a two-part AI model that first filters out unnecessary information and makes informed assumptions about what is hidden around the corner. The system then reconstructs a 3D model, providing it with a a visual representation of what lies behind the wall. Finally, it was time to put the system to the test by mounting it on a mobile robot and sending it to roam around campus. It navigated three distinct corners and consistently detected objects and researchers hidden from view.

Can radio waves make driverless cars safer? 

Earlier robotics research has attempted to solve the problem of seeing around corners, albeit with limited success. In 2019, a team from MIT used a camera mounted on top of a self-driving car to scan for shadows in the vehicle’s path. In theory, the system could detect the shadow of an approaching car at the far end of a parking garage and infer its presence. More recently, engineers at Stony Brook University used single-photon LiDAR to detect traces of light even after it bounced around corners. The researchers noted that this approach was inspired by human drivers who rely on convex mirrors at blind intersections to see beyond their immediate field of view.

Enhancing Autonomous Navigation by Imaging Hidden Objects using Single-Photon LiDAR

However, the problem with both of those and other approaches is that they fundamentally rely on light waves to see. That means they simply won’t work as well in mixed lighting conditions or in darkness. HoloRadar sidesteps those limitations by using radio waves. That said, it isn’t a one-stop solution. The Penn researchers make it clear that HoloRadar is intended to supplement, rather than replace, the array of sensor systems already used on robots and autonomous vehicles.

For better or worse, self-driving vehicles are becoming an increasingly common presence on city streets around the world. Accidents, and at least a dozen deaths in the United States , have already occurred, a figure that is almost certain to rise as their rollout accelerates. Equipping these robots on wheels with the ability to “see” through walls could offer an alternative.

 

2025 PopSci Best of What’s New

The 50 most important innovations of the year