Summary The device, called EyeDAR, is intended to give autonomous vehicles an additional view of surrounding traffic beyond their direct line of sight, according to Rice University postdoctoral researcher Kun
LOS ANGELES (Reuters) – Researchers at Rice University have developed a small roadside radar sensor designed to help self-driving cars detect vehicles, cyclists and pedestrians that may be obscured from a vehicle's onboard sensors at intersections or in poor visibility.
The device, called EyeDAR, is intended to give autonomous vehicles an additional view of surrounding traffic beyond their direct line of sight, according to Rice University postdoctoral researcher Kun Woo Cho.
"This helps capture reflections that a single vehicle would miss and effectively eliminates those blind spots. So instead of each car sensing alone, we are conceptually creating these distributed sensing system that is more reliable and scalable," said Cho.
Autonomous vehicles typically rely on a mix of lidar, radar and cameras to perceive their surroundings, though cameras can struggle in poor visibility such as fog, rain or glare.
Cho said EyeDAR is a low-power millimeter-wave length radar sensor about the size of an orange that could be mounted on infrastructure such as streetlights, traffic lights and stop signs.
The roadside unit sends out radar signals, uses the reflections to build a picture of nearby traffic, and alerts nearby autonomous vehicles to approaching cars, cyclists or pedestrians that may be hidden from view. It does this using a 3D-printed lens and antennas that help detect the direction of radar signals while reducing the processing power typically needed.
"EyeDAR is what we call a talking sensor, it not only detects objects, but also sends this information back to vehicles using the incoming radar signals arriving at the sensor," Cho said.
She added that tests showed the device could pinpoint the direction of incoming radar signals more than 200 times faster than traditional radar designs, which could help self-driving cars identify hidden hazards more quickly.
The research comes as self-driving technology moves gradually from trials into commercial deployment in some parts of the world. But wider adoption remains constrained by concerns over safety, regulation, cost and the challenge of making autonomous systems work reliably in dense, unpredictable urban environments.
