A team of electrical engineers from the University of California San Diego have developed a novel stereo RADAR implementation, which they claim, can replace LiDAR at a lower cost while operating perfectly at night and through fog and other atmospheric conditions.
RADAR, which uses bounced radio signals for direction and ranging, is a low-cost and well-proven sensing method but operates at a very low resolution; LiDAR, which uses laser-based light, offers a higher-resolution and more accurate view of surroundings but can't see through fog, rain, snow, dust, and other atmospherics. UCSD's twin-RADAR system, though, is claimed to offer the best of both worlds — effectively a "LiDAR-like RADAR," according to Professor Dinesh Bharadia.
The system, designed for autonomous vehicles, sees two RADAR sensors fitted to the hood of a car at either side. Data from these sensors are then combined to improve resolution and predict 3D geometry of detected objects — effectively creation a stereo-vision RADAR system.
"This is the problem with using a single radar for imaging: It receives just a few points to represent the scene, so the perception is poor. There can be other cars in the environment that you don't see," explains Kshitiz Bansal of the problem the twin-RADAR approach solves. "So if a single radar is causing this blindness, a multi-radar setup will improve perception by increasing the number of points that are reflected back."
In testing, the team found that their RADAR system performed equally as well as an off-the-shelf LiDAR system in sensing during clear conditions; when tested in simulated fog, though, the RADAR system continued to operate where the LiDAR system failed.
The team is presented the work at the Sensys conference this week, but has not yet made the paper available for public dissemination. More information is available on the UCSD website.