This article also appears in
Subscribe now »

New automotive radars take tech from a blip to a boom

Automotive radar rides the 77-GHz technology wave toward greater capability and vehicle safety.

Since it first appeared in automobiles in the 1980s, radar has become a reliable and increasingly cost-efficient tool for detecting objects on the road and along the roadside. The technology is forecast to be in 50% of all new cars and light trucks by 2020, where it will be “fused” with other on-board sensors – cameras, LiDARs, ultrasound – to offer multiple layers of detection and reliability in various environments and scenarios.

“Making traffic accidents a thing of the past is going to require a full stack in the car, the brain, and using sensor fusion,” stated Michael Peredo, senior solutions engineer for Velodyne LiDAR. Speaking with SAE’s Autonomous Vehicle Engineering at the 2019 CES, he noted that while his company obviously focuses on LiDAR, radar remains a vital component of tomorrow’s vehicle-sensor suite.

In July 2017, the FCC adopted new rules to expand the spectrum for automotive radar use. The agency is phasing out 24 gigahertz (GHz) and is opening frequencies for automotive use in the 77-GHz band, part of the contiguous five GHz of millimeter waves from 76-81 GHz. In addition to providing higher resolutions in both depth perception and range, experts note the change to 77-GHz also enables a reduction in antenna area and less interference with other on-board systems.

The FCC’s rule change has triggered greater focus on radar technologies among supply leaders Bosch, Aptiv, Continental, Denso, Hella, Valeo, Veoneer, and newcomer Magna, whose Icon radar was developed in collaboration with Uhnder. Semiconductor suppliers in this space, such as Infineon, NXP, STMicroelectronics and Texas Instruments have followed suit with new signal-processing developments.

“The sensors, ultimately, will be millimeter-wave radar,” Peredo observed, which is “less susceptible to snow, rain, and dust. It’s really good at spotting a car.” Its capability also dovetails with that of LiDAR. For example, 77-GHz automotive radar translates to 2.3 mm wave radar. Velodyne’s LiDAR operates near-infrared at 905 nanometers. Modern radar thus shines, so to speak, in low-visibility situations where LiDAR suffers.

“You don’t want [the sensors] to disagree, and you want [the artificial intelligence algorithm] to weigh radar more heavily in its dynamic assessment of the data” in snowstorms and heavy rain, for example, Peredo said. “Because of these operational differences, they really complement each other in a self-driving vehicle or in a driven vehicle with these additional safety features.”

Higher-res solutions
Sensor fusion is also the name of the game at ZF. At CES, the German tech giant showed three new radar sensors: the Gen21 full-range radar (for SAE Level 4 and 5 autonomous driving), the Gen21 mid-range radar (for Level 2 and up), and the Gen5 short-range corner radar (to detect stationary and moving objects, up to 160 meters away). All operate at 77 GHz.

“Each and every sensor has its pros and cons,” explained Damit Bartulovic, ZF’s global sales lead for ADAS. “With radar, you can just get rid of the negative aspects of any light device, like camera and LiDAR. When it comes to automated driving, you will want to have not only two sensors being redundant but also a third one due to the fact that if one is telling you, ‘I see an object’ and the other one is telling you ‘I don’t,’ you need a third one to arbitrate.

Martin Randler, ZF’s director of sensor technologies and perception systems, said his company’s challenge was to build a radar that has LiDAR-like resolution. He said the company’s new offerings are “very close” in resolution performance. “For Level 4, you always need a mixture of technologies,” Randler said. “If you’re going to build a redundancy to LiDAR, you need to have similar capabilities. If the redundancy is only one-tenth the resolution, that’s not enough.”

ZF’s full-range device can bring high-resolution radar to Level 4 and 5 autonomous vehicles. The antenna pattern is larger and the resulting point cloud looks more like something like you would see with LiDAR. “This is why we believe radar is going to be important in the future, as performance increases in a way that’s suitable for Level 4 and 5 automated driving,” Bartulovic said.

Continue reading »