Will high-res radar make tomorrow’s cars safer?

There’s a problem with one of the safety systems in the 2018 Nissan Sentra: It doesn’t work.

Let me clarify: There’s a problem with one of the safety systems in lots of new Nissans, including Sentras, Altimas, Muranos, and Pathfinders, and in lots of cars on the road today from other major brands, and it underscores a basic problem with modern mobility: As cars increasingly rely upon technologies such as radar, lidar, and cameras, the limitations of those existing systems is becoming increasingly obvious.

“Automotive radar systems are shortsighted,” the trade magazine Microwaves & RF wrote recently. “They lack the resolution to distinguish objects close to the vehicle, while discarded soda cans and other metallic objects can cause false alarms that increase the risk of accidents. They also struggle to process information fast enough for highway driving.”

That seems ideal for a crucial safety feature, right? Hardly.

We’re talking specifically about the front collision warning system in the 2018 Sentra, and the array of driver assist features that protect the driver; Nissan calls it a Safety Shield, and said in May that six models — roughly 1,000,000 Nissans in total — will include Automatic Emergency Braking as a standard feature. AEB relies on radar in the front grill to detect obstacles in front of the car, warn of possible collisions, and hit the brakes if the driver can’t. Too bad it doesn’t work properly. A problem with the Bosch-built radar module has some sensors deactivating over and over again during trips, with a loud beep and pair of orange alerts glaring from the instrument console.

Other eyes for today’s vehicles, including video cameras and lidar, carry their own issues. The problem of sensors is pervasive across models and across the automotive industry — and new tech may be the solution.

Here comes “high-res radar”

Video cameras are two-dimensional, unable to detect depth — and therefore not really useful when it comes to spotting a pedestrian in the crosswalk ahead of you. That’s where radar comes in, and its three-dimensional mapping skills. Here’s the thing: Radar hasn’t really changed in decades. You know that sweeping, pinging thing airport towers use? Yep, radar. And experts are divided on whether it’s good enough for the precision demands of cars.

“For today’s safety guidelines … radar is absolutely sufficient,” Kamal Khouri, VP of Advanced Driver Assistance Systems at NXP Semiconductors, told Digital Trends. “While radar may not be able to classify a detected object, it has the ability to detect objects at a long range and determine its direction of travel and velocity. This is not something that a camera is able to do.”

NXP is the world leader in automotive radar systems, but even Khouri is willing to acknowledge the shortcomings of radar. In an interview with Connected Car Magazine, he noted “the weakness of automotive radar today is its resolution and its ability to identify objects clearly and classify them. A radar system today can tell you that there is an object present, but can’t tell you what the object is.”

But like music and TV before it, radar is going high-definition, thanks to a new generation of startups that may have the answer to a problem you didn’t even know your car had. Enter high-resolution radar. But what exactly is it?

According to Roger Keen, Radar Segment Manager at NXP, high-resolution radar means improvements in a number of key attributes: range, doppler and azimuth or angle.

“Let’s briefly look at angular resolution,” Keen told Digital Trends. “For autonomous vehicles, improvement in angular resolution is most noticeable because it not only enables radar systems to distinguish two objects at the same range but it also enables radar systems to distinguish a pedestrian from a light vehicle from a large truck. This is obviously very important when architecting sensors systems of the autonomous vehicles of tomorrow.”

That’s why NXP is working on high-res chips of its own. At CES in 2018, the company announced the MR3003 Radar Transceiver, a high-res part specifically designed for automotive applications. But a whole slew of startups aim to challenge the incumbent giant in the race to high-res.

There’s a company called Uhnder that together with parts supplier Magna have built a high-resolution radar, for example. Called ICON RADAR, it “helps close the gap between level 3 and level 5 to reach full reliable autonomous driving,” the companies write.

Then there’s Ghostwave, which is revolutionizing radar — you can tell because that’s the title of the company’s homepage, so it must be true (that’s sarcasm, if you weren’t certain). And Oculii, which makes solid-state HD radar that relies on advanced signal processing and a dizzying array of buzzwords to…well, you get the idea.

Digital Trends reached out to several of these secretive startups, and none wanted to participate in this story — an effort to protect “trade secrets,” perhaps. It may also signal that HD radar is still just an idea in a folder on the desk of someone in marketing. But there may be another solution. 

Can Lidar solve the problem?

Lidar is like radar but with light beams instead of sound waves, and systems for today’s cars consist of large spinning buckets that sit on the roof and cost thousands of dollars apiece. Your vehicle needs several to paint a 3D picture of the road ahead — but that doesn’t really seem like a practical solution, does it? So lots of companies are trying new stuff, notably solid state devices that shrink down dramatically and save money. But are they better?

“Certainly none of them come even remotely close to actually meeting the key performance specifications that are truly required,” Austin Russell, founder and CEO of Luminar, told us earlier this summer. “If you want a $100 solid state lidar you can go down to Best Buy and pick up one. It’s a golf range finder. Sure, it sees out technically to a couple hundred meters — tells you how far away the flag is. But is that applicable to autonomous cars? Absolutely not.”

Russel has spent years coming up with a better system, using alternate chemistry and rethinking the physics of lidar with a completely different wavelength of light. The challenge is, there weren’t any off-the-shelf components readily available. So Luminar built them. All of them.

“Instead of going down to our fab in San Jose and picking up some silicon photo diodes and some 905-nanometer lasers from the same vendors that everybody else buys them from, we had to make these 1,550-nanometer lasers from scratch. You know, design them. We had to make our own special material — indium gallium arsenide photo diodes, that’s the material you have to use to see this wavelength of light, you can’t use silicon any more, which is really easy to use…we even had to make our own chips for the system.”

Luminar’s chips may improve on existing lidar sensors, but does that mean car makers could scale down the number of sensors some day — especially if, as we’ve seen, radar maybe doesn’t work too well?

For now: No solution on the radar

Like a healthy breakfast, cars need to have a balance of different sensors. Radar will remain key, experts say — but increasingly, cars will rely on lidar.

“You need a combination of cameras, radar, and lidar in order to create a self-driving system,” Jada Tapley, VP of Advanced Engineering at Aptiv, told us earlier this year.

“The industry uses three sensing systems for a reason,” Khouri told Connected Car Mag. “That is because no one system is complete in its ability to enable autonomous driving or ADAS (advanced driver-assistance systems) functions. All systems — camera, LiDAR, radar — have their own strengths and weaknesses.”

And given the weaknesses that exist today, technology appears to be a limiting factor, both in safety systems like AEB and in the quest for fully autonomous vehicles. Keep in mind that engineers are still racing for a solution — and you should maybe race a little less aggressively.