New “4D Imaging Radar” for Automotive Industry in the Works. Is the Name Accurate?
This week, Xilinx and Continental announced the auto industry’s first "production-ready" 4D imaging radar. But do 4D radars actually exist beyond the theoretical?
The autonomy of a self-driving vehicle is often measured on a scale of the “levels of automation.” Briefly, this is a scale from 0 to 5—0
meaning absolutely no autonomy and 5 referring to complete, driverless functionality. For reference, most Tesla vehicles today fall at a 3,
and that is considered a generous score.
Explanation of the levels of automation. Image used courtesy of GeoSpatial World
There are some serious improvements that need to be made before we can obtain completely autonomous vehicles. Among these necessary
improvements, many experts believe that a shift to 4D radar imaging will be a vital technology.
4D Radar: What is It?
The original radar had the capability to measure in two dimensions. Utilizing a specialized rotating antenna and listening for echo signals,
a 2D radar has the ability to acquire two coordinates and determine the position of the object.
Then 3D radar came along. The 3D radar would rotate like the 2D radar, but after each scanning rotation, the antenna elevation is changed
for the next sounding. This scenario will be repeated on many angles to scan all the volume of air around the radar within the maximum range.
In this way, a 3D radar could detect three dimensions: azimuth, elevation angle, and speed.
3D radar depiction. Image used courtesy of Radar Tutorial
Now, the 4D imaging radar is starting to be introduced. From a theoretical physics perspective, time is considered the 4th dimension. When applied
to radar, this would be Doppler frequency, which shows whether an object is moving toward the vehicle or away.
A Quick Note on Jargon
Some people argue that 4D radar is a misnomer and a marketing ploy, because these radars aren’t really mapping time, but rather utilize time to
understand the environment in 3D. Furthermore, they argue that modern reconnaissance radars already have the capability to detect Doppler shifts,
so the technology also isn’t novel.
These points, however, boil down to semantics. The point is that 4D radar integrates the fourth dimension of measurement into 3D radars to help
better understand and map the environment. Even if the specific technology isn’t novel, the integration of it is new and worth investigation.
Xilinx and Continental Team Up on 4D Radar Hardware
Looking to push ADAS technology closer to level 5 automation, Xilinx and Continental have teamed up to develop 4D radar hardware.
4D vs. current radars. Image used courtesy of Spar3D
creating what they claim to be the automotive industry’s first production-ready 4D imaging radar. This new radar system is said to offer a
300-meter and ± 60° field-of-view.
What’s Going on Under the Hood?
This system specifically relies on the Xilinx Automotive Zynq UltraScale+ MPSoC to power Continental’s imaging radar. The MPSoC is an
adaptable platform, allowing the radar to operate in multiple sensor-platform configurations and adapt to OEM specification.
Block diagram of Zynq UltraScale+ EV. Image used courtesy of Xilinx
This is made possible by parallel processing within the device’s programmable logic, enabling the fully independent, yet simultaneous
processing pipelines that are critical to 4D sensing. Furthermore, having multiple DSP threads enables hardware acceleration of real-time
radar sensor inputs.
A Step Closer to Autonomous Driving?
This new system being developed by Xilinx and Continental aims for the flexible and dynamic implementation of 4D radars into autonomous
vehicles. This is important because 4D radar is supposed to provide greater range, field of view, perception, and tolerance in bad weather conditions.
While Xilinx's claims of the first “production-ready” 4D radar seem a little unwarranted—the radar is not available for sale currently—the
implications are still relevant. Combining Xilinx’s silicon expertise with Continentals advanced radar technology looks to be a good match for
developing more sophisticated vehicles in the near future.