Imaging radar, the next big thing in automotive radar – A dual interview with both Xilinx & Continental

Radar is a well-established sensor in automotive. This market reached US$5.5 billion back in 2019, and is expected to grow at an 11% CAGR between 2019 and 2025, announces Yole Développement (Yole) in its “Status of the Radar Industry: Players, Applications and Technology Trends 2020” report. At the end of this period, the market should be worth nearly US$10.5 billion.

In the automotive field, radar has now become standard equipment. Following more stringent test scenarios, two trends are emerging. One consists of moving forward with imaging radar capable of more accurately describing the scene in front of and around the car. The other is to increase the number of sensors around the car and coordinate them to improve scene perception.

In both cases, a strong emphasis on signal processing and computing is emerging, while cost issues are particularly important in automotive. Multiple questions have been raised about where to process the signals and how better to exploit radar sensor data. This will likely contribute to a major transformation of the automotive radar industry… In a context of market growth, without doubt, the automotive radar industry is experiencing a transformation following more stringent safety requirements and driving automation.

Last month, Continental and Xilinx announced a significant collaboration to create the auto industry’s first production-ready 4D imaging radar – the ARS540 – for autonomous driving. According to both  these leading companies, this partnership will enable newly-produced vehicles equipped with the ARS540 to realize SAE J3016 Level 2 functionalities and will help pave the way to Level 5 autonomous driving systems… Full announcement.

Yole’s analyst, Cédric Malaquin, specialized in RF electronics devices and technologies, had the opportunity to debate Willard Tu, Sr. Director – Automotive Business Unit at Xilinx and Norbert Hammerschmidt, Vice President Program Management Radar, Advanced Driver Assistance Systems business unit, Continental.

Discover the details of this discussion below:

Cédric Malaquin (CM): To Xilinx – Could you introduce yourself and Xilinx activities related to automotive radar to our readers?

Willard Tu (WT): I am Willard Tu, Sr. Director and Global Automotive Lead and responsible for revenue creation, margin, product definition, vertical marketing, automotive messaging and position, and ecosystem/partner development. 

Xilinx is #1 or #2 player is most of the ADAS systems. We are #1 in LiDAR, 4D imaging radar, DMS, and Full Display Mirror.  We are #2 in Forward Camera, and within the top 2 suppliers for Surround View systems (including Automated Parking Assist).

We have been involved in radar for multiple generations – initially as a companion chip in the early 2D radar systems and now as the main SoC for 4D imaging radar.

CM: To Continental – Could you introduce yourself and remind Continental’s positioning on the automotive radar market to our readers?

Norbert Hammerschmidt (NH): Within Continental we have a long and outstanding history on the automotive radar market. When the first factory-built vehicle equipped with a long-range radar used for adaptive cruise control came onto the market 20 years ago in the Mercedes S-Class (W220 series), we played a key role in this joint development with Mercedes. The range of the sensor at the time was 150 meters, making it the most advanced system of its day. Today, with our latest, fifth generation long-range radar, with a range of up to 300 meters and unprecedented resolution, we offer sensors that are an enabling technology for oncoming NCAP requirements and also for future automated mobility.

CM: To Xilinx – Could you explain on which part of the imaging radar Xilinx has worked?

WT: Xilinx Zynq UltraScale+ MPSoC is utilized for its digital signal processing (DSP) capabilities.  In the digital processing, the device aggregates the inputs from the 192 virtual antenna channels and does complex signal processing, FFTs, to create the point cloud data.  The data is then extracted to create the point cloud range (distance) azimuth (angle relative to ego vehicle), relative speed, and elevation (height).

CM: To Continental – Currently, legacy radar sensors meet level 2 requirements best case. Could you explain the market dynamics that drove Continental to develop a level 3 – level 4 radar?

NH: One trend will continue to dominate development in future: increasingly precise vehicle surroundings monitoring, where radar, lidar and camera data are combined to generate a detailed and seamless view of the entire vehicle vicinity. This high-precision model is important to enable driver tasks to be performed by the vehicle in automated driving. By combining sensors, the aim is to achieve an understanding of the vehicle’s surroundings which is as good as or better than human perception. Long-range radar will also be able for “glancing back” in the future. If automated vehicles change lanes independently, they must be able to identify precisely whether and how fast other road users are approaching from behind. 20 years after its world premiere, Continental’s long-range radar has become a key component in automated driving. It is also a driving force towards new, future-oriented and safer mobility.

CM: To Xilinx – Could you describe the computing performance of your imaging radar device and how it compares/differentiates to/from the competition?

WT: The Xilinx Zynq UltraScale+ MPSoC is a heterogenous computing platform.  We have a processing system that includes a quad Cortex A53, a dual Cortex R5, and a Mali 400 GPU, DDR controller (gen3 and gen4), security features, power management, high speed connectivity, and general connectivity.  In addition, we have our secret sauce – Programmable Logic (PL).  The PL is what scales within the MPSoC family. The processing system is similar to what other competitors have, but the PL is what makes us unique.  Programmable Logic allows Continental to create a customized and highly efficient signal processing chain, due to the parallel processing capability inherent within the PL architecture.  Additionally, the PL can be leveraged to do the AI processing required for detection and classification.  Without the flexibility of the PL, there would not be enough parallelism to complete the signal processing, detection and classification within the necessary thermal limits.

CM: To Continental -The Covid19 pandemic has strongly affected the automotive market. Do you confirm the pull from automotive OEM for an imaging radar and when do you see it?

NH: We definitely see a pull and demand for our ARS 540. The premium, long-range 4D imaging radar with high resolution and 300-meter range will enter series production mid of next year in passenger cars of a European OEM.

CM: To Xilinx -Will Xilinx also offer the RF part of the imaging radar?

WT: Imaging radar is an innovative product, that requires a very mature technology which further enhances its abilities.

The primary challenge for the system designers is to master the complex signal chain to generate the point cloud. The next task is to do the detection and classification so that it is a “smart sensor”. After that, RF integration is a possibility.  We are offering the RF (A/D interface) for many of the Lidar companies today with our RFSoC devices.  It is possible that we could use those same devices or derivates of those devices with the future generations of imaging radar. 

CM: To Continental – What differentiates the imaging radar you introduce today from the high-end radar version Continental is already offering?

NH: The ARS540 is a high performance premium long-range radar sensor which enables highly automated driving in combination with other technologies. It provides best radar performance in a state-of-the-art sensor size. One of the benefits of the ARS540 lies in its compactness and flexible usage which makes it easier to use across entire vehicle platforms.

One of the features of the new radar generation is a higher resolution compared with the previous radar generations, with which a more exact picture of the traffic situation can be gained. In addition, road limits such as curbstones as well as the height of objects like the tails of a traffic jam under a bridge are detected thanks to the sensors’ elevation measurement accuracy. Thanks to the worldwide trend of using 77 GHz technology, the resolution of the sensors is becoming higher and facilitating, for example, more accurate detection of smaller objects such as a lost spare wheel or an exhaust that has fallen off.

Interviewees

Willard Tu is Sr. Director and Global Automotive Lead, responsible for revenue creation, margin, product definition, vertical marketing, automotive messaging and position, and ecosystem/partner development.

Tu has spent over two decades at the axis of the semiconductor, automotive and computing industries. He was previously at Arm, where he evangelized CPU IP and developed ecosystems to support Arm’s growth in automotive. At NEC Electronics, (now Renesas), Tu led the North American Automotive sales and marketing teams, growing sales to over $150 million. Tu holds a BS degree in Electrical Engineering from the University of Michigan and an MBA from the University of Phoenix.

Norbert Hammerschmidt is Vice President Program Management Radar, Advanced Driver Assistance Systems business unit at Continental. He has 31 years of experience in automotive research & development, project management, sales and general management. Prior to Continental, Hammerschmidt worked for Dornier GmbH, Temic GmbH and, since 2003, for Continental AG. Norbet Hammerschmidt has 23 years of experience in ADAS in different positions for assisted and autonomous driving. He also has 5 years of experience abroad by living and working in Asia

Interviewer

As a Technology & Market Analyst specialized in RF devices & technologies within the Power & Wireless division at Yole Développement (Yole), Cédric Malaquin is involved in the development of technology & market reports as well as the production of custom consulting projects. Prior to his mission at Yole, Cédric served Soitec first as a process integration engineer for 9 years, then as an electrical characterization engineer for 6 years. He contributed significantly to FDSOI and RFSOI product characterization. He has also authored or co-authored three patents and five international publications in the semiconductor field. Cédric graduated from Polytech Lille in France with an engineering degree in microelectronics and material sciences.

Related reports

Status of the Radar Industry: Players, Applications and Technology Trends

Worth more than $20B in 2019, the radar industry is experiencing a major transformation prior to entering the commercial era.

Sensing and Computing for ADAS Vehicle 2020

Advanced Driver Assistance System functionality will attract customers and restart growth of the automotive business.