Imaging radar, a leap forward for driving automation – An interview with Uhnder

The automotive radar market remains highly dynamic, with an expected 14% CAGR at both the system and silicon levels, accounces Yole Intelligence in its Radar for Automotive report (2022 edition). This evolution is driven in volume and value by solid market drivers, those of safety regulations and driving automation.

However, the industry is entering a paradigm change as cost optimization will no longer be the central focus. Indeed, the mission for radar sensors is evolving significantly as OEMs demand highly accurate, robust 360° sensing to achieve driving automation. Radar has a clear role to play here, but angular resolution must improve dramatically. Recent advances in RF performance and ongoing developments in radar computing are leading toward so-called imaging radar. The industry is actively developing such solutions.

In that context, Cédric Malaquin, Team Lead Analyst of the RF activity at Yole Intelligence, had the chance to interview Max Liberman Vice President, Chips at Uhnder, one of the pioneering imaging radar companies. Yole Intelligence is a Yole Group company.

Cédric Malaquin (CM): Please introduce yourself and your company to our readers

Max Liberman (ML): My name is Max Liberman, and I am vice president of chips at Uhnder. At Uhnder, we are focused on revolutionizing automotive and automated mobility with the world’s first digital imaging radar-on-chip (RoC) to enable safe mobility for both people and goods as well as make roads safer for all users. Our fully software-defined imaging digital radar chip and sensor modules perceive objects that cameras, LiDAR, traditional analog radar, and the human eye can easily miss due to either poor ambient conditions or performance limitations.

Cofounded in 2015 by Manju Hegde and Curtis Davis and headquartered in Austin, Texas, Uhnder has grown to more than 300 employees worldwide, and we’re expanding as we continue to grow in both the automotive Tier 1 and automated mobility segments. We are focused on revolutionizing the world of mobility by offering the best-performing radar sensors.

The company’s development has been successful so far with, among other achievements, the closing of a Series C funding at the end of 2020 and qualification for Fisker’s Ocean from Magna.

CM: What are the next milestones we can expect from Uhnder?

ML: We achieved a major milestone in this quarter with an automotive release of the world’s first digital radar-on-chip (RoC), the S80. The S80 is AEC-Q104 qualified and has an ISO26262 Automotive Safety Integrity Level (ASIL)-B rating certified by TUV-Sud. We have also successfully completed the first PPAP for the S80 with our lead automotive Tier 1 customer for its ADAS system. Our future milestones will further improve upon our industry-leading performance by leveraging the inherent advantages of Digital Code Modulation (DCM) over traditional analog radar architectures.

Uhnder highlights Radar-on-Chip (RoC) and Digital Coded Modulation (DCM) as key differentiators.

CM: Could you explain what problems Uhnder is trying to solve with RoC and DCM?

ML: With RoC and DCM, we are solving the two biggest safety issues with traditional analog radar found on vehicles today: (1) lack of resolution and contrast to be able to accurately and reliably detect and distinguish objects, such as a bicyclist next to a guardrail or a child next to a large truck, and (2) mutual radar interference, which is becoming a growing concern as the proliferation of radar sensors is projected to explode in the coming years, and it can have significant safety implications for the performance of ADAS, ADS, and Avs.

CM: Could you describe the RoC approach, its benefits, and its limitations?

ML: We developed a fully integrated 77 GHz, 4D Imaging Radar-on-Chip (RoC) that uses a DCM (Digital Code Modulation), MIMO (multiple-input multiple-output) radar architecture capable of processing up to 192 virtual channels in a small 12.8 mm x 8.21 mm wafer-level package. It supports 12 Tx (transmit) antenna channels and 16 Rx (receive) antenna channels, whereas similar competitive analog radar solutions require four or more RF MMICs and a separate processor to accomplish this level of performance yet are way higher on power and cost.  The S80 is fully software-defined, has built-in processors to run algorithms on-chip, and can also enable optimization with deep learning neural networks and AI (artificial intelligence) found in the most advanced automated perception systems. The integrated approach of the RoC architecture reduces the complexity of the radar hardware design, increasing reliability and easing system integration. It also reduces the overall size of a radar sensor module, which is critical for logistics automation and benefits automotive OEMs, as they have more freedom in the placement of the radar sensor module and better industrial design.

Uhnder’s RoC architecture addresses 12 transmit and 16 receive antenna arrays. Larger antenna arrays can be achieved by cascading up to 4 RoCs. The master/slave architecture is well established in the RF/analog part.

CM: How does it work on the processing/computing side?

ML: From a high level, compared to traditional analog radar, we do as much as we can in the digital domain. For example, we do all the transmit signal generation and modulation in the digital domain as well as the receive signal processing. We also have functional safety and security built into our digital chip, which is critical for automotive applications to make the device more reliable and safer.

CM: Radar can be very computationally intensive. Could you elaborate on the RoC’s computing power? Will it compete with domain controllers or centralized architecture?

ML: Our S80 device has two Cortex-R5F ARM CPUs and two Tensilica-P5 DSPs integrated on-chip. Being fully software-defined, it can support user and third-party algorithms on-chip as well as over-the-air updates with best-in-class HSM security. The S80 can output a variety of data types, from ‘raw’ radar detection data all the way up to object tracking lists. We want to give our customers complete flexibility to optimize their perception sensor fusion stack depending on the function they need to perform. We support centralized compute architectures and can also support edge computing on-chip for those customers wishing to optimize their overall architecture.

Uhnder’s 77 GHz, 4D Imaging Radar-on-Chip (RoC) with Digital Code Modulation is certified for use in key automotive safety applications, such as AEB (automatic emergency braking), lane keep assist, ACC (adaptive cruise control), and blindspot detection. The RoC uses a PMCW (phase modulated continuous wave), MIMO (multiple-input multiple-output) radar architecture capable of processing up to 192 virtual channels. It supports 12 transmit antenna channels (Tx) and 16 receive antenna channels (Rx) – Courtesy of Uhnder, 2022

CM: Could you elaborate on RoC’s High Contrast Resolution? What angular resolutions are the RoC and cascaded RoC capable of achieving?

ML: High contrast resolution allows the radar to distinguish between two objects, in particular a small RCS object in the proximity of a large RCS object, such as a young child next to a pick-up truck. We can achieve a high contrast resolution of 35 dB, which is 30 times better than a traditional analog radar.

In addition to the RoC’s number of channels, the radar antenna design is critical to achieving the highest angular resolution. We currently have a radar sensor module that can achieve 0.5° angular resolution and a roadmap to improve that soon.

CM: Could you explain the operating principles of DCM, its benefits, and limitations?

ML: DCM stands for Digital Code Modulation. With DCM, each radar transmitter has its own unique digital code modulated onto the signal so we can easily identify the correct corresponding return radar signal. As such, DCM provides high contrast resolution, delivering maximum discrimination and high-confidence detection of independent targets, including vulnerable road users (VRUs), such as pedestrians and cyclists, while minimizing mutual interference from neighboring radar.

As the number of radars used on vehicles continues to grow exponentially, interference is becoming a growing safety concern, especially among traditional analog radars. NHTSA has already commissioned a radar congestion study that concluded that existing analog radar systems might be significantly impacted by coexistent radar interference. In addition to DCM, the Uhnder RoC has other interference-mitigating technology to help reduce any interference it might have on other non-DCM radars.

Interference management/mitigation is a major concern in the radar industry that Uhnder is addressing with DCM.

CM: Do you see a push from other players for DCM or similar technics?

ML: Yes, many companies, both start-ups and mature, are currently developing imaging radar. Digital Imaging radar provides the resolution and contrast needed to detect, track, and distinguish objects on our roadways, such as other vehicles, pedestrians, motorcyclists, and bicyclists, accurately and reliably. No other perception technology is proven to reliably and under all extreme ambient conditions do this at the price point of imaging radar, hence why it is such an attractive technology.

All the other radar players are based on a traditional FMCW (frequency-modulated continuous wave) analog radar, whereas Uhnder’s products are the first and only to use a fully integrated and software-defined digital radar architecture based on DCM.

Many new companies are currently developing imaging radar.

CM: What is Uhnder’s added value or differentiator compared to these new players?

ML: Uhnder’s added value or differentiator compared to other imaging radars is our unique DCM architecture that allows us to achieve 16 times better angular resolution and 30 times higher contrast resolution compared to traditional analog radar with much higher mutual interference mitigation. We offer a smaller radar footprint using fewer components to reduce complexity, size, and cost, as well as increase reliability. This integration will also reduce overall system power consumption.  In some cases, we see that our competitors have 100% higher power consumption. 

Imaging radar enables better resolution than legacy radar. But there may be compromises on the system’s complexity, the cost, and probably the power consumption and size.

CM: Where do you see the sweet spot for imaging radar regarding all these parameters?

ML: Everywhere. In safety-critical applications, the more accurate and reliable data that you have from your sensor, the better the algorithms will be to make decisions to have a safer outcome.

Mobileye is already a major player on the vision side in cars and is currently developing high-performance imaging radar.

CM: In your opinion, could they become a key player on radar as well, and how would you compete with such a player?

ML: In general, we welcome and encourage the mainstream adoption of imaging radar to replace the traditional radar systems found on most vehicles today. Competition will drive all of us to develop a better solution to make a safer vehicle.  Automotive companies, both Tier 1 and Tier 2, will quickly see the limitations of using analog chips, such as size, power, and cost, in addition to the mutual interference concerns.  

CM: Do you have a final word for our readers?

ML: Uhnder’s focus and passion are to make roadways safer for all road users, especially vulnerable road users (VRUs), as the percentage of fatalities of individuals outside the vehicle continues to grow. In addition, we want to enable the safe deployment of automated logistics vehicles, so they can safely transport items. This requires precise perception sensing technology that can accurately and reliably detect all objects of interest, especially under conditions when human, camera, or LiDAR visibility is reduced due to snow, ice, rain, fog, smoke, darkness, or sun glare. Digital imaging radar is best positioned to meet these demands with the lowest risk given radar’s long proven history in automotive.

About the interviewee

Max Liberman is Vice President of Chips at Uhnder where he oversees global sales and marketing for the company’s digital imaging radar-on-chip designed to enable the highest-resolution digital perception for ADAS and autonomous vehicles. Max joined Uhnder in 2020 after a distinguished 19-year career with Analog Devices having served most recently as director of sales for the automotive business across Asia, Europe, and the Americas.
Max earned his MBA in entrepreneurial studies from Babson F.W. Olin Graduate School of Business and his Bachelor of Science degree in electrical engineering from Worcester Polytechnic Institute.

About the interviewer

Cédric Malaquin is Team Lead Analyst of the RF activity within the Power & Wireless Division at Yole Intelligence, part of Yole Group.
In this position, Cédric is managing the technical expertise and the market knowledge of his team for the company.

Cédric graduated from Polytech Lille (France) with an engineering degree in microelectronics and material sciences and holds a DEA in microwave and microtechnology from the University of Lille.