The automotive industry needs a reliable sensing technology today to improve safety outcomes for planned hands-off and eyes-off features tomorrow. Manufacturers, aware of the advantages and disadvantages of traditional radars, have tried to enhance radar data by moderately expanding the channel array while still relying on legacy chips. When it comes to reliability in complex environments or at higher speeds, however, these smaller array radars simply fall short. Ultimately, the dream of safe hands-free and eyes-free operation will rise and fall based on the robustness of the sensing technology.
To meet the expectations of the L3 customer experience, radar systems should offer diversity and real redundancy to cameras (and sometimes LiDAR), facilitating perception across long ranges without speed limitations. But if radar is to provide data detailed enough to be relevant for truly safe autonomous driving without trade-offs, it must be designed to do so from the ground up. By harnessing cutting-edge radar technology and meticulously engineering a next-generation radar chipset to meet the rigorous demands of ADAS and AV performance, Arbe offers an optimal, production-ready solution, achieving a massive channel array and resolving traditional design obstacles.
Radar technology holds the promise of extending the detection range beyond cameras and LiDAR, while also improving detection capabilities in adverse weather and lighting conditions.
However, “traditional” or legacy radar technology, with its limited channel count (usually 3×4), is akin to relying on “blurry”, low-resolution image to navigate the world. Legacy systems struggle to deliver adequate resolution, detect smaller objects at a distance, and lack the ability to perceive elevation – a critical aspect of understanding the three-dimensional world we operate in.
Some manufacturers have opted for radar solutions with a higher channel count, typically ranging from 12×16 to 16×16, while leveraging the legacy 3×4 chips. These solutions, sometimes referred to as “Imaging Radar,” do provide improved resolution and elevation capabilities compared to legacy radars, enhancing performance for basic ADAS features like AEB and ACC. Based on this, these companies state the achievement of L3 autonomy.
However, a closer look reveals a different story. In reality, the use of smaller array radar systems for this purpose has revealed reliability gaps, necessitating the imposition of speed limitations for L3. The performance simply isn’t sufficient in terms of resolution or sidelobe reduction for hands free or eyes-off features. Detecting small objects in challenging environments presents additional challenges, including dynamic range issues, separation of multiple targets, considerations of elevation, and more. In numerous tests, radars with fewer than 300 channels have proven inadequate in discerning static objects, detecting lost cargo, or accurately resolving their position in complex scenarios. The radar’s role in perception remains marginal, hindered by its inherent limitations.
This is not to say that a smaller array radar adds no value to the sensor suite; in fact, they are very important within a specific context. Arbe’s 24×12 radar, for instance, serves as a valuable solution for lower-tier applications and surround implementations, particularly for corner and rear installations. Rather than being contenders to power full-fledged L3 autonomy, these radars play a complementary role. Paired with high-channel count front-facing radars in an L2+ or L3 setup, smaller arrays contribute to a 360-degree view around the vehicle. This configuration not only offers cost-effective benefits but also delivers true long-range performance, high resolution, and a wide field of view. Further, such radars can serve as a front Long-Range Radar (LRR) solution when catering to a limited set of driver-assist applications lower than L2+.
Achieving enhancement in advanced driving solutions requires sensor fusion, which demands a robust point cloud that can only be achieved with a wide channel array and a powerful radar processor. As such, high-channel count radars offer a multitude of advantages:
In his presentation at Tech.Ad Berlin 2024, Dr. Jürgen Dickmann, Head of Radar and Radar-Perception at Mercedes-Benz Group, stated clearly his belief that safe ADAS capability demands a minimum of 1024 channels, a 32×32 array. Likewise, a recent blog post from Continental Automotive noted that “higher channel counts lead to more accurate sensors,” acknowledging that innovative solutions with 2000+ channels contribute to safety.
However, the Continental blog goes on to suggest that such solutions may potentially be associated with tradeoffs like power consumption, compute power, design complexity, physical size, and heat dissipation. In reality, this argument only holds water for a 2000+ channel radar system was designed with existing legacy 3×4 or 4×4 MMICs and standard, general-purpose DSPs or FPGAs (even the best in class).
Arbe took a different approach from day one. We designed a next-generation radar chipset solution specifically built to support a 48×48 array, eliminating the limitations inherent in legacy systems. This proprietary chipset underpins Arbe’s radar technology, optimized for L2+ and beyond. It features high port density RF chips: 24-channel Tx and 12-channel Rx chips, along with a dedicated radar processor capable of handling the massive data generated.
An industry first, Arbe’s chipset offers 2,304 channels at comparable cost, same power consumption, and similar system size compared to the 192 or 256 channel “4D imaging” radars based on 3×4 MMICs. Want proof that this chipset supports a massive channel array radar without the tradeoffs? You need only to look to Arbe’s tier 1s, who have each designed 2,304 channel B-sample radars that meet all automotive radar requirements – including heat dissipation, complexity, and mass production maturity – at competitive cost.
The Continental blog also mentions a common misconception – namely, that although higher levels of autonomy will require more channels to be safe, “suppliers should gauge the best balance of performance and cost for each generation and only increase channels as needed.”
Arbe has a different viewpoint.
One cannot simply “gradually add channels” above 300 without switching to another chipset technology, as it is not feasible due to the associated cost, power consumption, heat dissipation, size constraints, and the computational power needed. Expanding will necessitate the development of a new radar system on alternative chips.
The rapid pace of innovation in sensing technologies makes it nearly impossible to predict and specify capabilities five or ten years down the line. Many OEMs are hesitant to invest heavily in maximizing performance from inherently limited hardware, especially since a transition to a different architecture is likely on the horizon. OEMs that underestimate the required channel count now could find themselves handicapped in future feature launches. Aiming low is a significant risk.
Instead, to develop and roll out advanced features, OEMs need to collect massive amounts of data in as many real-world scenarios as possible using the very same sensors that will be in production. Integrating the relevant radar infrastructure as early as possible provides access to a huge volume of quality data, facilitating and catalyzing the roll out of new features to existing and new customers via software updates alone. To that end, a true imaging radar with a high channel count is an invaluable element of the software defined vehicle sensor suite.
Higher channel counts lead to more reliable sensors, and solutions with 2000+ channels demonstrably contribute to safety. Arbe’s 48*48 massive MIMO radar takes this a step further. It delivers unmatched performance at a competitive cost, size, and power consumption when compared to smaller array designs. This translates to a staggering hundredfold increase in point cloud density, effectively eliminating ambiguities and false alarms that plague lower channel count systems. The better news? Arbe’s solutions are available today. By proactively integrating these high-performing radars now, automakers can secure a pole position in the race toward autonomous driving leadership.
————————————
This blog contains “forward-looking statements” within the meaning of the Securities Act of 1933 and the Securities Exchange Act of 1934, both as amended by the Private Securities Litigation Reform Act of 1995. The words “expect,” “believe,” “estimate,” “intend,” “plan,” “anticipate,” “may,” “should,” “strategy,” “future,” “will,” “project,” “potential” and similar expressions indicate forward-looking statements. Forward-looking statements are predictions, projections and other statements about future events that are based on current expectations and assumptions and, as a result, are subject to risks and uncertainties, including the risk and uncertainties resulting from the October 7th attack upon Israel, conflicts and potential conflicts involving Israel and the effect of the reaction to the war against Hamas on Israeli companies, particularly high tech companies as well as market acceptance of Arbe’s radar processor and Arbe’s radar processor performing in the manner which Arbe anticipates, and the risk and uncertainties described in “Cautionary Note Regarding Forward-Looking Statements,” “Item 5. Operating and Financial Review and Prospects” and “Item 3. Key Information – Risk Factors” Arbe’s Annual Report on Form 20-F/A for the year ended December 31, 2023, which was filed with the Securities and Exchange Commission on March 28, 2024 as well as other documents filed by Arbe with the SEC. Accordingly, you are cautioned not to place undue reliance on these forward-looking statements. Forward-looking statements relate only to the date they were made, and Arbe does not undertake any obligation to update forward-looking statements to reflect events or circumstances after the date they were made except as required by law or applicable regulation. Information contained on, or that can be accessed through, Arbe’s website or any other website or social media is expressly not incorporated by reference into and is not a part of this blog.
Connect to learn more
