Imaging Radar’s Potential to Undercut LiDAR for Autonomous Driving

This article explores the evolution of automotive radar, focusing on the competitive landscape between imaging radar and Light Detection and Ranging (LiDAR) in autonomous driving. As automakers aim to achieve SAE Level 3 driving experiences, the need for advanced radar systems becomes essential. While current Level 3 systems operate at low speeds, imaging radar presents a promising alternative due to its cost-effectiveness and improved performance. This article discusses the advantages of imaging radar, the necessary advancements in radar technology, and the potential for imaging radar to replace LiDAR in future autonomous vehicles.

Autonomous Driving Radar Market Dynamics

As the automotive industry advances toward autonomous driving, developing effective radar systems is crucial for achieving safe and reliable SAE Level 3 driving experiences. This involves creating systems that can handle highway driving at conventional speeds. Currently, the two Level 3 systems available can only operate at low speeds, which limits their effectiveness and impacts the consumer experience.

To enable safe, unsupervised operation at highway speeds, automakers must enhance their sensor systems. This requires adding a secondary sensor to complement existing vision sensors. The new sensor must meet several criteria:

• Long range (over 300 meters)
• Accurate direction (sub-1° azimuth and elevation)
• Ability to detect various road users, including vulnerable ones
• Reliable performance in difficult weather and lighting conditions

Currently, autonomous vehicle companies are considering two sensors to meet these requirements: Light Detection and Ranging (LiDAR) and imaging radar. LiDAR is far more prominent than imaging radar for autonomous vehicles; however, things may change at the turn of the decade.

As Chart 1 indicates, imaging radar shipments are expected to increase rapidly through the next 10 years, reaching more than 55 million annual shipments by 2035.

The Rise of Imaging Radar

Imaging radar is gaining traction due to its improved performance and cost advantages, posing a competitive threat to LiDAR. The threat is especially prevalent for short-range applications. Here’s why imaging radar could become a key player in self-driving:

• Cost Efficiency: Replacing LiDAR with imaging radar can lower production costs for Level 3 vehicles, making them more accessible to mass markets.
• Enhanced Capabilities: Imaging radar vendors are not positioning their products as alternatives to LiDAR; they emphasize improvements over traditional radar systems. This includes better resolution and performance under various conditions.

In the short term, Level 4 vehicles and robotaxis will still rely heavily on LiDAR. There is a strong emphasis on safety and reliability, especially after high-profile incidents involving autonomous vehicles. However, as technology advances, the reliance on LiDAR could decrease, particularly in forward-facing applications.

Table 1: Analysis of Imaging Radar Technology

	ADAS Radar	Sparse MIMO Radar	Massive MIMO Radar	Virtual Aperture Radar (VAR)
Hallmarks	Smaller aperture, 12 to 16 virtual channels	Larger aperture, 192 to 256 virtual channels	Typically larger aperture, up to 2,000+ virtual channels (depending on aperture size)	Coherent fusion between at least 2 smaller radar sensors, creating a larger virtual aperture
Example Vendors	Tier One suppliers (Valeo, Conti, Aptiv, etc.), supported by Tier Two semiconductor vendors (NXP, Infineon, TI, etc.)	Tier One suppliers (Conti, ZF, Aptiv, etc.), supported by Tier Two semiconductor vendors (NXP, Infineon, TI, etc.)	Mobileye/Valeo, Arbe	Zendar, ongoing research at other vendors
Example Products	Conti ARS510, Aptiv SRR6	Conti ARS540, Conti ARS640, Aptiv FLR4+	Arbe Phoenix, Mobileye Imaging Radar, Mobileye Compact Imaging Radar	No named products in market or Original Equipment Manufacturer (OEM) evaluation
Strengths	Low cost, wide supplier base, small form factor, low power	Leverages commonly available silicon, wide supplier base, relatively low power consumption, relatively low complexity, easier integration, big improvement in resolution and reliability over Advanced Driver-Assistance Systems (ADAS) radar	Excellent range, resolution, dynamic range, and separation of objects, including objects of differing size and reflectivity	Extracts higher resolution from commonly available front end technology; can be applied to Multiple Input, Multiple Output (MIMO) radar in the future
Weaknesses	Poor resolution, grating lobes, poor dynamic range	Less dynamic range (relative to Massive MIMO (mMIMO)); limited headroom to scale "cascading" approach (noise, coherence)	Integration challenges (size, power consumption, system complexity); novel silicon needed to process massive number of channels	Relatively novel technology; outstanding questions over calibration and maintenance; relatively low dynamic range
Opportunities	ADAS and active safety mandates, quasi-mandates	Level 2+, some Level 3	Some Level 2+, Level 3, Level 4, robotaxis	Level 2+, some Level 3

(Source: ABI Research)

Future of Radar in Autonomous Vehicles

The next step in autonomous driving is achieving reliable Level 3 operation at highway speeds. This demands a highly effective perception stack that can handle potential failures, while providing long-range performance.

• Reaction Time Improvements: Detecting obstacles 150 meters away while traveling at 120 Kilometers per Hour (km/h) gives drivers a 4.5-second reaction time, compared to just 8.5 seconds at 64 km/h, as the Mercedes-Benz Level 3 DRIVE PILOT achieves.
• Current Radar Limitations: Traditional radar systems have some drawbacks, including poor resolution, false positives from sidelobe effects, and challenges in separating different objects.

Despite these issues, radar technology has advantages for self-driving, such as solid performance in bad weather and reliable long-range detection. To meet the needs of autonomous driving, radar systems must evolve in the following ways:

• Larger Radar Apertures: This will improve resolution and allow the detection of stationary objects without relying on speed. For example, Berkley, California-based company Zendar, is working closely with NXP to extract maximum value from existing radar technology through coherence between multiple radar sensors. This will create a synthetically larger aperture to improve resolution.
• Higher Density of Transmit and Receive Elements: More elements in the radar system will enhance dynamic range and reduce sidelobes, improving object separation.

Looking Ahead: Imaging Radar’s Potential to Challenge LiDAR

Imaging radar is expected to be adopted rapidly in the autonomous driving industry. It can provide significant improvements in resolution and range at a reasonable cost.

• Adoption in Vehicles: Large aperture radar will lead the market, with smaller corner radar following. Vehicles may feature between three and eight radar systems, depending on their application.
• Market Dynamics: While sparse imaging radar will initially serve Level 2+ vehicles, dense imaging radar will become crucial for unsupervised autonomous driving. This is where the technology must perform reliably without human oversight to compete with LiDAR.

Currently, imaging radar and LiDAR will serve different markets. Most imaging radar sales will be in Level 2+ vehicles, where LiDAR isn't commonly used. However, by 2030, dense array imaging radar may start to replace LiDAR in some Level 3 systems.

Key Companies

• Aptiv
• Arbe
• Continental
• Magna International, Inc.
• Mobileye
• NXP Semiconductors
• Texas Instruments Inc
• Valeo
• ZF Friedrichshafen AG

Conclusion

In summary, the evolution of automotive radar, particularly imaging radar, unlocks new autonomous driving experiences. While LiDAR currently plays a critical role, imaging radar's cost and performance advantages may challenge its market share in the years to come.

For more in-depth insights on this topic, download ABI Research’s report: Replacing Automotive LiDAR with Imaging Radar.