Automotive and Industrial RF Automotive Radar Informational

What is the required angular resolution for an automotive radar to support autonomous driving?

Q: Can radar ever match LiDAR angular resolution?

Conventional radar cannot match the 0.05-0.1 degree resolution of current automotive LiDAR because the RF wavelength at 77 GHz (3.9 mm) is about 4,000x longer than LiDAR wavelength (905-1550 nm). However, radar provides velocity measurement that LiDAR does not, and imaging radar with thousands of virtual channels can achieve 0.5-1 degree resolution that is sufficient for most autonomous driving perception tasks.

Q: Why is elevation resolution important for autonomous driving?

Elevation resolution prevents false alarms from overhead objects (bridges, signs, overpasses) that appear as obstacles to a 2D (azimuth-only) radar. Without elevation capability, the radar cannot distinguish a sign overhanging the road from a vehicle stopped in the lane. False braking events from overhead objects were a significant problem for early ADAS radar systems.

Q: What is the minimum number of MIMO channels for 4D imaging radar?

Practical 4D imaging radar requires at least 48-96 virtual channels (e.g., 12TX x 8RX = 96) to achieve useful resolution in both azimuth and elevation simultaneously. The most capable production systems use 192+ virtual channels (e.g., 4-chip cascade with 12TX x 16RX). Research systems from Arbe and others use 2,000+ virtual channels for ultra-high resolution imaging.

The required angular resolution for an automotive radar to support autonomous driving depends on the autonomy level and the specific detection task. For Level 2 ADAS functions like adaptive cruise control in a single lane, an azimuth resolution of 3-5 degrees is sufficient to detect and track vehicles directly ahead. For Level 3+ highway automation requiring multi-lane awareness, an azimuth resolution of 1-2 degrees is needed to distinguish vehicles in adjacent lanes at 100+ meter range (lane width of 3.7 meters at 100 meters subtends approximately 2.1 degrees). For Level 4/5 autonomous driving in urban environments, sub-1 degree azimuth resolution and 2-5 degree elevation resolution are required to classify objects (pedestrians, cyclists, vehicles, infrastructure), separate closely spaced targets (parked cars, pedestrians near vehicles), and detect elevated objects (overpasses, signs) versus ground-level obstacles. This sub-degree resolution drives the industry toward 4D imaging radar with MIMO virtual arrays of 100+ virtual channels, achieving angular resolution approaching 1 degree in both azimuth and elevation. The elevation dimension is critical for distinguishing overhead objects (bridges, signs) from road-level obstacles, preventing false emergency braking events.

Category: Automotive and Industrial RF

Updated: April 2026

Product Tie-In: Radar ICs, PCB Materials, Antennas

Angular Resolution Requirements for Autonomous Driving Radar

Angular resolution is arguably the most critical radar parameter for autonomous driving because it determines the radar's ability to create a detailed image of the driving environment, distinguishing individual objects rather than merging nearby targets into a single unresolvable cluster.

Resolution Requirements by Autonomy Level

Level 2 (ADAS): 3-5 degrees azimuth resolution. Adequate for in-lane target tracking (ACC, AEB). No elevation resolution required for basic functions
Level 2+ (Highway Assist): 1.5-3 degrees azimuth. Need to resolve adjacent lane vehicles for lane change assist. Limited elevation capability desirable
Level 3 (Conditional Automation): 1-2 degrees azimuth, 3-5 degrees elevation. Multi-lane awareness, overhead object discrimination, basic object classification
Level 4/5 (Full Autonomy): < 1 degree azimuth, 2-3 degrees elevation. Full scene understanding, pedestrian detection and classification, urban environment mapping

4D Imaging Radar

4D imaging radar measures range, velocity, azimuth angle, and elevation angle simultaneously, creating a 3D point cloud similar to LiDAR but with added velocity information. This requires large MIMO virtual arrays (86-192+ virtual channels) achieved through cascade transceiver architectures. Companies like Arbe/Vayyar (2304 virtual channels), Continental (ARS540, 192 channels), and ZF (imaging radar) are pushing angular resolution below 1 degree in both planes.

Angular Resolution vs MIMO Array Size

Angular resolution: theta_3dB ~ lambda / (N x d x cos(theta_scan))
At 77 GHz (lambda = 3.9 mm), N = 86 virtual elements, d = lambda/2:
theta_3dB ~ 3.9mm / (86 x 1.95mm) = 1.3 degrees
Cross-range resolution at range R: delta_cr = R x theta_3dB [radians]
At R = 100 m, theta = 1.3 deg: delta_cr = 2.3 m

Common Questions

Frequently Asked Questions

Can radar ever match LiDAR angular resolution?

Conventional radar cannot match the 0.05-0.1 degree resolution of current automotive LiDAR because the RF wavelength at 77 GHz (3.9 mm) is about 4,000x longer than LiDAR wavelength (905-1550 nm). However, radar provides velocity measurement that LiDAR does not, and imaging radar with thousands of virtual channels can achieve 0.5-1 degree resolution that is sufficient for most autonomous driving perception tasks.

Why is elevation resolution important for autonomous driving?

Elevation resolution prevents false alarms from overhead objects (bridges, signs, overpasses) that appear as obstacles to a 2D (azimuth-only) radar. Without elevation capability, the radar cannot distinguish a sign overhanging the road from a vehicle stopped in the lane. False braking events from overhead objects were a significant problem for early ADAS radar systems.

What is the minimum number of MIMO channels for 4D imaging radar?

Practical 4D imaging radar requires at least 48-96 virtual channels (e.g., 12TX x 8RX = 96) to achieve useful resolution in both azimuth and elevation simultaneously. The most capable production systems use 192+ virtual channels (e.g., 4-chip cascade with 12TX x 16RX). Research systems from Arbe and others use 2,000+ virtual channels for ultra-high resolution imaging.

Keep Reading