1. What is Angular Resolution?
Angular resolution refers to the angle formed by the central axes of two adjacent laser detection points at the ranging center during the scanning and detection process of a LiDAR. Depending on the scanning direction, it can be divided into horizontal angular resolution and vertical angular resolution.

Horizontal angular resolution: The angle between the central axes of adjacent detection points at the ranging center along the azimuth direction. It determines the LiDAR's ability to resolve lateral targets (such as vehicles, pedestrians, and road shoulders).
Vertical angular resolution: The angle between the central axes of adjacent detection points at the ranging center along the elevation direction. It determines the LiDAR's ability to resolve height information (such as the ground, curbs, and obstacle heights).

The physical significance of angular resolution can be intuitively reflected by the spatial resolution size:
At a target distance d, the minimum resolvable size $\Delta L$ corresponding to the angular resolution $\theta$ satisfies:
$$\Delta L = d \cdot \tan(\theta) \approx d \cdot \theta \quad (\theta \text{ 以弧度计})$$ For the same angular resolution, the farther the target distance, the larger the minimum resolvable size and the weaker the spatial resolution capability; for the same target distance, the higher the angular resolution (the smaller the value), the smaller the minimum resolvable size and the stronger the spatial resolution capability.

Taking Figure 1 as an example, at a distance of 30 meters: a horizontal resolution of 0.10° can resolve an object of about 5 cm; a vertical resolution of 1° can resolve an object of about 50 cm. This indicates that a higher angular resolution means a stronger ability to capture details.

If a LiDAR is compared to a camera, angular resolution is equivalent to pixel density. For example, at a distance of 200 meters, under low angular resolution, a pedestrian might be covered by only a few laser points, resulting in insufficient features and making it difficult to distinguish. However, under high angular resolution, more than a dozen points can be distributed on the pedestrian, allowing the system to clearly see the contours of the person's limbs, thereby achieving precise recognition.

2. Influencing Factors
(1) Scanning Timing and Sampling Matching:
Angular resolution is affected by the synchronization between the scanning mechanism (motor/micro-mirror) and the laser point frequency. If the step angle (sampling interval) is too large, the point cloud distribution will be too sparse, causing small distant targets to fall into the gaps of the point cloud, resulting in information loss and perception blind spots; if the step angle is too small, it will lead to data redundancy and cannot effectively improve resolution when exceeding the physical limit. Testing must ensure that the actual step angle is consistent with the theoretical design to ensure the uniformity of spatial sampling.

As shown in Figure 2 [2], the more vertical channels, the higher the laser point frequency and the smaller the vertical step angle, resulting in a denser point cloud and higher vertical angular resolution, which restores object and scene details more clearly.

(2) Physical Characteristics Constraints of the Laser Beam:
The laser divergence angle determines the physical upper limit of resolution. As the distance increases, an excessively large divergence angle will cause physical overlap and echo aliasing of adjacent spots in space, making the system unable to distinguish two closely adjacent obstacles (such as a distant pedestrian and a lamppost), thereby weakening the spatial resolution capability of the LiDAR.

As shown in Figure 3 [3], when the aircraft altitude is 1000 m, the laser divergence angle on the left is 0.7 mrad, and the ground spot diameter is 0.7 m; while the laser divergence angle on the right is 0.3 mrad, and its ground spot diameter is 0.3 m. This indicates that the larger the laser divergence angle, the larger the spot at the same distance, and the larger the minimum resolvable target size, thereby limiting the spatial resolution of the system.

(3) Scanning Pointing and Control Accuracy:
The linearity of angle control in the scanning mechanism and the assembly tolerances of optical components directly affect pointing accuracy. If dynamic jitter or nonlinear deviation occurs in the system, it will lead to interleaved point cloud arrangement and blurred edge contours, thereby reducing the accuracy of object geometric feature extraction by autonomous driving algorithms.

As shown in Figure 4 [4], in Figure (a), the point cloud within the red box exhibits obvious interleaved arrangement and edge distortion, and the originally smooth road edge presents a jagged misalignment. In Figure (b), the point cloud of the same road section is neatly arranged with clear edges and accurate contour restoration. This intuitively reflects the impact of the accuracy of the scanning mechanism and optical components on pointing accuracy: insufficient accuracy will lead to point cloud distortion and reduce the perception accuracy of autonomous driving.

3. How to Measure Angular Resolution?
Angular resolution is a core performance indicator of automotive LiDAR, directly determining the density of the point cloud and the ability to extract target geometric features. The following details the test conditions, operational procedures, and determination methods for angular resolution in accordance with the relevant provisions of GB/T 45500-2025. {Note: This test method is not applicable to LiDAR using MEMS technology.}

(1) Test Conditions:
Environment Setup:

Recommended Equipment:
The Automotive LiDAR Automated Testing System independently developed by Yanding matches the test methods specified in the national standard GB/T 45500-2025. It can fully automatically complete test items such as ranging and angular resolution, achieving one-stop standardized verification from equipment control and data collection to report output. The core equipment used in this test, including the 90% diffuse reflectance target, automatic chart switching bracket, and moving guide rail, are all integrated into this system.

Parameter Requirements:

Division of 9 Fields of View:

(2) Operational Procedures:

(a) According to the test setup in Figure 5, install the LiDAR on a high-precision turntable. At a horizontal distance of $(d_{\text{Nmin}} + 1\ \text{m})_{-5}^{+5}\ \text{cm}$m from the LiDAR, place a 90% diffuse reflectance target perpendicular to the ground (recommended size is 1 cm × 1 cm). Connect the LiDAR to a point cloud visualization device to observe the point cloud data output by the LiDAR in real time.
(b) Rotate the LiDAR around the ranging center to bring the diffuse reflectance target into a certain field of view area of the LiDAR. Slowly rotate the LiDAR horizontally, and record the angle rotated by the high-precision turntable when the point cloud hitting the left or right edge of the diffuse reflectance target changes from the “2nd point/column of point cloud” to the “3rd point/column of point cloud” (i.e., switching from Scene A to Scene B, refer to Figure 8).
© Repeat the above measurement steps no less than 5 times, and take the average of the multiple measured angles as the horizontal angular resolution for this field of view area.
(e) Rotate the LiDAR vertically in the same way, record the angle rotated by the high-precision turntable when one point or a column of point cloud at the upper or lower edge of the diffuse reflectance target changes, repeat the measurement no less than 5 times, and calculate the average of the angles rotated by the high-precision turntable as the vertical angular resolution for this field of view area.
(f) Repeat operations b to d to measure the horizontal and vertical angular resolution of other field of view areas.

(3) Result Analysis
When the difference between the angular resolution of the LiDAR and the nominal value is not greater than 0.25 times the nominal value, it meets the standard requirements.

Image Sources:
[1] https://www.analog.com/en/resources/analog-dialogue/articles/lidar-for-autonomous-system-design-object-classification-or-object-detection.html
[2] https://inertiallabs.com/wp-content/uploads/2024/09/Screenshot-2024-09-10-162547.jpg
[3] https://www.researchgate.net/figure/llustration-of-LIDAR-beam-divergence-Horizontal-and-vertical-distances-are-drawn-in_fig2_237372271
[4] https://www.researchgate.net/figure/An-example-of-point-cloud-distortion-correction-a-The-point-cloud-before-correction_fig13_331102582