라이다-자동차 장착오차 자동 검사 시스템 개발
Accurate Alignment Inspection System for Low-Resolution Automotive LiDAR
We have developed an automatic and accurate inspection system for LiDAR alignment for application in the automobile industry.
A misalignment of LiDAR as low as a few degrees could cause a significant error in obstacle detection and mapping that could cause safety and quality issues.
The proposed methods estimate the orientations and the horizontal position of the LiDAR attachment with sub-degree and millimeter level accuracy.
(기법 #1) 1D Planar Target based LiDAR Alignment Inspection
Oh, Seontaek, Ji-Hwan You, Azim Eskandarian, and Young-Keun Kim. “Accurate Alignment Inspection System for Low-Resolution Automotive LiDAR.” IEEE Sensors Journal 21, no. 10 (2021): 11961-11968.
It is a static method for the stationary target board and vehicle during the inspection.
Also, it only requires one pose of the target board and does not utilize other sensors, such as a camera or GPS.
The experimental results for a low-resolution 16 channel LiDAR (Velodyne VLP-16) confirmed that misalignment could be estimated with accuracy within 0.2 degrees and 4 mm.
(기법 #2) 1-D Photodetector Arrays based LiDAR Alignment Inspection
You, Ji Hwan, Seontaek Oh, Jae Eun Park, Hyeongseok Song, and Young Keun Kim, ‘A Novel LiDAR Sensor Alignment Inspection System for Automobile Productions Using 1-D Photodetector Arrays’, Measurement, 183 (2021), 109817 https://doi.org/10.1016/J.MEASUREMENT.2021.109817
LiDAR extrinsic calibration using a target board with embedded photodetector arrays.
LiDAR sensor pose estimated within 0.1 degree and 3 mm of precision.
Requires a single target at static pose, applicable to manufacturing automobiles.
(기법 #3) Cubic Target based LiDAR Alignment Inspection
Song, H. S., You, J. H., Park, J. E., Eskandarian, A., & Kim, Y. K. (2021). Automatic inspection system for automotive LiDAR alignment using a cubic target. IEEE Sensors Journal, 22(3), 2793-2801.
The proposed inspection system is a one-shot and static method that uses only a single cubic-shaped target at a close range of less than 2 m.
The proposed method with a low-resolution LiDAR (Velodyne VLP-16) resulted in measurement errors of less than 0.04◦ and 1 mm for rotation and translation alignment, respectively.
(기법 #4) NIR Camera based LiDAR Alignment Inspection
H. -S. Song and Y. -K. Kim, "Automatic One-Shot LiDAR Alignment Inspection System Using NIR Camera," in IEEE Sensors Journal, vol. 23, no. 7, pp. 7798-7805, 1 April1, 2023, doi: 10.1109/JSEN.2023.3245219.