Effects of Target Poses on the Performance of Lidar-Camera Calibration

This work explores the enhancement of the Velo2Cam LiDAR-camera calibration algorithm by introducing a LiDAR rotation step to improve adaptability across different sensor configurations. The study also investigates optimal sensor-target-backwall poses and distances for effective calibration. The proposed method was validated using an Ouster OS0-64 LiDAR and a ZED 2i stereo camera in a lab setting. This method was validated using an Ouster OS0-64 LiDAR sensor and a ZED 2i stereo camera in a lab setting. The findings demonstrate successful calibration within specific distance ranges for various target sizes. The calibration is effective within 0.6-0.9 m for small targets and 0.6-1.8 m for medium targets, with optimal target-wall distances of 0.2-0.3 m and 0.2-0.7 m, respectively. This work contributes to a better understanding of LiDAR-camera calibration and provides practical insights for achieving user-friendly calibration processes. This approach simplifies calibration, reducing the need for recalibration after sensor misalignment.