• Development of an intrinsic calibration method for solid-state LiDARs, including a concentric circular feature target, an asymmetric Gaussian feature detection algorithm and overexposure correction. • Deployment on an accelerated real life LiDAR test bench, characterizing intrinsic parameter drifts over temperature and lifetime. • Impact analysis of intrinsic parameter drift on 3D mapping accuracy. Reliable 3D perception is vital for automated driving, with LiDAR sensors offering superior detection accuracy compared to other active sensors like Radar. However, precise LiDAR calibration directly impacts detection accuracy. This work examines the effects of temperature and aging on the intrinsic projection model of automotive Flash LiDARs, as well as its influence on spatial mapping accuracy. For that, accelerated lifetime tests were performed on three sensors across 0 ∘ C – 80 ∘ C and a novel calibration approach was developed, which achieves projection errors of less than 0.2 pixel by using 2D Gaussian feature fitting, overexposure correction, and model-based pose estimation. The results show negligible drift related to aging, but indicate temperature-induced, elastic model parameter drifts, which show small variations between all three test devices and cause up to 130 mm translational detection error in 25 m measurement distance. While long-term intrinsic recalibration appears unnecessary, temperature compensation may be needed for high-precision applications. Thus, the findings provide critical insight for LiDAR deployment in automated transportation systems. In addition, the developed calibration method can be utilized for precise extrinsic calibration in vehicle production or to compensate for sensor misalignment over lifetime.
Kettelgerdes et al. (Fri,) studied this question.