Accurate Calibration of LiDAR-Camera Systems using Ordinary Boxes

This paper deals with the calibration of a visual system, consisting of RGB cameras and 3D Light Detection And Ranging (LiDAR) sensors. Registering two separate point clouds coming from different modalities is always challenging. We propose a novel and accurate calibration method using simple cardboard boxes with known sizes. Our approach is principally based on the detection of box planes in LiDAR point clouds, thus it can calibrate different LiDAR equipments. Moreover, camera-LiDAR calibration is also possible with minimal manual intervention. The proposed algorithm is validated and compared to state-of-the-art techniques both on synthesized data and real-world measurements taken by a visual system consisting of LiDAR sensors and RGB cameras.