Real-Time Point Cloud Fusion of Multi-LiDAR Infrastructure Sensor Setups with Unknown Spatial Location and Orientation

The use of infrastructure sensor technology for traffic detection has already\nbeen proven several times. However, extrinsic sensor calibration is still a\nchallenge for the operator. While previous approaches are unable to calibrate\nthe sensors without the use of reference objects in the sensor field of view\n(FOV), we present an algorithm that is completely detached from external\nassistance and runs fully automatically. Our method focuses on the\nhigh-precision fusion of LiDAR point clouds and is evaluated in simulation as\nwell as on real measurements. We set the LiDARs in a continuous pendulum motion\nin order to simulate real-world operation as closely as possible and to\nincrease the demands on the algorithm. However, it does not receive any\ninformation about the initial spatial location and orientation of the LiDARs\nthroughout the entire measurement period. Experiments in simulation as well as\nwith real measurements have shown that our algorithm performs a continuous\npoint cloud registration of up to four 64-layer LiDARs in real-time. The\naveraged resulting translational error is within a few centimeters and the\naveraged error in rotation is below 0.15 degrees.\n