Automated rotational calibration of multiple 3D LIDAR units for intelligent vehicles

A new calibration method is introduced for automatically and simultaneously determining roll, pitch, and yaw offsets of multiple 3D Light Detection and Ranging (LIDAR) sensors from their theoretical values. The application focuses on autonomous ground vehicles with a LIDAR unit on each of four sides, but can be extended to other multi-LIDAR configurations as well. The data from the LIDAR units is combined to form a 3D point cloud representing the vehicle's surroundings. There are two parts to the calibration: the ground plane alignment that determines and corrects the roll and pitch offsets of each LIDAR independently, and the genetic algorithm based alignment that determines and corrects the yaw offsets of all LIDARs simultaneously. In the ground plane alignment, a scan from each LIDAR is used to calculate the roll and pitch offset of the sensed ground plane from the theoretical ground plane. A correction is applied to the data and a genetic algorithm is then used to concurrently determine the yaw offsets.