Extrinsic Calibration between a Rotating Laser Rangefinder and a Camera Using a Photogrammetric Control Field and Ping Pong Balls

The combination of 2D laser rangefinder(LRF) and camera has many application in robotics, mapping,self-driving vehicles to map and capture  color and texture information of objects. to fuse LRF and camera’s data, these  sensors must be carefully calibrated related to each other. Extrinsic calibration between a LRF and camera is often performed by common features. Therefore extrinsic calibration between LRF and camera is necessary process.  Extrinsic calibration between a Camera and a LRF  is often performed by common features in data captred by two sensors. In this research, three difrent methods for extrinsic calibration between a LRF and a camera using photogrammetric control field and ping pong balls are presented. In calibration process using 3D and 2D point clouds,ping pong balls are used as common targets that can be identified in the camera and LRF data. In calibration using the 3D point clouds, the calibration parameters are calculated using the point clouds generated from the test field at the main station and performing the bundel adjusment using a set of images taken from the control field. In this method, a 3D point cloud is obtained from combination of 2D LRF and servo motor data. The 2D LRF  is connected to the servo motor by a gimbal so that the extension of the servo motor shaft pass through the center of the LRF. The LRF rotation is measured by the encoder in the servo motor, and by rotating the LRF, all 2D scans at a station can be registered relative to each other. On the other hand, through images taken from the control field and bundel adjusment technoque, the position of the balls in the photogrammetric model is obtained. The second method is similar to the first method,except that a 2D point cloud is used instead of a 3D point cloud. In the first and second methods, the coordinates of the centers of ping pong balls are available in two different coordinate systems, so the relationship between the two coordinate systems with the 3D conformal equation is obtained. In the third method, through the pyramid created in the corner of the room and solving the perspective-three-point problem,  the 2D LRF calibration parameters at the main station relative to the control field are calculated with just one scan. The calibration parameters of camera in the main station relative to the control field are calculated using bundel adjusment and 3D conformal equations. Finally, the calibration parameters of the LRF relative to the camera are obtained. According to the RMSE calculated for checkpoints, calibration using 3D point cloud,calibaration using room corner and calibration using 2D point cloud were the most accurate, respectively.