Evaluation of calibration methods to construct a 3-D environmental map with good color projection using both camera images and laser scanning data

Abstract

In mobile robot navigation, restoration of the environment around the robot in a 3-D map is necessary for self-location, route planning, and detecting surrounding obstacles. We construct the 3-D color environmental map using both camera images and laser scanner (LIDAR) point cloud data. In the map, the RGB values are provided to the LIDAR point cloud data by projecting the point cloud onto the simultaneously acquired image. The projection parameters can be determined by measuring the calibration boards using both the camera and the LIDAR. In this paper, we have constructed a method to evaluate the accuracy of projection applicable to any calibration methods. And, then, we found that the calibration points in the central position of an image are important to obtain good projection parameters and that additional points at side positions also can improve the accuracy of the projection.