Out Door Map Feature Point Generating Mechanism Design and Algorithm Robust to Lidar Sparsity

Abstract

In this paper, we introduce the software/hardware system that can reliably calculate the position of the model from the sensors regardless of point cloud density in an outdoor environment. 3D point cloud map is used in various fields such as construction environment and autonomous driving as it provides 3D information. As different methodologies and sensors are used in the process of generating a 3D point cloud map, the density of the result should be inevitably different. Generation of specific points is essential as evaluating accuracy of a 3D point cloud map or merging different 3d point cloud maps can be done based on these reference point. Currently, method of placing an object such as a band with high reflectivity at a specific location and manually extracting a point that is judged to represent the object is used to generate a specific point in the point clud. The result of the manual process has a significant difference depending on the Lidar sparsity level and the person performing the process. To overcome these problems, this paper presents a hardware design that can generate specific points in the 3D point cloud map independent of the robust outdoor environment and Lidar sparsity. As will be shown, the system performance is verified at both indoor environment and outdoor environment. Furthermore, two different 3D point cloud map generating methods with different density levels of the output were used to verify the methodology.