Lidar-Based Fast 3D Stockpile Modeling

Abstract

A lidar-based 3D stockpile modeling system was designed and implemented, which allows the operator to quickly obtain a 3D stockpile model with detailed features by simply carrying a portable scanning device to circle around the target stockpile. The system uses a line scanning lidar to scan a stockpile with high-precision sensors to measure the position and attitude of itself in real time, and then integrates discrete scanning points to form a 3D point cloud of the stockpile. In data processing, the algorithm was optimized according to the measurement characteristics of the line scanning lidar and the structural characteristics of the stockpile. Noise reduction for the scanning points and ground points removal was performed concurrently in the measurement process. In addition, the point cloud was projected onto a 2D plane for a quick Delaunay triangulation, thus improving the speed of 3D stockpile modeling effectively. Field experiments showed that the system can complete 3D modeling of large stockpiles within a few minutes. The modeling time mainly depends on the walking speed of the operator who circles around the stockpile, and the waiting time for the 3D modeling after the measurement is generally less than 30s.