Development of optimized point cloud merging algorithms for accurate processing to create earthwork site models

Because of the nature of earthwork, there are difficulties in improving productivity due to situational variability and task complexity. Therefore, automation is receiving attention as a method to improve productivity. This study was conducted as a part of the IES (Intelligent Excavator System). The IES (Intelligent Excavator System) is intended for development of unmanned and remote controlled excavation systems for earthwork. The purpose of this study was to develop algorithms to merge two models acquired by a 3D laser scanner and Stereovision system in real time. To develop accurate and fast merging algorithms, ICP (Iterative Closest Point)-based merging algorithms and normal vectors were applied. Algorithms to merge the two models and to reflect the characteristics of the necessary earthwork will be proposed.

► We acquire a global model and local models for merging scan data and local model.
► Developed algorithms will increase merging accuracy for scan data and local model.
► The algorithm will increase efficiency of heterogeneous scan data.