Direct Georeferencing 3D Points Cloud Map Based on SLAM and Robot Operating System

Abstract

Recently, 3D data has undergone an explosion in demand by many applications such as the creation of 3D maps, road condition monitoring, and autonomous navigation as well. The concept of collecting geospatial data from the road environment has been recognized by the term “Mobile Mapping”, which represents a cost-efficient and quick method of acquiring data from dynamic fields such as highways, underground streets, and urban zones, where the process of data collection has been difficult due to the road traffic or to bad weather conditions. Traditional mobile mapping systems can face several problems including reliability issues and poor accuracy of produced data, particularly in the GNSS outage case. For this reason, the integration of the Simultaneous Localization and Mapping (SLAM) method in the Mobile Mapping System (MMS) represents a robust mapping solution in complex environments to optimize the accuracy of the generated data even in the case of GNSS signal loss. The present paper proposes an efficient approach for collecting, modeling, and generating large-scale 3D point cloud maps with real-world coordinates based on the SLAM algorithm and Robot Operating System (ROS). The proposed approach can effectively generate high-definition georeferenced point cloud maps with an accuracy up to ± 5 cm.