A SEMANTICALLY CLASSIFIED GEO-SPATIAL 3D OCTREE VOXEL BASED SYSTEM FOR GEOTECHNICAL SITE CHARACTERIZATION

Geotechnical site characterization is the process of collecting geophysical and geospatial characteristics about the surface and subsurface to create a 3-dimensional (3D) model. Current Robot Operating System (ROS) world models are designed primarily for navigation in unknown environments; however, they do not store the geotechnical characteristics requisite for environmental assessment, archaeology, construction engineering, or disaster response. The automotive industry is researching High Definition (HD) Maps, which contain more information and are currently being used by autonomous vehicles for ground truth localization, but they are static and primarily used for navigation in highly regulated infrastructure. Modern site characterization and HD mapping methods involve survey engineers working on-site followed by lengthy post processing. This research addresses the shortcomings for current world models and site characterization by introducing Site Model Geospatial System (SMGS). This site model leverages an octree spatial data model to store heterogeneous geotechnical information in a Volumetric Pixel (Voxel) grid, which allows for more efficient algorithms in data analysis and fusion. SMGS provides a real-time, dynamically updated, 3D data model with semantically derived costmaps for navigation and Engineer operations, ground truth localization without GPS, and produces standard Geographic Information System (GIS) maps.

Citation: M. Richards, K. Murphy, I. Lopez Toledo, A. Soylemezoglu, “A Semantically Classified Geo-spatial 3D Octree Voxel Based System for Geotechnical Site Characterization,” In Proceedings of the Ground Vehicle Systems Engineering and Technology Symposium (GVSETS), NDIA, Novi, MI, Aug. 15-17, 2023.