Growing demand for digitizing indoor and outdoor environments
Applications 3D content generation for mapping service applications, e.g. google maps, bing maps video games based on real locations interactive museum exhibits Pre-visualization and virtual sets for the film industry Virtual Reality Systems Realistic simulators for robotic algorithms (e.g. ROS/gazebo)
Goal: autonomous system for efficient creation of photo realistic 3D models of indoor and outdoor environments
Robot hardware including range and imaging sensors Frontier based exploration for active environment exploration 2D mapping and localization (Grisetti et. al, 2005) 5MP Camera Scanning Lidar
prev. 3D scans next scan trajectory
Stereo Cameras
SLR Camera
Current location
Tilting Lidar Pan-tilt unit
frontier
Exploration Lidar Base Lidar
Bosch’s 3D Mapping Robot Research and Technology Center North America
Reconstruction of material properties (e.g. color, specularity, temperature) Inference of material types (e.g. wood, plastic, vegetation) Enhancing visual impression by adding more realism
Surface Model Research and Technology Center North America
Color Reconstruction (1) 1. Surface partitioning Surface segmentation into nearly planar regions Merging of similar regions using adjacency graphs 2. Unfolding Mapping of surface into texture domain Finding a conformal mapping which preserves the angular distortion Minimizing the complex conformal energy (Levy et al., 2002):
3. Color Reconstruction Unprojection of 3D points into camera images Research and Technology Center North America
Office model (50m x 140m) 28M Vertices, 55M Faces 772 MByte Geometry 600 MByte Texture
Generation of scene graphs by iteratively dividing 3D model Model simplification of scene graph nodes to limit number of elements per node View-dependent culling and loading
