Conclusions

This paper has presented a framework for the reconstruction of textured 3D maps with an autonomous mobile robot, a 3D laser range finder and two pan-tilt color cameras. The developed systems allows to gage environments in 3D and fuse the data with camera images. A wide range of applications using 3D models benefit from the proposed automatic acquisition method, e.g., virtual reality applications, architecture, factory and facility management and rescue and inspection robotics.

Needless to say, much work remains to be done. Future work will concentrate on four aspects:

• Build 3D maps with texture information involving 6D robot poses, including loop tours, i.e., 6D SLAM [#!IAV!#].
  
• Use local color/texture correction in addition to our global correction method.
  
• Calibrate the camera parameters dynamically without the chess board quad, i.e., calculate it from the 3D scene and the corresponding images.
  
• Generate high level descriptions and semantic maps of environments including the 3D information, e.g., in XML format. The semantic maps contain spatial 3D data with descriptions and labels [#!IAS2004!#].