Automated, 3D Modeling of Building Interiors

Three-dimensional models of objects, sites, buildings, structures, and urban environments made of geometry and texture are important in a variety of civilian and military applications such as training and simulation for disaster management, counteracting terrorism, virtual heritage conservation, virtual museums, historical sites documentation, mapping of hazardous sites and underground tunnels, and modeling of industrial and power plants for design verification and manipulation.

While object modeling has received a great deal of attention in recent years, 3D site modeling, particularly for indoor environments, poses significant challenges. The main objective of this proposal is to design, analyze, and develop architecture and algorithms, as well as associated statistical and mathematical framework for a human operated, portable, 3D indoor/outdoor modeling system, capable of generating photo-realistic rendering of the internal structure of multi-story buildings as well as external structure of a collection of buildings in a campus.

Key technical challenges consist of system architecture, sensor choice, calibration, and error characterization, local and global localization algorithms, sensor integration, registration, and fusion, and complete and accurate coverage of all details. Using a human-operated backpack system equipped with 2D laser scanners, cameras, and inertial measurement units, we develop four scan-matching-based algorithms to localize the backpack and compare their performance and tradeoffs. In addition, we show how the scan-matching-based localization algorithms combined with an extra stage of image-based alignment can be used for generating textured, photorealistic 3D models. We present results for two datasets of a 30-meter-long indoor hallway.

Full 3D Models Available for Download

Download 3D models for both datasets 1 and 2 here. Note zip file size is approx 280MB.

The zip file includes two 3D models, one for dataset 1 and one for dataset 2. Both datasets were acquired on July 1, 2009, on the fourth floor of Cory Hall at UC Berkeley. The two datasets were acquired by two different human operators with two different walking gaits. Each model is a textured, triangular mesh saved as a .ive file, which is a native format of the software osgviewer. To view these models first download OpenSceneGraph. Once you have downloaded OpenSceneGraph, in Windows you should open a command prompt, cd to the directory in which the .ive files are stored, and enter a command of the form "osgviewer cory4_20090701_setX_blended.ive".

Screenshots of 3D Models

Here are screenshots of the blended 3D model for dataset 1.

References

G. Chen, J. Kua, S. Shum, N. Naikal, M. Carlberg, and A. Zakhor. "Indoor Localization Algorithms for a Human-Operated Backpack System," submitted to 3D Data Processing, Visualization, and Transmission 2010, Paris, France, May 2010. [Adobe PDF]

N. Naikal, J. Kua, G. Chen, and A. Zakhor. "Image Augmented Laser Scan Matching for Indoor Dead Reckoning," IEEE/RSJ International Conference on Intelligent RObots and Systems (IROS), St. Louis, MO, October 2009. [Adobe PDF]