|
Cover Story |
|
|
Navigating Real and Virtual Worlds
Researchers create tools for more effic- | |
ARCHITECTS, COMPUTER SCIENTISTS and a roboticist are navigating spaces – both real ones and their virtual counterparts – and applying their findings to such diverse purposes as better office workstation layout and mapping of enemy buildings in urban warfare.
|
In one project, researchers are creating computer models to evaluate the use of space in the average office environment.
"Differences in the proportion of one floor plate (perimeter shape) to another might make a huge difference in the flexibility of furniture layout," explains Assistant Professor of Architecture and Computing Ruth Dalton, a GVU Center faculty member.
Dalton and her collaborators – Associate Professor John Peponis, Assistant Professor Sonit Bafna, computing Ph.D. student Markus Haas and architecture PhD. student Ermal Shpuza – are creating computer models for three types of offices, in particular the ubiquitous "cube farm." Using a generative algorithm that essentially generates a 3D layout, they are determining the most efficient of 50 different floor plates.
For the General Services Administration, which is funding this research, the savings could be substantial, Dalton says. Computer models could help the agency evaluate whether certain buildings are a better investment than others because of the flexibility of their floor plates.
In another study Dalton, roboticist and GVU Assistant Professor of Computing Frank Dellaert and computing Ph.D. students Spencer Charles Brubaker and Ananth Ranganathan are improving robots' ability to construct topological maps of the built environment.
"If a robot is navigating an unfamiliar environment, and it arrives at what it thinks is a new space, how does the robot evaluate this?" Dalton asks. "How does the robot distinguish between a space it has been to before, perhaps viewed from a different perspective, and an entirely new space?" The researchers' novel approach uses probabilistic methods and expert knowledge of environment types.
"If you have a number of key spaces, what is the likelihood that it's a new space, given a linear-path sequence through the building?" Dalton says. "What we know about the nature of buildings may begin to help us reduce the probabilities."
This tool might be helpful to military forces sending a robot into hostile territory to survey a building or environment and hence construct a crude map, Dalton says. In addition to military scenarios, this work would also be useful for search and rescue operations, office tasks and cleaning, Dellaert adds.
Researchers hope to conduct a follow-up study as their initial results are promising,
For more information, contact Ruth Dalton, 404-385-2913 or
ruth.dalton@arch.gatech.edu; or Frank Dellaert, 404-385-2923 or frank@cc.gatech.edu.
Contents
Research Horizons
GT Research News
GTRI
Georgia Tech
Send questions and comments regarding these pages to
webadmin@edi.gatech.edu
Last updated: Dec.11, 2003