VR accommodates reality

By Eric Smalley, Technology Research News

One reason flight simulators are much more compelling than other virtual environments is that only the scenery is virtual. The cockpits are often exact replicas of the real things.

For many training, design verification, telepresence and phobia treatment applications, the ideal environment is a mix of virtual and real.

Researchers from the University of North Carolina and Disney Corporation have advanced their method for representing real objects in virtual environments by devising a way for real and virtual objects to interact.

The researchers tested the system by having NASA engineers simulate a space shuttle payload assembly task. "The ideal virtual environment system would have the participant fully convinced he was actually performing the task being simulated," said Benjamin Lok, now an assistant professor of computer and information science and engineering at the University of Florida. "Parts and tools would have mass, feel real, and handle properly with appropriate visual and haptic feedback."

The principal drawback to today's fully virtual environments is that there is nothing to feel and to constrain the user's movements. "Imagine trying to simulate a task as basic as unscrewing an oil filter from a car engine in a virtual environment," said Lok. The researchers' goal is to expand the applicability of virtual environments, he said.

They have taken a step toward mixed-reality environments by allowing real and virtual objects to coexist in a shared virtual space. The approach is the inverse of augmented reality systems, which incorporate a small number of virtual objects in real environments by, for example, projecting interactive images onto a desk.

In one example of the researchers' hybrid virtual environment, a user can part a real window curtain so that she feels curtains. At the same time she sees virtual representations of her hands moving virtual curtains to reveal a virtual window and a virtual scene beyond.

The heart of the researchers' system is a method for determining when real and virtual objects collide and providing a plausible response, said Lok.

The key is keeping the virtual representations of the real objects as simple as possible. The system uses four cameras and object recognition software to determine the shapes and positions of real objects in the environment. The camera data is used to generate virtual three-dimensional shells in the shapes of the real objects, and the shells are forbidden zones for all of the virtual objects in the environment.

Virtual objects can have virtual properties like velocity, acceleration and deformability, or give. Calculating these properties in real-time for the representations of the real objects would be extremely difficult, so only virtual objects move or change shape in reaction to collisions. "This allowed real objects to be integrated into a virtual environment without additional modeling or tracking," said Lok.

When a virtual object and a shell collide, the system determines the point of contact and shifts the virtual object to keep the shell and the object from overlapping.

The researchers tested the hybrid virtual environment with a group of NASA engineers. The tests show that the environment is more effective for evaluating hardware designs and for planning assembly tasks than fully virtual environments, said Lok.

The engineers determine the optimal design for payloads like satellites and scientific equipment, balancing the need to take up as little room as possible against the requirements for technicians to be able to assemble the payload systems.

The researchers simulated a shuttle payload in a hybrid environment. "We mocked up a common payload integration task and asked [the engineers] to estimate how much space was required to perform the task," said Lok. The researchers set the simulation to those spacings and had the engineers try it out, he said.

The task consisted of fitting a real tube into a hole in a virtual box, attaching the tube to a real mount on a real table beneath the virtual box, feeding a real cable through the tube and plugging the end of the cable into a real socket on the table. The virtual box was surrounded by virtual walls that constrained the engineer's movements by flashing red whenever any of the real objects, including the engineer's hands, came into contact with the walls.

The simulation showed that the engineers' estimates were off. On average, the engineers underestimated the required spacing by an average of 5.6 centimeters, according to the researchers. The engineers were surprised that both additional space and a tool were required to complete to task, said Lok.

Hybrid environments provide greater realism than fully virtual environments, and because they are easier to set up than full mockups of payloads, they can be used earlier in the design process when changes are easier to make, Lok said.

It will be about five years before hybrid reality systems are available for specialized applications, and 20 years for general applications, which require quick reconstruction of real objects in virtual environments, said Lok. Computer vision and image-based rendering researchers would call these problems "holy grails in their fields," he said. "Perhaps in 20 years we can have very compelling hybrid realities with production quality solutions on real-world industrial applications."

The researchers are working on generating better virtual representations of real objects, using hybrid realities, and determining the role of real objects in effective virtual environments, according to Lok.

Lok's research colleagues were Samir Naik of Disney Corporation, and Mary Whitton and Frederick P. Brooks Jr. of the University of North Carolina at Chapel Hill. The researchers presented the work at the Association of Computing Machinery (ACM) Symposium on Interactive 3D Graphics, held in Monterey, California, April 27 to 30, 2003.

The research was funded by L-3 Communications Corporation, the Office of Naval Research (ONR), the National Institutes of Health (NIH), the National Science Foundation (NSF), the University of North Carolina at Chapel Hill and the University of North Carolina at Charlotte.

Timeline:   5 years
Funding:   Government, University
TRN Categories:  Data Representation and Simulation; Human-Computer Interaction
Story Type:   News
Related Elements:  Technical paper, "Incorporating Dynamic Real Objects into Immersive Virtual Environments," Association of Computing Machinery (ACM) Symposium on Interactive 3D Graphics, April 27-30, 2003, Monterey, California


July 30/August 6, 2003

Page One

VR accommodates reality

Fractals support growing organs

Eyes off, screen off

Chip senses trace DNA

News briefs:
Laser bursts pierce fog
Electricity loosens tiny bits
Nano light stores data in polymer
See-through magnets hang tough
Munching microbes feed fuel cell
Crystal cracks nurture nanowires


Research News Roundup
Research Watch blog

View from the High Ground Q&A
How It Works

RSS Feeds:
News  | Blog  | Books 

Ad links:
Buy an ad link


Ad links: Clear History

Buy an ad link

Home     Archive     Resources    Feeds     Offline Publications     Glossary
TRN Finder     Research Dir.    Events Dir.      Researchers     Bookshelf
   Contribute      Under Development     T-shirts etc.     Classifieds
Forum    Comments    Feedback     About TRN

© Copyright Technology Research News, LLC 2000-2006. All rights reserved.