VR
tool keeps line of sight in hand
By
Ted Smalley Bowen,
Technology Research NewsOne of the challenges of navigating computer-generated three-dimensional environments is figuring out how to avoid bumbling into things or losing sight of what you're interested in.
Three-dimensional simulations could make it easier to see and work with many types of information, from architectural designs to more abstract quantities like economic data.
Researchers at the University of North Carolina have written algorithms designed to keep users' lines of sight clear as they move through and manipulate these virtual worlds using touch-sensitive, or haptic controls.
These haptic controls translate the force and torque of a person's movements -- usually hand movements -- into changes within the virtual environment. These changes can be anything from the movement of a cursor-like probe to the shaping of a virtual object, assembly of virtual parts, or operations on symbolically represented data sets.
In order for a person to follow the virtual action visually, his field of vision must be tracked and oriented by virtual cameras. This tracking and orientation, however, can easily fall out of synch with the movements of the haptic controls.
Some existing tracking schemes use head and eye movements to gauge what area of a virtual world a person is interested in at any given moment. The UNC researchers sought to gauge interest by instead tracking the way a person uses a three-dimensional haptic control arm, which is an elaborate type of joystick. They also experimented with virtual painting, which translates users' strokes with a haptic brush device into marks in a computer-generated picture.
The researchers wrote software algorithms that repositioned the virtual camera based on the previous few movements of the haptic control device and the way objects are situated in the virtual space. The software also accounts for the virtual camera's field of view and focus distance.
The algorithms react directly to the control device's movements, said Ming Lin, associate professor of computer science at the University of North Carolina. "We... infer users' intentions from the motion of the haptic probe, [and] contact regions are used to determine regions of interest during the manipulation."
At this point in their research the scientists have not tapped into the force of the user's movements, she added. "We haven't explored the concept of using the amount of force to deduce users' intentions."
Using the motion information, however, "we can implicitly define the objects of interest as [those] that the user is grabbing [or] interacting with," said Lin.
The algorithms can be pre-set to address specific virtual worlds and data, or can adapt as a virtual space is navigated and manipulated, said Lin. "One of the key issues is the relative size of objects and the virtual probe. [The algorithms] could potentially be self-adaptable," she said.
The researchers' algorithms include a method for adding a second camera for a second view into the virtual space to handle crowded areas where the main camera's view could become obstructed.
They tested the camera adjustment techniques with 10 computer science students in virtual environments for rendering polygonal models, for painting with a haptic brush interface, and navigating three-dimensional models.
The algorithms developed for the tests would require only minor adjustments to work with current virtual reality and three-dimensional display systems, Lin said. "The techniques... are quite easy to implement and are also independent from each other. We do not anticipate too much effort to be required to port these techniques to commercial systems," she said.
Further research would put the algorithms through their paces with more complex models, said Lin. "Our current system is interactive with medium size models [of] several thousand polygons, but it might not be fast enough for haptic rendering of massive models consisting of several millions of polygons [like] datasets from scientific computation and medical visualization of human organs," she said.
The methods could improve virtual navigation with haptic input devices like joysticks that have a single point of contact, but would not do much for more complex input devices like gloves, said Grigore Burdea, associate professor of electrical and computer engineering at Rutgers University. The algorithms "will not... work, in my view, for dexterous haptic interactions using gloves, where multiple contact points exist simultaneously," he said.
More extensive user testing would be needed to ultimately gauge the techniques' usefulness and ergonomic impact, Burdea said. One crucial aspect of the work that looks good on the researchers' video is smooth viewpoint transition, he added. One problem with virtual environments is they can potentially lead to "simulation sickness if there's no filtering -- for example if the probe is in contact with a bumpy surface, leading to rapid changes in the detailed camera view," said Burdea.
Lin's research colleague was Miguel A. Otaduy. They presented the research at the October IEEE Visualization 2001 conference in San Diego. The work was funded by the National Science Foundation (NSF), the Office of Naval Research (ARO), the Department of Energy (DOE), the Army Research Office (ONR), the Government of the Basque Country in Spain, and Intel Corp.
Timeline: Now
Funding: Government, Corporate
TRN Categories: Applied Computing, Software Design and Engineering
Story Type: News
Related Elements: Technical paper, "User-Centric Viewpoint Computation for Haptic Exploration and Manipulation", IEEE Visualization 2001 conference, San Diego, October 21 to 26, 2001. >
Advertisements:
January 9, 2002
Page One
Sunlight turns water to fuel
Search method melds results
VR tool keeps line of sight in hand
Laser pulse penetrates glass
Model tracks desert spread
News:
Research News Roundup
Research Watch blog
Features:
View from the High Ground Q&A
How It Works
RSS Feeds:
News
| Blog
| Books 
Ad links:
Buy an ad link
| Advertisements:
|
 |
Ad links: Clear History
Buy an ad link
|
TRN
Newswire and Headline Feeds for Web sites
|
© Copyright Technology Research News, LLC 2000-2006. All rights reserved.
