Pages: pp. 12-15
Ten years ago, experts predicted great success for virtual-reality technology. The idea of users immersing themselves in virtual worlds for games, simulations, aerospace engineering, scientific research, and other activities seemed to have many commercial possibilities, said Jackie Fenn, a vice president and Fellow with Gartner Inc., a market research firm.
The future seemed so bright that VR became the subject of high-profile newspaper articles and movies such as Tron and Lawnmower Man.
Along the way, though, the original VR concept—complete with head-mounted displays (HMDs) and data gloves that track hand movements—remained an expensive niche application used only in high-end fields such as military and medical simulations, automotive and other complex product design work, and major university research projects.
Until recently, the computing power and storage capacity necessary to run VR systems were expensive. But even as those costs dropped, goggles, gloves, and some other types of equipment remained expensive, unwieldy, and not useful for many applications, noted Ben Delaney, president of Cyberedge Information Services, a market research firm.
And far fewer users than expected were interested in escaping into immersive VR worlds.
Now, though, VR systems are becoming less expensive and easier to use. This has occurred in part because computing and storage costs have fallen considerably. And instead of using supercomputers, some VR systems are working with clusters of commodity PCs or even single PCs, which are much less expensive.
Moreover, companies and researchers are developing scaled-down VR approaches that aren't as complex or costly as the original systems and are thus suitable for less demanding applications such as smaller research and art projects.
Also, some games and other graphics-related applications are incorporating features that have turned them into low-end VR systems. These include World of Warcraft and other popular multiplayer 3D games, which show the image of a 3D scene from each player's perspective, and interactive 3D worlds such as Linden Lab's Second Life.
Researchers at several universities are developing open source, low-cost VR platforms for scientific visualization, art, and training, noted Norman Badler, director of the University of Pennsylvania's Center for Human Modeling and Simulation.
However, users won't widely adopt VR until the cost of some system elements fall further, development tools become less expensive and easier to use, and proponents address several other problems.
VR systems use software and hardware to create and manage a virtual, interactive 3D environment that includes visual and sometimes audio and tactile elements. They generally include various types of display, sensor, and user-tracking and -navigation technologies. The systems can either simulate a real environment, such as a building, or create an imaginary one.
In 1965, Ivan Sutherland—then a researcher at the US Defense Advanced Research Projects Agency (DARPA) and now a Sun Microsystems Fellow and vice president—described some early concepts for a virtual environment interface.
Four years later, then-University of Wisconsin researchers Dan Sandin, who subsequently moved to the University of Illinois at Chicago, and Myron Krueger collaborated on an early prototype VR environment called Glow Flow. The environment used black-and-white graphics and responded to touch sensors in the floor.
Sandin—a professor emeritus who is still codirector of the University of Illinois at Chicago's Electronic Visualization Lab—went on to create the first data-glove interface in 1976. He also helped develop one of the first major VR research projects: the Cave Automatic Virtual Environment (CAVE), demonstrated at the 1992 ACM Siggraph conference.
In the early 1990s, according to Cyberedge's Delaney, VPL Research built the first commercial VR system, which NASA used.
A typical VR system includes computers for processing data; graphics processors for rendering the data into views for each of a user's eyes; multiple projectors that use polarizing filters to project separate views for each eye; a viewing system such as an HMD; and tracking systems that follow users' motions so that the views they see can change as they move.
In addition to being expensive, some of the high-end systems require large amounts of equipment and thus take up more space than many potential users have available.
Also, the systems have been so complex that they have been tricky to set up and program, and, in some cases, their interfaces have been too difficult for anyone who isn't technically knowledgeable or highly trained to use.
Compared to some of today's scaled-down VR applications, Delaney said, the high-end systems have more complicated software, more processing power, bigger image generators, bigger displays, and more storage. They are better suited for complex, data-intensive applications such as weather forecasting, fluid-flow dynamics, simulations of multiple orbiting planetary bodies, or training large numbers of people to perform complex activities.
Companies such as Evans & Sutherland Computer are building commercial traditional VR systems for industrial, military, medical, academic, and other applications. For example, visualization and VR vendor Fakespace Systems recently sold a $3.5 million VR center to the Los Alamos National Laboratory for nuclear weapons simulation. The center can display 43 million pixels of data (compared to 768,000 for a standard computer monitor) at one time on three walls, the floor, and the ceiling.
Newer VR systems are less expensive and complex than their predecessors and thus are accessible by more people for simpler projects.
Advances in and increased production efficiencies for commodity components have reduced their costs and have contributed to lower prices for VR systems. For example, processors, projectors, and HMDs are less costly than earlier versions with the same performance.
In addition, researchers have developed simpler, less-expensive VR applications that project images onto a single wall rather than multiple walls. Multiwall systems create a more immersive experience but are more costly and complex to run.
The smaller systems can work with fewer cameras, which also saves money and simplifies operations.
Moreover, developers can build VR applications with recently developed open source software tools, which are less expensive than commercial or custom-built products.
Thus, Cyberedge's Delaney said, building a CAVE-like system that would have cost $500,000 about 15 years ago would cost only about $20,000 today.
Lower prices and ease of implementation have made the applications affordable and accessible to users who couldn't have worked with VR in the past, such as artists, students, scientists and engineers working on smaller projects, architects, and museum technical staff.
Some users, such as geologists, are also working with scaled-down systems for data visualization, according to professor emeritus Sandin. Others are using VR to generate 3D displays of buildings or rooms, such as those used in museum or art exhibitions.
There are several academically developed less-complex VR systems.
The University of Illinois has developed a scaled-down, less ex- pensive version of CAVE that is now used on about 175 projects world- wide, noted Henry Kaczmarski, director of the school's Integrated Systems Laboratory.
Using PC clusters rather than a single high-performance machine dropped CAVE's computing costs for a typical system to about $45,000, he explained. In addition, projector costs have fallen to about $2,000 each. The polarizing screens cost up to $6,000, and tracking systems also run about $6,000, said Sandin. Thus, an entire system is affordable for many smaller projects.
Researchers at the University of Illinois at Urbana-Champaign developed the Collaborative Advanced Navigation Virtual Art Studio, a three-screen, rear-projection, portable, CAVE-based VR laboratory for fine- and applied-arts research and teaching projects. CANVAS supports collaboration, advanced navigation, and virtual art.
Sandin developed the C-Wall (Configurable Wall), a high-quality stereoscopic VR application that is configurable for front or rear projection and compatible with CAVE software. In essence, the application functions as a one-wall CAVE. Users have worked with the C-Wall to build, for example, the Looking for Water application, which displays animations of several months of weather conditions.
CANVAS and the C-Wall are less expensive than full CAVE implementations in part because they don't have as much computing power and they use less-costly projectors.
High-end science and industrial customers currently spend from $250,000 to $1.5 million for advanced CAVE versions, used for industrial and scientific work, detailed simulations, and other complex projects.
Indiana University's Advanced Visualization Lab has developed the John-e-Box, a portable, rear-projection system with a 4 × 3 foot screen.
A preintegrated system can cost as little as $20,000. John-e-Box uses commodity components, including small digital projectors, standard Intel PC processors, Nvidia graphics cards, and open source software tools.
The system renders a 3D view of a scene and lets users navigate via a SpaceBall motion controller from 3Dconnexion.
CAE-net Inc.—which sells modeling, simulation, and 3D stereo-visualization hardware, software, and services—is commercializing the John-e-Box.
Several Indiana University campuses have deployed the application for data visualization projects.
Researchers at Indiana University and the University of Illinois' Urbana-Champaign and Chicago campuses are jointly and, in some cases, separately developing open source VR tools.
Open source platforms are less ex- pensive than proprietary ones and thus are reducing VR environments' costs and making them more accessible.
The University of Illinois at Urbana-Champaign's Integrated Systems Laboratory has created the Syzygy (pronounced si-zid-ji) open source toolkit ( www.isl.uiuc.edu/syzygy.htm) for cluster-based VR.
"Syzygy is about taking clusters of PCs and turning them into a high-performance, high-quality, low-cost VR platform. It is the software glue that binds the PCs together. This helps make VR cheaper," said Syzygy inventor Benjamin Schaeffer, formerly with the University of Illinois and now a Wachovia Corp. trading analyst.
Ygdrasil and Electro, which the University of Illinois at Chicago's Electronic Visualization Laboratory developed, are two important scripting systems that enable artists and scientists to create VR worlds without having to be hardcore programmers, said professor emeritus Sandin.
The Ygdrasil framework is a tool for creating networked virtual environments. It focuses on building the behaviors of virtual objects from reusable components and sharing an environment's state via distributed technologies. It is presently being used to build several artistic and educational applications.
Electro is an environment for simplifying the development of applications that span multiple processors and displays. It works with cluster-based and desktop systems and supports Linux, Macintosh, and Windows environments.
Multiplayer 3D games and inter- active worlds are among the less- complex commercial VR systems on the market.
There are no off-the-shelf commercial VR systems, but several companies build and integrate components for specialized applications, including Advanced Visual Systems, Fakespace, and L-3 Communications.
For example, Fakespace offers components for a basic VR system called the Workz Bundle, which includes a $45,000 stereoscopic projector and a single 6 × 8 foot screen with a resolution of 1,400 × 1,050 pixels, noted Jeff Brum, the company's vice president of business development and marketing.
This type of system, consisting of the elements shown in Figure 1, might be good for prototyping small parts, showing a proposed museum exhibit, or research in which resolution is not as important as the basic representation of content or interaction capabilities, he explained.
Figure 1 Fakespace Systems'Workz Bundle is a set of components—including a high-performance workstation, a stereoscopic projector, a 6 × 8 foot screen, and stereoscopic glasses—that can be integrated into a low-cost virtual-reality system.
Customers use the less complex commercial VR applications for various purposes, including military and surgical training, art exhibits, data visualization, architectural walk-throughs, and design collaboration.
One new type of VR application combines information from security-system sensors into a navigable, 3D virtual world that enhances surveillance capabilities.
For example, Vistascape has developed an application that fuses data from cameras and alarm sensors into a single virtual view of a facility. Guards can thus view the facility as a whole and focus on anomalous events, such as someone being in area that is supposed to be unoccupied, noted Don Campbell, the company's vice president of software and services.
Currently, guards typically try to detect problems by watching dozens of video screens.
The less expensive, easier-to-use VR systems have performance limitations and can't be used in as many ways as the high-end systems. For example, the new systems aren't as fast, provide lower graphics quality, and don't offer an immersive experience.
There are several other barriers to wider VR deployment.
The skills required to develop virtual worlds have been beyond the reach of many programmers, said professor emeritus Sandin. New tools are only beginning to address this issue.
Also, commercial tools from companies such as CEI and Virtools are beginning to make creating and implementing virtual-reality systems easier by, for example, providing GUIs and more efficiently distributing VR applications across computer clusters, added Matt Szymanski, vice president of engineering at VRCO Inc.
Roy Latham, president of Computer Graphics Systems Development Corp., a vendor of VR and simulation products, said another big challenge is the continued lack of low-priced tracking devices and HMDs. The market for this equipment isn't big enough yet to create manufacturing economies of scale and otherwise drive down prices.
Another factor limiting VR adoption is that many systems still require users to wear goggles, which are uncomfortable when worn for long periods of time, or glasses, which limit the field of vision, explained Dan Mapes, director of business development at 3D-equipment vendor Deep Light.
In response, Deep Light and several other organizations are developing machine-vision approaches that track users' movements without their having to wear special equipment, as well as stereoscopic displays that let users view 3D images without eyewear.
Fakespace's Brum predicted that companies will begin adopting VR systems more widely for team-based environments to help workers collaborate on design and other projects.
Delaney predicted VR won't become a mass consumer technology in the near future because not enough applications require it. However, he added, because VR has become accessible to more users, some companies, such as architectural and engineering firms, will begin adopting the technology for product and architectural design projects.
According to professor emeritus Sandin, if the technology's prices fall to near those of standard visualization, users will more widely adopt VR for the enhanced visualization it creates.