Virtual reality (VR) is a new human-machine interface technology based on the full sensory immersion of participants in the virtual world, generated in real time by the high-performance computer. Virtual worlds can range from physical spaces such as those modelled by dynamic terrain viewers or architectural walk-through tools, through a variety of ``fantasy lands'' to entirely abstract cognitive spaces, generated by dynamic visualization of low-dimensional parametric subspaces, extracted from complex nontopographic databases.
The very concept of the immersive interface and the first prototypes were already known in 1960s [Sutherland:68a] and 1970s [Kilpatrick:76a]. In the 1990s, VR technology is becoming affordable. Most current popular hardware implementations of the interface are based on a set exotic peripherals such as goggles for the wide solid-angle three-dimensional video output, head-position trackers, and gloves for sensory input and tactile feedback. Another immersion strategy is based on ``non-encumbered'' interfaces [Krueger:91a], implemented in terms of the real-time machine vision front-end which analyzes participants' gestures and responds with the synchronized sensory feedback from the virtual world.
VR projects cover the wide range of technologies and goals, including high-end scientific visualization (UNC), high-end space applications (NASA Ames), base research and technology transfer (HIT Lab), and low-end consumer market products (AGE).
The VR domain is growing vigorously and already has reached the mass media, generating the current ``VR hype.'' According to VR enthusiasts, this technology marks the new generation of computing and will start a revolution comparable in scope to personal computing in the early 1980s. In our opinion, this might be the correct assessment since VR seems to be the most natural logical next step in the evolution of human-machine interfaces and it might indeed become the ``ultimate solution'' for using computers because of its potential for maximal sensory integration with humans. However, the explicit implementations of VR will most probably vary very rapidly in the coming years, in parallel with the progress of technology, and most of the current solutions, systems, and concepts will become obsolete very soon.
Nevertheless, one is tempted to immediately start exploring this exciting field, additionally encouraged by the rapidly increasing affordability of VR peripherals. The typical cost of a peripheral unit for a VR environment has gradually decreased from $1M (Super Cockpit) in the 1970s through $100K in the 1980s (NASA) down to $10K (VPL DataGlove) in the early 1990s. The new generation of low-cost ``consumer VR'' systems which will reach the broad market in the mid-1990s comes with a price tag of about $100. This clearly indicates that the time to get involved in VR is-now!
VR opponents predict that VR will have its major impact in entertainment rather than R&D or education. However, there is already a new buzzword in VR newspeak, suggesting a compromise solution: edutainment! From the software engineering perspective, the edutainment argument can be formulated as follows: the software models and standards generated today will mature perhaps five to ten years from now and hence they will be used by the present ``Nintendo generation.'' There is no reason to expect that these kids will accept anything less intuitive and natural for user interfaces than the current Nintendo standards, which will evolve rapidly during the coming years towards the full-blown VR interfaces.
Leaving aside longer term prognoses, we would expect that a few years from now, VR will be available on all systems in the form of an add-on option, more or less as the mouse was for personal computers a few years ago. We will be witnessing soon the new generation of consumer VR products for the broad entertainment market and, in the next stage, the transfer of this technology to the computer interface domain. These low-cost gloves and headsets will probably appear more and more frequently attached to conventional monitors and easy to use. VR applications will coexist with standard applications within the existing windowing systems. We will still be using conventional text editors and other window tools, whereas the add-on VR peripherals and software layers will allow us to enter virtual worlds (i.e., dynamic three-dimensional-intensive applications) through conventional two-dimensional windows.