VIRTUAL REALITY INTERFACES

                           Christopher M. Smith
                           Lynellen D.S.P. Smith
                                Steven Yang

                             A Final Project
                        Submitted to the Faculty of
                       Mississippi State University
                in Partial Fulfillment of the Requirements
                              for CS4663/6663
                   in the Department of Computer Science

                      Mississippi State, Mississippi
                              April 25, 1995

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1

Visual Aspects of Virtual Reality. . . . . . . . . . . . . . . . . . . .  1
     Types of Visual Display . . . . . . . . . . . . . . . . . . . . . .  2
          LCD Flicker Lens . . . . . . . . . . . . . . . . . . . . . . .  2
          Head Mounted Displays. . . . . . . . . . . . . . . . . . . . .  3
          LCD display HMD. . . . . . . . . . . . . . . . . . . . . . . .  3
          Projected HMD. . . . . . . . . . . . . . . . . . . . . . . . .  3
          Small CRT HMD. . . . . . . . . . . . . . . . . . . . . . . . .  4
          Single Column LED HMD. . . . . . . . . . . . . . . . . . . . .  4
          Binocular Omni-Orientation Monitor (BOOM). . . . . . . . . . .  4
     Advantages and Disadvantages. . . . . . . . . . . . . . . . . . . .  5

Graphic Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . .  7
     Depth Cueing. . . . . . . . . . . . . . . . . . . . . . . . . . . .  7
     Lighting Models . . . . . . . . . . . . . . . . . . . . . . . . . .  8
     Shading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  8
     Radiosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  9
     Ray Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . .  9

3-D Audio. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
     Evolution of 3D Sound . . . . . . . . . . . . . . . . . . . . . . . 11
     Realistic Sound . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Tactile and Force Feedback . . . . . . . . . . . . . . . . . . . . . . . 15
     Force Feedback. . . . . . . . . . . . . . . . . . . . . . . . . . . 16
          Motion Platforms . . . . . . . . . . . . . . . . . . . . . . . 16
Gloves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
          Exoskeletons . . . . . . . . . . . . . . . . . . . . . . . . . 17
Butlers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
     Tactile Feedback. . . . . . . . . . . . . . . . . . . . . . . . . . 18
          Texture. . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
     Tracking Devices. . . . . . . . . . . . . . . . . . . . . . . . . . 19
          Mechanical Trackers. . . . . . . . . . . . . . . . . . . . . . 21
          Electromagnetic Trackers . . . . . . . . . . . . . . . . . . . 21
          Ultrasonic Trackers. . . . . . . . . . . . . . . . . . . . . . 22
          Infrared Trackers. . . . . . . . . . . . . . . . . . . . . . . 23
          Inertial Trackers. . . . . . . . . . . . . . . . . . . . . . . 24
     Interaction Devices . . . . . . . . . . . . . . . . . . . . . . . . 25
          Gloves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
          3D Mice. . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
          Joysticks. . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Human Factors in Virtual Environments. . . . . . . . . . . . . . . . . . 28

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
          Virtual Reality hype is becoming a large part of
     everyday life.  This paper explores the components of actual
     virtual reality systems, critiqueing each in terms of human
     factors.  The hardware and software of visual, aural, and
     haptic input and feedback are considered.  Technical and
     human factor difficulties are discussed and some potential
     solutions are offered.
          Virtual reality is a new technology for simulation,
     design, entertainment, and many other pursuits.  Simulation
     applications range from testing a non-existent object before
     it is created, to military training, to practicing a
     maneuver for the next shuttle mission.  The purpose of our
     paper is to identify weaknesses in virtual reality
     interfaces.  To accomplish this task, we have divided the
     typical virtual reality interface into four specific areas:
     audio, visual, tactile, and navigation.  We will point out
     the limitations of current solutions to problems in these
     areas, possible areas of improvement, and those problems
     that remain completely unsolved at this point.
                  Visual Aspects of Virtual Reality
          The main tradeoffs in this area are image detail versus
     rendering speed, and monoscopic versus stereoscopic vision. 
     In most applications of virtual reality, visual feedback is
     required.  In fact, visual cues are perhaps the most
     important feedback required in the virtual reality system. 
     To achieve reality, the pictures sent to the display have to
     be real-time to avoid discontinuity.  Therefore, the
     trade-off between the rendering time and the graphic
     resolution for the 3-dimensional graphic scene and the
     2-dimensional graphic scene are investigated from both the
     software and hardware perspective.
                         Types of Visual Display
                                     LCD Flicker Lens
          The LCD (liquid crystal display) flicker lens has the
     appearance of a pair of glasses.  A photosensor is mounted
     on these LCD glasses with the sole purpose of reading a
     signal from the computer.  This signal tells the LCD glasses
     whether to allow light to pass through the left lens or the
     right lens.  When light is allowed to pass through the left
     lens, the computer screen will be showing a left eye scene,
     which corresponds to what the user will see through his left
     eye.  When light passes through the right lens, the scene on
     the computer screen is a slightly offset version of the view
     from the left eye.  The glasses switch between the two
     lenses at 60 Hertz, which causes the user to perceive a
     continual 3D view via the mechanics of parallax (Blanchard
     and Tsuneto).
     Head Mounted Displays
          Head mounted displays place a display screen in front
     of each of the viewer's eyes at all times.  The view, the
     segment of the virtual environment generated and displayed,
     is controlled by orientation sensors mounted on the
     "helmet".  Head movement is recognized by the computer, and
     a new perspective of the scene is generated.  In most cases,
     a set of optical lens and mirrors are used to enlarge the
     view to fill the field of view and to direct the scene to
     the eyes (Lane).  Four types of Head Mounted Displays (HMDs)
     will be discussed below.
     LCD display HMD
          This type of HMD uses LCD technology to display the
     scene.  When a liquid crystal pixel is activated, it blocks
     the passage of light through it.  Thousands of these pixels
     are arranged in a two dimensional matrix for each display. 
     Since liquid crystals block the passage of light, to display
     the scene a light must be shone from behind the LCD matrix
     toward the eye to provide brightness for the scene
     (Aukstakalnis and Blatner 1992).
     Projected HMD
          This type of HMD uses fiber optic cables to transmit
     the scene to the screen.  The screen is similar to a cathode
     ray tube (CRT) except the phosphor is illuminated by the
     light transmitted through fiber optic cables.  Ideally, each
     fiber would control one pixel.  But due to the limitation in
     cost and manufacturing, each fiber controls a honeycomb
     section of pixels (Lane).
     Small CRT HMD
          This type of HMD uses two CRTs that are positioned on
     the side of the HMD.  Mirrors are used to direct the scene
     to the viewers eye.  Unlike the projected HMD where the
     phosphor is illuminated by fiber optic cables, here the
     phosphor is illuminated by an electron gun as usual (Lane).
     Single Column LED HMD
          This type of HMD uses one column of 280 LEDs.  A mirror
     rapidly oscillates opposite from the LEDs, reflecting the
     image to the user's eye.  The LEDs are updated 720 times per
     oscillation of the mirror.  As the LED column updates for
     each column of the virtual screen, the mirror redirects the
     light to the viewers eye, one column at a time, to form the
     image of the entire virtual screen (Aukstakalnis and Blatner
     Binocular Omni-Orientation Monitor (BOOM)
          The binocular omni-orientation monitor is mounted on a
     jointed mechanical arm with tracking sensors located at the
     joints.  A counterbalance is used to stabilize the monitor,
     so that when the user releases the monitor, it remains in
     place.  To view the virtual environment, the user must take
     hold of the monitor and put her face up to it.  The computer
     will generate an appropriate scene based on the position and
     orientation of the joints on the mechanical arm
     (Aukstakalnis and Blatner 1992).
     Advantages and Disadvantages
     LCD flicker lens' are light weight and cordless.  These
     two features makes them easy to wear and to remove. 
     Unfortunately, the user would have to stare only at the
     computer screen in order to see the 3-D scene.  Since the
     field of view is limited to the size of the computer screen,
     the surrounding real world environment can also be noticed. 
     This does not provide an immersive effect.
          The LCD display HMD is lighter than most HMDs.  As with
     most HMDs, it does provide an immersive effect, but the
     resolution and the contrast is low.  The problem associated
     with low resolution is inability to identify objects and
     inability to locate the exact position of objects.  Since
     the crystals are polarized to control the color of a pixel,
     the actual polarizing of the crystal creates a small delay
     while forming the image on the screen.  Such a delay may
     cause the viewer to misjudge the position of objects (Bolas
          The projected HMD provides better resolution and
     contrast than LCD displays.  This HMD is also light weight. 
     Higher resolution and contrast means that the viewer is able
     to see an image with greater detail.  The downside of this
     type of HMD is that it is expensive and difficult to
     manufacture (Bolas 1994).
          The CRT HMD is in many ways similar to the projected
     HMD.  This type of HMD is heavier than most other types
     because of added electronic components.  These electronic
     parts also generate large amounts of heat.  The user wearing
     this type of HMD may feel discomfort due to the heat and the
     weight of the HMD (Bolas 1994).
          Single Column LED HMDs allow the user to interact with
     a virtual world and the real world simultaneously.  This
     type of display can be used to create a virtual screen that
     seems to float in the real world.  
          One of the common problems of HMDs is that the cable
     connecting the HMD and a computer restricts the mobility of
     the user.  The user can only move as far as the cable
     allows.  If the cable is not properly managed, one could
     trip over it or become entangled in it.  Another problem
     occurs when switching frequently between a virtual world and
     the real world. Continually wearing and removing the HMD is
     tedious and tiresome.
          Some of the problems associated with HMDs can be solved
     by using a BOOM display.  The user does not have to wear a
     BOOM display as in the case of an HMD.   This means that
     crossing the boundary between a virtual world and the real
     world is simply a matter of moving the eyes away from the
     Graphic Techniques
                                   Depth Cueing
          Depth Cueing is a technique to provide a 3-D
     perspective of a scene.  Basically, it adds depth to a 2-D
     picture.  There are two ways to add such perspective.  The
     first method is to change the color of pixels.  For example,
     a light source located farther away looks dimmer than one
     located nearer the viewer.  When the scene is generated, the
     red, green, and blue color values of the light source
     located farther away can be reduced by the same amount to
     represent its distance.  The second method is to add color
     to the scene.  For example, fog may exist between a light
     source and the viewpoint.  The color value corresponding to
     that light source in the fog is accumulative from the
     viewpoint.  The deeper the light is in the fog, the whiter
     its light.  In this manner the user can judge the depth of
     the scene and navigate through the virtual world with less
     effort (Foley et al. 1992).
                           Lighting Models
          Lighting is essential in creating a realistic 3-D
     image.  In an unlit atmosphere, a sphere looks the same as a
     disk.  In other words, a 3-D scene becomes a 2-D scene.  As
     a light shines on a sphere, some of the light is reflected
     diffusely and thereby gives an indication of three-
     dimensionality.  The other reflected light allows the viewer
     to see the sphere's specular reflection and shadow, which
     provide further spatial information.  Due to the complexity
     of the lighting formula, computing the intensity of light at
     each pixel reduces the screen update rate (Foley et al.
     1992).  When the update rate falls too low, flickers are
     noticeable and can cause visually induced motion sickness. 
     Motion sickness occurs whenever the body receives
     conflicting information from the eyes and the brain (Bolas
          Shading is used in conjunction with lighting.  As the
     lighting model is computed, the intensity of the light at
     each vertex of a polygon is calculated.  Two types of
     shading can be used to render a scene.  The first is flat
     shading, where the entire polygon takes on the value of the
     color of one of the vertices.  The second is Gouraud
     shading, where pixel color within the polygon is linearly
     interpolated from the color at each vertex to create a
     smoothly varying shade.  Since linear interpolation requires
     little computational power, it is often used to render
     realistic scenes quickly (Foley et al. 1992).
          Radiosity is most often used for rendering a static
     scene.  This technique pre-calculates the light interaction
     among the objects in the 3-D environment.  The calculation
     of light interaction for a complex environment usually takes
     a long time.  However, upon completion of the calculation,
     the user can view this environment from any angle, on the
     fly, with no further exhaustive computation (Foley et al.
                             Ray Casting
          Ray casting is a real-time technique for rendering a
     scene.  A ray is traced from the viewpoint through a pixel
     of the screen to intersect an object defined in the virtual
     space.  The color of the closest object intersected is
     recorded for that pixel of the screen.  If an object is
     moved, ray casting can trace the new position of the object,
     thus creating a dynamic scene (Foley et al. 1992).
          Distortion correction is needed in most of HMDs.  The
     optic lenses used in the HMD magnifies the image displayed
     on the screen.  This magnification causes the image to be
     distorted radially.  Since this distorted image is not what
     one would see in the real world, the viewer may feel nausea
     in the virtual world (Watson and Hodges).  
     3-D Audio
           The main research area in audio is the simulation of
     sound origin.  "It has been demonstrated that using sound to
     supply alternative or supplementary information to a
     computer user can greatly increase the amount of information
     they can ingest" (Aukstakalnis and Blatner 1992).  This is
     no less true in the virtual world.  Although a virtual
     environment provides a multitude of visual cues, a person
     can't see what is behind him/her.  However, in reality
     someone can hear what is behind them and it should be so in
     a virtual world. 
          The main problem with producing sound is that it is
     impossible to replay previously recorded sound in a manner
     that moves a sound from behind you to in front of you as you
     turn your head.  Crystal River Engineering has developed a
     process for producing a sound such that it seems like it is
     coming from a particular direction.  Since these sounds are
     computed and produced in real time, there is not a problem
     with playback. 
          In addition to visual output, a complete virtual world
     must incorporate a three dimensional sound field that
     reflects the conditions modeled in the virtual environment. 
     This sound field has to react to walls, multiple sound
     sources, and background noise, as well as the absence of
     them.  This requires massive computational power and speed
     because hearing is a complex system which uses the shape of
     the outer ear and microsecond delays in the arrival of sound
     to the two ears to determine position and location of the
     source of the sound.
                        Evolution of 3D Sound
          The evolution of 3D sound can be traced through time
     and technology starting with monophonic sound.  "Mono", the
     latin word for "one", sends one signal to every speaker.  It
     appears that all of the sounds of the environment are coming
     from each individual speaker.  If there is only one speaker,
     then all sounds seem to be coming from that point.  
          Stereophonic sound is the next step in the development
     of realistic sound production.  It allows for the sounds to
     seem as if they are coming from anywhere between two
     speakers.  This is accomplished by delaying the signals
     between the two speakers by a few microseconds.  The smaller
     the delay, the closer to the center the source appears to be
          Surround sound, used in many theaters, uses the idea of
     stereo but with more speakers.  They delays can be set so
     that a sound can seem to move from behind the listener to in
     front of the listener.  An example problem with this system
     is that a plane taking off behind the listener will appear
     to go by the listener's elbow instead of overhead (Crystal
     River Engineering 1994).
          A solution to the problem of creating a three
     dimensional sound field comes from production of sound which
     is tuned to an individual's head.  When sound reaches the
     outer ear, the outer ear bends the sound wavefront and
     channels it down the ear canal.  The sound that actually
     reaches the eardrum is different for each person
     (Aukstakalnis and Blatner 1992).  To resolve this problem,
     the computer must create a sound that is custom designed for
     a particular user.  This is done by placing small
     microphones inside the ear canal, then creating reference
     sounds from various locations around the listener.  Then the
     computer solves a set of mathematical relationships that
     describe how the sound is changed from being produced to
     being received inside the ear canal.  These mathematical
     relationships are called Head Related Transfer Functions
     (HRTFs) (Crystal River Engineering 1994).
          To simulate a virtual sound environment, a computer
     must first determine the position of the source relative to
     the listener.  It also must calculate the effects of the
     environment.  For example, to simulate an echo due to a
     wall, the computer must first determine the source's
     location relative to the subject and the wall, then place
     another source the at the appropriate distance and location
     on the opposite side of the wall (Tonnesen and Steinmetz).
                           Realistic Sound
          An additional computational burden is the production of
     background noise.  This is very important if the person in
     the virtual environment wishes to be immersed in a
     "believable" world.  However, since the noise is background,
     it does not need to take advantage of 3D sound technology. 
     This limits the interactivity of the user with the virtual
     environment.  In the real world, a person can pick out
     sounds from the background.  This ability is commonly called
     the "cocktail party affect", because of the ability of a
     person to focus on different conversations from the
     background noise.  This can only be done in a 3D acoustic
     field (Steinmentz and Lee) and background noise in the
     virtual world does not use a 3D field.  
          Some researchers suggest the use of prerecorded sounds
     so that all computational power is devoted to determining
     the location and direction of the source.  This, however,
     can not work in a 3D sound field.  Although the sounds can
     be accurately placed in a 3-D sound field, the listener can
     not interact with the environment--he/she can only observe
     it.  In a 3D acoustic field played through headphones, when
     the listener turns around, the sounds that were behind the
     listener should now be in front.  However, with
     prerecorded/playback methods, the sounds that were behind
     the listener are still behind the listener (Aukstakalnis and
     Blatner 1992).
          HRTF measurements can not accurately simulate the
     acoustical environment when used alone.  The problem lies in
     trying to nonintrusively make measurements.  When the
     microphone is placed in the ear canal, it changes the
     acoustical track, thereby changing the HRTF.  Also, this
     method does not attempt to take into consideration the
     middle or inner ear (Steinmetz and Lee).
          A realistic sound environment has great potential to be
     an interface for visually impaired or blind people.  For
     example, a virtual environment can be created where the
     objects in it are application software.  Then the users can
     learn their way around the environment, much like they can
     learn their way from home to the store without ever having
     to see.
                      Tactile and Force Feedback
          One of the biggest complaints about top of the line
     virtual environment packages is the "lack of tangibility." 
     Although the area of tactile feedback is only a few years
     old, it has produced some impressive results.  These
     solutions are critiqued below.  There is no interface
     currently built that will simulate the interactions of
     shape, texture, temperature, firmness, and force.  
          Being able to produce a realistic interface means
     having to produce tactile and force feedback to correspond
     to the objects in the virtual world.  Dr. Fred Brooks of the
     University of North Carolina at Chapel Hill is noted for
     introducing the problem of "shin-knockers" (Brooks 1995). 
     This was originally in reference to modeling of a submarine. 
     "How are you going to let the person know when he knocks his
     shin on a pipe that is sticking out in his way?"  
          The area of touch has been broken down into two
     different areas.  Force feedback deals with how the virtual
     environment affects a user.  For example, walls should stop
     someone instead of letting him/her pass through, and pipes
     should knock a user in the shin to let him/her know that
     they are there.  Tactile feedback deals with how a virtual
     object feels.  Temperature, size, shape, firmness, and
     texture are some of the bits of information gained through
     the sense of touch. 
                            Force Feedback
          There are several different types of devices that allow
     a user to "feel" certain aspects of the virtual environment. 
     Motion platforms for simulators and simulated rides, force
     feedback gloves, exoskeletons, and butlers are all forms of
     force feedback.  
     Motion Platforms
          The motion platform was originally designed for use in
     flight simulators which train pilots.  A platform is bolted
     to a set of hydraulic lift arms.  As the motion from a
     visual display changes, the platform tilts and moves in a
     synchronous path to give the user a "feeling" that he/she is
     actually flying.  This has a serious limitation in that it
     can only go so far.  If the user's visual cues are that the
     plane is upside down, the hydraulics can not simulate this. 
     However, it does give the middle ear sensations that
     correspond to the visual scene, making the simulation more
          For interaction with small objects in a virtual world,
     the user can use one of several gloves designed to give
     feedback on the characteristics of the object.  This can be
     done with pneumatic pistons which are mounted on the palm of
     the glove, as in the Rutgers Master II (Gomez, Burden and
     Langrana 1995).  When a virtual object is placed in the
     virtual hand, the users hand can close around it.  When the
     fingers would meet resistance from the object in reality,
     the pressure in the pistons is increased, giving the
     sensation of resistance from the virtual object.  
          Exoskeletons are also employed to simulate the
     resistance of objects in a virtual world.  An exoskeleton is
     basically a robotic arm strapped to a person.  At the
     University of Utah, researchers have developed a robotic arm
     which has 10 degrees of freedom.  The robot continuously
     updates the force to each of its ten joints, and can make it
     appear that the 50 pound arm is weightless.  "However, when
     the operator touches something, the virtual forces become
     actual forces felt through the exoskeleton" (Lane and
     Smith).  This would make the operator's arm stop when it hit
     a virtual wall or feel the weight of a virtual object.
          The butler is a robot that basically gets in the way
     whenever you try to move through an object.  If a user
     reaches out his/her hand to touch a wall, desk, or any other
     virtual object, the butler robot will place a real object at
     the location where the virtual object is supposed to be. 
     This technique is currently being researched at the
     University of Tokyo by Susumu Tachi (Tachi 1995).  The
     butler being worked on "provides mechanical impedance of the
     environment, i.e., inertia, viscosity and stiffness" (Tachi
     1995).  The major drawback of the butler robot is that it
     can only present these properties for a single point at a
                           Tactile Feedback
          The butler robot under development can give the
     impression of stiffness and viscosity, but it can't present
     the information needed by a human to know what the object
     feels like.  The temperature and texture are totally unknown
     to the user.  It is possible to display the temperature by
     heat resistive wires sown into the lining of a glove.  
          The texture of a surface is probably the hardest
     feature of tactile feedback to simulate.  The closest
     documented attempt is the Sandpaper system.  This system,
     developed by a research group which includes members from
     MIT and UNC, can accurately simulate several different
     grades of sandpaper (Aukstakalnis and Blatner 1992).  Other
     systems, like the Teletact Commander, use either airfilled
     bladders sown into a glove, or piezo-electric transducers to
     provide the sensation of pressure or vibrations.  These
     systems have problems with the unreliability of compressors
     and interference between the piezo-electric transducer
     electromagnetic fields and the electromagnetic field used by
     a Polhemus tracking system (Stone 1993).  
          Any attempt to model the texture of a surface faces
     tremendous challenges because of the way the human haptic
     system functions.  There are several types of nerves which
     serve different functions, including:  temperature sensors,
     pressure sensors, rapid-varying pressure sensors, sensors to
     detect force exerted by muscles, and sensors to detect hair
     movements on the skin.  All of these human factors must be
     taken into consideration when attempting to develop a
     tactile human-machine interface.
                          Tracking Devices
          The purpose of a tracking device is to determine the x,
     y, and z position, and the orientation (yaw, pitch, and
     roll) of some part of the user's body in reference to a
     fixed point.  Most types of virtual reality interaction
     devices will have a tracker on them.  HMDs need a tracker so
     that the view can be updated for the current orientation of
     the user's head.  Datagloves and flying joysticks also
     usually have trackers so that the virtual "hand" icon will
     follow the position and orientation changes of the user's
     real hand.  Full body datasuits will have several trackers
     on them so that virtual feet, waist, hands, and head are all
     slaved to the human user.  
          When designing or evaluating a virtual reality system
     that will receive tracking information, it is important to
     pay attention to the latency (lag), update rate, resolution,
     and accuracy of the tracking system.  Latency is the "delay
     between the change of the position and orientation of the
     target being tracked and the report of the change to the
     computer" (Baratoff and Blanksteen).  If the latency is
     greater than 50 milliseconds, it will be noticed by the user
     and can even cause nausea or vertigo.  Update rate is the
     rate at which the tracker reports data to the computer, and
     is typically between 30 and 60 updates per second. 
     Resolution will depend on the type of tracker used, and
     accuracy will usually decrease as the user moves farther
     from the fixed reference point (Baratoff and Blanksteen). 
     Six-degree-of-freedom tracking devices come in several basic
     types of technology: mechanical, electromagnetic,
     ultrasonic, infra-red, and inertial.  
     Mechanical Trackers
          A mechanical tracker is similar to a robot arm and
     consists of a jointed structure with rigid links, a
     supporting base, and an "active end" which is attached to
     the body part being tracked (Sowizral 1995), often the hand. 
     This type of tracker is fast, accurate, and is not
     susceptible to jitter.  However, it also tends to encumber
     the movement of the user, has a restricted area of
     operation, and the technical problem of tracking the head
     and two hands at the same time is still difficult.
     Electromagnetic Trackers
          An electromagnetic tracker allows several body parts to
     be tracked simultaneously and will function correctly if
     objects come between the source and the detector.  In this
     type of tracker, the source produces three electromagnetic
     fields each of which is perpendicular to the others.  The
     detector on the user's body then measures field attenuation
     (the strength and direction of the electromagnetic field)
     and sends this information back to a computer.  The computer
     triangulates the distance and orientation of the three
     perpendicular axies in the detector relative to the three
     electromagnetic fields produced by the source (Baratoff and
          Electromagnetic trackers are popular, but they are
     inaccurate.  They suffer from latency problems, distortion
     of data, and they can be thrown off by large amounts of
     metal in the surrounding work area or by other
     electromagnetic fields, such as those from other pieces of
     large computer equipment.  In addition, the detector must be
     within a restricted range from the source or it will not be
     able to send back accurate information (Sowizral 1995), so
     the user has a limited working volume.
     Ultrasonic Trackers
          Ultrasonic tracking devices consist of three high
     frequency sound wave emitters in a rigid formation that form
     the source for three receivers that are also in a rigid
     arrangement on the user.  There are two ways to calculate
     position and orientation using acoustic trackers.  The first
     is called "phase coherence".  Position and orientation is
     detected by computing the difference in the phases of the
     soundwaves that reach the receivers from the emitters as
     compared to soundwaves produced by the receiver.  "As long
     as the distance traveled by the target is less than one
     wavelength between updates, the system can update the
     position of the target" (Baratoff and Blanksteen).  The
     second method is "time-of-flight", which measures the time
     for sound, emitted by the transmitters at known moments, to
     reach the sensors.  Only one transmitter is need to
     calculate position, but the calculation of orientation
     requires finding the differences between three sensors
     (Baratoff and Blanksteen).
          Unlike electromagnetic trackers that are affected by
     large amounts of metal, ultrasonic trackers do not suffer
     from this problem.  However, ultrasonic trackers also have a
     restricted workspace volume and, worse, must have a direct
     line-of-sight from the emitter to the detector.  Time-of-
     flight trackers usually have a low update rate, and phase-
     coherence trackers are subject to error accumulation over
     time (Baratoff and Blanksteen).  Additionally, both types
     are affected by temperature and pressure changes (Sowizral
     1995), and the humidity level of the work environment
     (Baratoff and Blanksteen).
     Infrared Trackers
          Infrared (optical) trackers utilize several emitters
     fixed in a rigid arrangement while cameras or "quad cells"
     receive the IR light.  To fix the position of the tracker, a
     computer must triangulate a position based on the data from
     the cameras.  This type of tracker is not affected by large
     amounts of metal, has a high update rate, and low latency
     (Baratoff and Blanksteen).  However, the emitters must be
     directly in the line-of-sight of the cameras or quad cells. 
     In addition, any other sources of infrared light, high-
     intensity light, or other glare will affect the correctness
     of the measurement (Sowizral 1995).
     Inertial Trackers
          Finally, there are several types of inertial tracking
     devices which allow the user to move about in a
     comparatively large working volume because there is no
     hardware or cabling between a computer and the tracker. 
     Inertial trackers apply the principle of conservation of
     angular momentum (Baratoff and Blanksteen).  Miniature
     gyroscopes can be attached to HMDs, but they tend to drift
     (up to 10 degrees per minute) and to be sensitive to
     vibration.  Yaw, pitch, and roll are calculated by measuring
     the resistance of the gyroscope to a change in orientation. 
     If tracking of position is desired, an additional type of
     tracker must be used (Baratoff and Blanksteen). 
     Accelerometers are another option, but they also drift and
     their output is distorted by the gravity field (Sowizral
                         Interaction Devices
          Virtual reality and virtual environments go far beyond
     typical interfaces in the realism of the visual metaphor. 
     Point and click with a table-top mouse is wonderful in some
     situations, but not nearly sufficient for an immersive
     environment.  So instead of a keyboard and mouse,
     researchers are developing gloves, 3D mice, floating
     joysticks, and voice recognition.  This paper will not
     attempt to cover voice recognition because it is such a
     large domain.
          For sensing the flexion of the fingers, three types of
     glove technology have arisen: optical fiber sensors,
     mechanical measurement, and strain gauges.  The Dataglove
     (originally developed by VPL Research) is a neoprene fabric
     glove with two fiber optic loops on each finger.  Each loop
     is dedicated to one knuckle and this can be a problem.  If a
     user has extra large or small hands, the loops will not
     correspond very well to the actual knuckle position and the
     user will not be able to produce very accurate gestures.  At
     one end of each loop is an LED and at the other end is a
     photosensor.  The fiber optic cable has small cuts along its
     length.  When the user bends a finger, light escapes from
     the fiber optic cable through these cuts.  The amount of
     light reaching the photosensor is measured and converted
     into a measure of how much the finger is bent (Aukstakalnis
     and Blatner 1992).  The Dataglove requires recalibration for
     each user (Hsu).  "The implications for longer term use of
     devices such as the Dataglove--fatigue effects,
     recalibration during a session--remain to be explored"
     (Wilson and Conway 1991).
          The Powerglove was originally sold by Mattel for the
     Nintendo Home Entertainment System but, due to its low
     price, has been used widely in research (Aukstakalnis and
     Blatner 1992).  This Powerglove is less accurate than the
     Dataglove, and also needs recalibration for each user, but
     is more rugged than the Dataglove.  The Powerglove uses
     strain gauges to measure the flexion of each finger.  
               A small strip of mylar plastic is coated with an
          electrically conductive ink and placed along the length
          of each finger.  When the fingers are kept straight, a
          small electrical current passing through the ink
          remains stable.  When a finger is bent, the computer
          can measure the change in the ink's electrical
          resistance (Aukstakalnis and Blatner 1992).
          The dexterous hand master (DHM) is not exactly a glove
     but a exoskeleton that attaches to the fingers with velcro
     straps.  A mechanical sensor measures the flexion of the
     finger.  Unlike the Dataglove and Powerglove, the DHM is
     able to detect and measure the side-to-side movement of a
     finger.  The other gloves only measure finger flexion.  The
     DHM is more accurate than either of the gloves and less
     sensitive to the user's hand size, but can be awkward to
     work with (Hsu).
          The main strength of the various types of gloves is
     that they provide a more intuitive interaction device than a
     mouse or a joystick.  This is because the gloves allow the
     computer to read and represent hand gestures.  Objects in
     the environment can therefore be "grasped" and manipulated,
     the user can point in the direction of desired movement,
     windows can be dismissed, etc (Wilson and Conway 1991). 
     "Gestures should be natural and intuitive in the particular
     virtual environment.  Actions should be represented visually
     and be incremental, immediate, and reversible to give a
     person the impression of acting directly in an environment"
     (Dennehy).  Wilson and Conway (1991) say that a basic set of
     command gestures for gloves has been developed, but that
     more work is needed to expand the set beyond the current
     simple mapping.  Another area of improvement is  feedback
     for the user to aid hand-eye coordination and proprioceptive
     feedback to let a user know when an object has been
     successfully grasped (Wilson and Conway 1991).
     3D Mice
          There are several brands of 3D mice available, all with
     basically the same technology: A mouse or trackball has been
     modified to include a position and orientation tracker of
     some kind (Aukstakalnis and Blatner 1992).  This modified
     mouse is fairly familiar and intuitive to users--simply push
     the mouse in the direction you want to move.  However, these
     mice are not very useful for interactions other than
     navigation and selection of objects (Hsu).
          The final category of interaction device is the wand or
     floating joystick.  Basically, this device works exactly the
     same as a conventional joystick, but it is not attached to a
     base that sits on a table top.  Instead, the joystick is
     equipped with an orientation tracker so the user simply
     holds it in their hand and tilts it.  Most flying joysticks
     also have some buttons on the stick for "clicking" or
     selecting, similar to a mouse (Hsu).
                Human Factors in Virtual Environments
          Kay Stanney (1995) has written an excellent critique of
     the areas that still need to be researched in order to make
     virtual environments a safe place to work.  These include
     health concerns such as "flicker vertigo" which can induce a
     seizure, auditory and inner ear damage from high volume
     audio, prolonged repetitive movements which cause overuse
     injuries (for example, carpal tunnel syndrome), and head,
     neck, and spine damage from the weight or position of HMDs. 
     Safety factors also need to be considered.  For example,
     when a user's vision is restricted by an HMD, they are
     likely to trip and fall over cables or other real world
     objects.  Also, how safe is the user from harm in the event
     of system failure?  Hands and arms might be pinched or over
     extended if a haptic feedback device fails; the user might
     be disoriented or harmed if the computer crashes and
     suddenly dumps the user into reality, disrupting the sense
     of "presence".
          Virtual reality holds promises of being the "ultimate
     human-computer interface".  This would incorporate a
     natural, intuitive interface between a human and a machine
     generated work environment.
               An intuitive interface between man and machine is
          one which requires little training . . . and proffers a
          working style most like that used by the human being to
          interact with environments and objects in his day-to-
          day life.  In other words, the human interacts with
          elements of this task by looking, holding,
          manipulating, speaking, listening, and moving, using as
          many of his natural skills as are appropriate, or can
          reasonable be expected to be applied to as task (Stone
          This paper has overviewed the technology currently in
     use, in addition to the open areas of research.  Visual
     display devices, graphics display techniques, 3D audio,
     haptic feedback, navigation, and interaction devices are all
     in need of more development.  Large areas of concern about
     the health and safety of the user are still in focus, not to
     mention the unsolved technical problems standing in the way
     of an intuitive immersive environment.  As the public market
     for virtual reality grows in the coming years, more money
     will be spent on quality interface improvements and some of
     these problems may be solved.
     Aukstakalnis, Steve, and David Blatner.  1992. Silicon 
          Mirage The Art and Science of Virtual Reality.
          Berkeley, California: Peachpit Press, Inc.
     Baratoff, Gregory, and Scott Blanksteen. Unknown.
          Encyclopedia of Virtual Environments. World Wide Web URL
     Blanchard, Jim, and Reiko Tsuneto. Unknown. Stereoscopic
          viewing. Encyclopedia of Virtual Environments. World
          Wide Web URL:
     Bolas, Mark T. 1994. Human factors in the design of an
          immersive display. IEEE Computer Graphics and     
          Applications January: 55-59.
     Brooks, Fred. 1995. Panel Discussion on March 13, 1995 at
          the IEEE Virtual Reality Annual International Symposium
          in Research Triangle Park, North Carolina.
     Dennehy, Michael. Unknown. Encyclopedia of Virtual
          Environments. World Wide Web URL:
     Foley, J., A. van Dam, S. Feiner, and J. Hughes. 1992.      
          Computer Graphics Principles and Application, second
          edition. Reading, Massachusetts: Addison-Wesley.
     Gomez, Daniel, Grigore Burdea, and Noshir Langrana. 1995.
          Integration of the Rutgers Master II in a Virtual
          Reality Simulation. In IEEE: Proceedings of the Virtual
          Reality Annual International Symposium in Research
          Triangle Park, NC, March 11-15, 1995, IEEE Computer
          Society Press. Washington: IEEE Computer Society Press,
     Hsu, Jack. Unknown. Encyclopedia of Virtual Environments. 
          World Wide Web URL:
     Lane, Corde, and Jerry Smith. Unknown. Encyclopedia of 
          Virtual Environments. World Wide Web URL:
     Lane, Corde. Unknown. Displays. Encyclopedia of Virtual
          Environments. World Wide Web URL: /projects/eve/eve-main.html
     Sowizral, Henry. 1995. Tutorial: An Introduction to Virtual
          Reality. Virtual Reality Annual International
     Stanney, Kay. 1995. Realizing the Full Potential of Virtual
          Reality: Human Factors Issues That Could Stand in the
          Way. In IEEE: Proceedings of the Virtual Reality Annual
          International Symposium in Research Triangle Park, NC, 
          March 11-15, 1995, IEEE Computer Society Press. 
          Washington: IEEE Computer Society Press, 28-34.
     Steinmetz, Joe, and Glen Lee. Unknown. Encyclopedia of
          Virtual Environments. World Wide Web URL:
     Stereo is dead. The AudioReality story. [1994]. Palo Alto,
          California: Crystal River Engineering.
     Stone, Robert J. 1993. Virtual reality systems. Edited by
          R.A. Earnshaw, M.A. Gigante, and H. Jones. Virtual
          Reality: A tool for telepresence and human factors
          research. London: Academic Press.  
     Tonnesen, Cindy, and Joe Steinmetz. Unknown. Encyclopedia of
          Virtual Environments. World Wide Web URL:
     Tachi, Susumu. 1995. Whither Force Feedback?. In IEEE:
          Proceedings of the Virtual Reality Annual International
          Symposium in Research Triangle Park, NC, March 11-15,
          1995, IEEE Computer Society Press. Washington: IEEE
          Computer Society Press, 227.
     Watson, Benjamin A., and Larry F. Hodges. Unknown. Using
          Texture Maps to Correct for Optical Distortion in
          Head-Mounted Displays. World Wide Web URL:"
     Wilson, Michael, and Anthony Conway. 1991. Enhanced
          Interaction Styles for User Interfaces. IEEE Computer 
          Graphics and Applications March: 79-89.