A Virtual Retinal Display For Augmenting Ambient Visual Environments
by Michael Tidwell
![[Previous Chapter]](/icons/button-back.gif)
![[Table of Contents]](/icons/button-contents.gif)
Chapter 9:Future Work and Conclusions
9.1 Conclusions
A full color (RGB), 40 [deg.] horizontal field, 30[deg.] vertical
field, 640 horizontal by 480 vertical picture element (VGA) retinal
scanning display has been demonstrated and characterized for this
thesis. Experiments show that individual picture elements and
scan lines are resolvable in the display which indicates that
the resolution is truly 640 by 480 elements. The display also
performed effectively in a see-through mode exhibiting sufficient
brightness to be viewed against a normal room illumination background.
The technology in this thesis should be extendible and adaptable
to an SVGA (1280 by 1024) display.
9.2 Areas for Improvement
Performance limiting technical issues were presented in the previous
chapter. Remaining technical problems and potential solutions
will be discussed in this chapter including work in the areas
of light sources (Section 9.3) , scanning technology (Section
9.4), viewing optics (Section 9.6), and visual cues (Section 9.7).
Ways to improve the small exit pupil of the system are also discussed
in Section 9.5.
It is necessary for any future work to include the needs of the
end user. The applications are diverse and some are highly technical.
For example, if a major application for an augmented vision system
is in anesthesiology and, by chance, anesthesiologists perform
better using yellow augmented displays instead of RGB augmented
displays, then the sooner it is known the better. Since little
work has been done in the area of see-through head mounted displays
for applications other than the military, more work is needed
to find out what a user needs from this type of display and why.
9.3 Light Sources - Light Emitting Diodes
An important characteristic of any light source for the VRD is
modulation bandwidth or more precisely, the modulation transfer
function. It is desired that the light source maintain a high
modulation depth at high frequencies. There is indeed a tradeoff
between modulation bandwidth and luminance in most LED structures.
Although lasers with external modulators are presently available,
directly modulated semi-conductor lasers are not presently available
in blue and green. Although semi-conductor laser are a potential
future solution to the problem of unsatisfactory light sources,
light emitting diodes may offer a more viable solution in terms
of cost and size.
The light sources for this project are a 650 [nm] semiconductor
laser, a 543.5[nm] HeNe, and an air cooled 488[nm] argon laser.
These light sources can be considered neither inexpensive nor
small (or lightweight). Light emitting diodes (LED's) may offer
an alternative to bulky and expensive lasers. LED's of sufficient
intensity could be coupled to an optical fiber. A different coupling
scheme from the coupling scheme described earlier, however, would
be required. If the emission area an LED were small enough (on
the order of 5[mm]) and of the correct
geometry (namely circular), an LED could be shined directly onto
the horizontal scanner with a minimum of corrective optics. It
should be noted that an LED with a traditional rectangular emission
area would have an astigmatic wavefront and require corrective
optics to compensate for the astigmatism. A rectangular emission
aperture causes orthogonal planes of the wavefront to have different
focal points (i.e. astigmatism).
9.4 Scanning Technology
9.4.1 Overview of Scanner Development
The optical beam scanner for this project is a novel is a novel
moving mirror device based on a resonant electromechanical system
[45]. The important scanner characteristics relevant to the VRD
are scan angle (or scan deviation), scanner aperture (beam size
at scanner), scanner speed, and also size and weight. The Mechanical
Resonant Scanner (MRS) has a small form factor and is light weight
at 2 [cm3] and approximately 2 [oz.]. The scanner design,
as partly shown in Section 6.2.2, is a complex problem. More work
is needed to reduce scanner volume while increasing scan frequency
and mechanical deviation.
9.4.2 Mechanical Resonant Scanner Phase Detection
The frequency of the scanner is temperature dependent and may
drift. One way to curtail the effects of drift on display quality
is to adjust the phase difference between the light modulation
and the scanner position by adding delay to the light modulation.
If the phase difference is to be adjusted automatically, the phase
difference between the light modulation and the scanner position
over time (or scanner timing) must be known. Although the modulation
timing is easily determined in the electronics, the scanner position
over time is more difficult to predict. One idea for measuring
scanner phase involves detecting back electro-motive forces in
the scanner coils [46]. Once the scanner timing can be detected,
an electronic circuit will continually reduce the phase difference
between the light modulation and scanner timing by adjusting the
light modulation delay.
9.4.3 Sinusoidal Scan Correction
Since the horizontal scanner is a mechanically resonant device,
the deflection of the scanner mirror is sinusoidal over time.
Therefore, the scanner is moving with greater speed at the center
of its motion (through the at rest position) and slower at the
extremes of its motion. Consequently, a constant modulation rate
results in a distorted image which is expanded in the middle of
the horizontal scan and compressed in the middle.
A solution for this problem is too modulate the pixels at a
faster rate in the middle of the display. Much of the noticeable
distortion occurs at the outermost 5% - 10% of the display. Blanking
the outermost 5% - 10% may make sinusoidal correction easier.
9.4.4 Increasing the MRS Scan Frequency to Super VGA
The SVGA graphics format (1280 1024 picture elements) is the next
step in terms of improving resolution for the Virtual Retinal
Display technology. A bi-directional horizontal scanner must have
a scanning frequency of 30.72 [kHz] to achieve SVGA resolution
with a 60 [Hz] frame rate. Initial experiments indicate that a
30.72 [kHz] scanning frequency is obtainable from the MRS with
only slight reduction in optical deflection [46].
9.4.5 Improving Mirror Surface Quality
By experimenting with different scanner mirrors, the scanner mirror
was found to be the main cause of the astigmatism afflicting the
optical performance of the system. However, improved polishing
techniques have proved to significantly decrease the astigmatism
introduced to the scanned wavefronts by the scanner mirror [46].
A Virtual Retinal Display using an improved scanner mirror has
recently been built which provides improved performance to the
extent that individual pixels are easily resolved.
9.5 Exit Pupil Size
9.5.1 Scanner Aperture
It is evident that a larger exit pupil is required for this display
to be practical in most applications. One way to achieve a larger
exit pupil with the MRS approach is to increase the mirror size
(see Section 6.2.2). However, it is a difficult mechanical engineering
problem to increase scanner size and speed simultaneously. In
the mechanical system of the MRS, increasing the mirror size tends
to decrease scanner frequency and/or deflection.
9.5.2 Exit Pupil Duplication
One way to increase the apparent exit pupil may be to duplicate
the exit pupil with a grating or diffuser [44]. This method implies
that a real image somewhere before the final eyepiece is being
diffused. Initial experiments show promise for inclusive systems
but the real image for an augmented vision display may not be
accessible.
9.6 Viewing Optics
A more compact optical design for the viewing optics may be necessary
for a head mounted display. In the current design, the location
of the scanner below the eye may disrupt downward vision. An approach
where the scan comes into a partially reflective element from
the side of the head would be more desirable for an operational
head mounted display. However, any system incorporating the scan
from the side of the head would probably be an off-axis optical
system. Off-axis optical systems (as opposed to on-axis systems
used in this project) inherently contain more complicated aberration
and optical problems [3]. The final reflective element may be
partially reflective plastic or perhaps a holographic combiner.
The most desirable configuration for a head mounted version of
the system would resemble eyeglasses more than it would the traditional
helmet type head mounted displays.
9.7 Variable Accommodation
One problem in current helmet mounted display technology is the
conflict between vergence and accommodation cues. Both cues assist
in telling the brain the depth of an object. The vergence cue
is related to the angle between the two eyes - the closer the
object the greater the angle. The accommodation cue is related
to where the eyes are focused - the closer the object the shorter
the focal length of the eye. In current helmet mounted displays,
the focal length must remain fixed because the screen is focused
to a constant distance. Vergence continually changes, however,
in stereoscopic helmet mounted displays. The discrepancy between
the vergence and accommodation cues may potentially cause a user
to feel somewhat ill [47].
The Virtual Retinal Display has a unique potential to create
picture element specific wavefronts entering the eye. The VRD
generates each pixel at different points in time and it is therefore
possible to modulate the light wavefront for each picture element.
The challenge for the implementation of this potential is to devise
a variable focal length lens with switching speeds on the order
of the pixel modulation switching speed.
9.8 Size and Weight
The eventual size and weight of a head mounted version of the
display should be minimized. The design and construction of a
head mounted display involves optical, mechanical, electrical,
and industrial design. Furthermore, the final decision of size
weight (and cost) should depend upon specific applications. Some
applications have stricter tolerances on size and weight. Also
the additional weight of the display produces a rotational moment
which may contribute to fatigue. It matters not only how much
the display weighs but also where the weight is in relation to
the center of inertia for the head's muscular and skeletal system.
Ideally for longer duration applications, however, the display
should weigh less than 16 [oz.].
9.9 Summary
Remaining research relates primarily to system size and system
performance. Smaller light sources are needed to make the display
portable. Light emitting diodes or directly modulated laser diodes
show the best potential to solve the problem. The exit pupil size
is a serious and challenging problem for this device. The exit
pupil problem must be solved to allow the user to gimbal the eye
while looking around in the display. The Mechanical Resonant Scanner
should incorporate phase detection in the future. Sinusoidal correction
in modulation is also needed to compensate for the sinusoidal
scan of the MRS.
The demonstration of a full color (RGB), augmented vision (see-through),
Virtual Retinal Display represents a major step in pioneering
efforts to create high performance, affordable, visually augmented
displays. The display technology in this thesis, the Virtual Retinal
Display, brings us closer to practical applications of virtual
reality and augmented vision in fields such as medicine, manufacturing,
and personal communications where overlaying computer information
over the real world can enhance performance.
Human Interface Technology Laboratory