Content uploaded by Stefan Rapp
Author content
All content in this area was uploaded by Stefan Rapp on Feb 23, 2014
Content may be subject to copyright.
Spotlight Navigation: a pioneering user interface for
mobile projection
Stefan Rapp
Conante
Fellenoord 130
5611ZB Eindhoven, The Netherlands
rapp@conante.com
alternative address
Dr. Stefan Rapp
Uhlandstr. 10
78559 Gosheim, Germany
ABSTRACT
Spotlight Navigation [4, 3] is - to the best of the author’s current
knowledge - historically the first general purpose user interface
(UI) that has specifically been designed for mobile projection de-
vices. From a user perspective, access to information through the
projection device is performed by panning over a projected infor-
mation space by moving the projection device as a whole, and by
zooming into interesting regions using dedicated controls or spe-
cific gestures. The rendering component of Spotlight Navigation
updates the projected image in real time according to the move-
ments of the device, such that for users it appears as if they would
interact on a huge virtual screen of which only the illuminated part
within the projector’s light cone becomes visible. In the paper we
discuss fundamental aspects that differ from traditional desktop or
hand held device interaction and describe interaction techniques
that have been found suitable for mobile projection. All features
and techniques presented here have been implemented on a work-
ing prototype that served as a basis for evaluation and elaboration
of the techniques. We also discuss the setup and implementation
architecture of our prototypes in more detail, and report on previ-
ously unpublished features of the prototype device, such as specific
gestures, ripples and a text selection widget.
1. INTRODUCTION
Mobile projectors can be considered just a miniaturization of pro-
jectors for business presentation or home cinema use. They may
also be seen as just another form of display technology for mo-
bile devices, one that allows for larger screen space, despite a small
form factor. Albeit all that is basically true, it is missing the essence:
mobile projection devices are a new device class that deserve to
have a specifically tailored user interface, to maximize its useful-
ness to the benefit of users. Spotlight Navigation has been devel-
oped along these lines, triggered by a simple question of how users
may access and alter ultra high resolution information content when
on the move. By using Spotlight Navigation as an interface, mobile
projectors can be useful as an efficient, general purpose informa-
tion processing device, and not just as a device to show movies or
slide presentations, when plugged to a phone. So if users want to do
more – how can they best operate a mobile projection device? Will
they be using dedicated controls such as keys or a joystick on the
device? Will they be using a pointing device, or interact directly on
the projected image? In the author’s view, none of these ’standard’
ways of interacting with computers that are now common place for
PC, tablet/PDA, (smart)phone, table-top surface or powerwall are
best suited for mobile projection. As the usage context is so differ-
ent from desktop or smart phone use, it is in order to rethink also
some fundamental interaction principles.
2. USAGE CONTEXT
But let us start by looking at the usage context of a mobile projector.
We postulate that users will want to use a projection device in ad-
hoc situations, at various places and intermittent or occasional over
a working day or during their free time, to quickly access informa-
tion, or show information to others. As they will probably not carry
a projection screen with them, they will want to project on any re-
alistically suitable surface, such as a table, a wall, a sheet of paper,
the floor, a book, or, if all else fails, a shirt, a curtain, a sleeve or
the palm of their hand. If we look around, there are always suitable
projection surfaces in convenient reach, and they tend to be surpris-
ingly large, much larger than what current technology can handle
with respect to brightness and contrast. The projection distance will
vary, but it will be limited to comparably small distances due to the
inferior light levels that are achievable with a battery-powered de-
vice of today’s technology. We observe that there will not always
be the possibility to lay down the device in an appropriate place,
so in general, the projection will be from the device in the user’s
hand to a rather homogeneous, flat surface. In this situation, due
to the naturally occurring small movements of the muscles, the so
called tremor, projection is not perfectly stable. In consequence,
if the physical resolution is relatively high, say XGA, it is advis-
able to prepare content as if for SVGA-resolution or lower, so that
the information is still well readable also under tremor condition.
The problem with tremor is not the translational variance but the
rotational. If the projector is jittering left-right perfectly parallel
(that is, without being rotated) by half a millimeter, the projected
image is also jittering half a millimeter only. But if a projector of
10 cm length is only slightly rotated around its center, so that the
front is shifted half a millimeter to the left, and on the back half a
millimeter to the right, the projected image will shift already for 1
cm at one meter projection distance. It is also evident that any op-
eration of keys or other controls that require force inevitably leads
to movements of the projected image, as it can not be avoided to
induce a tiny rotation on the device during operation. Rotations
dominate translations. On the other hand, it is very easy to adapt
the projector to a good position with respect to the projection sur-
face, by intuitive and direct translation and rotation of the projector
– everybody who ever operated a flashlight can do that, and it is
much simpler than with stationary projectors. Also scanning over
an extended surface or changing from surface to surface is instant
and intuitive when the projector is in the user’s hand.
3. THE FLASHLIGHT METAPHOR
People interacting with Spotlight Navigation appear as if they are
searching for something with a flashlight. They shine a light here
and there, stay on some spot for some moment and continue further.
Alas, while acting in the real world, they are searching in the virtual
world, and it is a portion of the virtual data space that they see in
the projected light spot. While operating the device, its posture
is continually measured and a corresponding image of the virtual
data space is generated in real time in such a way, that the illusion
is generated that the projected content appears stable and fixed to
the surface, as if it would be a part of the real world. This linking of
location between the virtual and the real world is a major factor in
the device’s intuitivity. The projected light is unveiling a part of the
virtual landscape, like making invisible ink visible with ultraviolet
light. Searching for information gets as easy as using a flash light.
Pan and zoom user interface. Access to information within Spot-
light Navigation is in general achieved by panning over a (theoret-
ically) unlimited and resolution-less two dimensional information
space. The panning is achieved directly by moving the device as a
whole, to steer the light spot into the direction of interest. As previ-
ously explained, rotation dominates translation, so the typical style
of panning the information space is to slightly rotate the device in
the hand, while keeping its position. With effortless small move-
ments, large distances on the projection surface can be quickly trav-
elled. As the content stays at the same spot, it is easy to build a
mental model of the information space, to quickly re-access previ-
ously seen items that are no longer visible with a flick of the hand.
Once an interesting aspect in the data space is pointed at, users can
zoom into the data space, either by a gesture or by simple controls
on the device. Such a gesture could be approaching the projection
surface with the whole device, or a gesture on a track pad or other
touch sensitive surface on the device. In the Spotlight Navigation
prototypes, the only way to zoom is via a scroll wheel on top of the
device and under the user’s thumb. Zooming in means that the area
pointed at is enlarged and more details or additional information is
becoming visible. While being independently developed, Spotlight
Navigation resembles in many aspects Raskin’s ’ZoomWorld’ con-
cept [5]. Whereas ZoomWorld is operating on a standard monitor
and with standard input devices, it is the author’s firm believe that
through the direct pointing gestures of Spotlight Navigation, also to
’off-screen’ locations, the panning and coordination with zooming
is much more intuitive, and that this is a key enabler for making
the pan and zoom interface actually work. Clutching. Panning and
zooming is sufficient to travel across the unlimited data space even
on a limited projection surface. If longer distances need to be trav-
elled, zooming out, panning in the miniaturized view and zooming
in again works. If needed, the sequence can be repeated after pan-
ning back. This operation is not very intuitive for beginners, so
we introduced ’clutching’, that can be used when through panning
and zooming users have reached the end of a suitable projection
area. During clutching (triggered by pressing the scroll wheel) the
tricky processing of keeping things where they were is temporarily
switched off, and the area currently shown is following with the
device’s movements over the projection surface. With it, the whole
data space is translated to the new position: clutching is dragging
the whole data space to a more convenient position, similar to mov-
ing a sheet of paper when sketching. Clutching becomes similarly
intuitive and unconscious after a very short time. Panning, zooming
and clutching together are a powerful way to explore and access an
unlimited data space along visual cues.
4. IMPLICATIONS FOR INTERACTION
Having an unlimited, unbounded work area immediately makes
several fundamental assumptions void, that we are so used to from
our PC or mobile phone experience. Over the past decades, screen
real estate has more or less stayed in the same order of magnitude,
despite the advancements in computational power or memory. The
predominant user interface paradigm can be seen as all about saving
pixel space: windows are overlapping because there is not enough
room for all; they are iconified to ease the management of screen
space; there are menus that drop down or up so that they do not
waste space while not in use; menus must be made hierarchic to
save space; there are scroll bars as we do not have enough space for
larger data sets or images etc. With an intuitive panning and zoom-
ing access to an arbitrarily large and resolution-less workspace,
many if not all of these interface design choices should be revis-
ited, to see if they are still reasonable. A reason to keep them might
be that people are used to the concepts from the PC or phone, and
that it may be easier to port existing applications or develop plat-
form independent solutions. To fully leverage the potential of the
interface, it is advisable to go for alternative approaches: Rather
than scrolling with a scroll bar, in Spotlight Navigation, users can
intuitively use panning and zooming directly. While scroll bars give
only little help in where in the overall picture the current view port
is, as the visual feedback is rather small and indirect, with Spotlight
Navigation there is a direct and prominent feedback via the location
of the projected light spot. Rather than overlapping windows, Spot-
light can use tiling of windows so that nothing is obscured. Appli-
cations can run concurrently side-by-side, and switching between
them needs nothing more than a flick of the hand, no need for task
managers or a task bar. There is also no explicit form of iconifica-
tion, of course items get smaller when zooming out and larger when
zooming in. This may be viewed as some form of iconification, but
it is gradual, in contrast to switching from one state to the other.
Focussed work style. In principle getting rid of the various forms
of hiding information to save space should make it easier to navi-
gate, as everything is readily visible. Of course, due to the zooming,
not all details will be always visible, but it is easy to zoom into and
check for the details, also see ’semantic zooming’ below. Hiding
also has the task of structuring (as in menus or folder hierarchies)
as well as focusing on relevant information and filtering out the ir-
relevant for the task at hand (such as opening only a tiny portion
of the documents available on storage). With Spotlight Navigation,
as we have unlimited workspace, the decision on what is relevant
and what is not is done by the user, selecting which portion of the
data to view. In Spotlight Navigation, we call this focused work
style. There are no categorical boundaries, everything is seamless:
users may focus on a set of architectural plans, on the layout of the
children’s rooms in the first floor, or on the placement of a window
or door. There are no abruptly changing views in spotlight navi-
gation. Recent developments such as the UI for the iPhone take
care that transitions in appearance are always animated, so that the
users get a clue on what is happening, that they do not get too much
confused by changing screen content. Spotlight Navigation precon-
ceived this and implements it throughout: it is the users who select
the good view and appropriate level of detail in their constant, daily
routine operations. For example when working with the calendar,
no explicit interactions are needed to switch from a week to a month
view or change to the next week, everything is accomplished by
the same panning and zooming used throughout the interface. This
smoothness and being in control makes the behavior of artifacts in
the virtual world more similar to what we are used to in the real
world. Users can easily lay out or rearrange information to their
needs, and can utilize their spatial memory to organize things. It is
advantageous to also include visual cues, or landmarks, in the data
space so that visual orientation is easier while navigating through
it. The interface thereby takes care to not overload users with irrel-
evant information through semantic zooming – depending on the
zoom level, the way that information is displayed can be different.
For example the clock, when viewed from a distance, only shows
the hour and minute hand, whereas if users zoom in, more and more
information is shown, such as the second hand, a day-of-month
display etc. Similarly, in the calendar, the individual scheduled ap-
pointments only get visible when users have zoomed in sufficiently
that this makes sense. Otherwise, every day is only represented by
its day-of-month number. In the Spotlight Navigation prototype,
semantic zooming is often implemented by mipmapped textures,
that have different information at the various mipmap levels. The
rendering hardware then automatically selects the appropriate rep-
resentation based on the currently visible size, and even implements
a smooth blending over at no extra implementation cost.
Gestures. Apart from the deictic pointing gestures for selecting the
view or point-and-click interaction, further gestures are used. First,
the orientation of the working surface in relation to the projection
device is defined by holding the device orthogonal to the projection
surface during power on. The device will store this orientation until
it is switched off. In practice, small misalignments are not prob-
lematic for interaction. Second, there is a specific home gesture.
When users get lost in the data space by zooming in or out too far,
quickly pointing to the ground and up again brings the data space
scaled back full screen in the projection. Finally, there is a gesture
for swapping walls that is especially useful for working in the cor-
ner of a room. As the device has no sense of where the corners of a
room are, the trick is to build on the assumption that in most cases,
adjacent walls are orthogonal to each other. In order to swap walls,
users can simply point to the other wall, and as soon as they exceed
somewhat more than 90 degrees, the virtual data space is swapped
to the other wall by rotating it by 90 degrees. If walls are not
perpendicular to each other, users have to switch off and on again
to define the orientation of the new projection surface. Swapping
walls also works with floor and ceiling. By turning the device up,
as soon as the zenith is exceeded by some degrees, the data space
is flipped on the ceiling, or similarly on the floor after pointing a
bit behind the stand point. On floor and ceiling, it is also possible
to rotate the data space by clutching and rotating the device in the
appropriate direction. Here, clutching not only translates the data
space but also rotates it, which is specifically useful, if users want
to show information to others sitting on the opposite side of a table.
Pointer interaction. Spotlight Navigation can build on many suc-
cessful interaction techniques developed in the last decades. For ex-
ample, point-and-click or direct manipulation techniques, as well as
drag-and-drop can directly by used. Rather than having a pointing
device that can be moved within the current view to define the ac-
tion point, Spotlight Navigation has a cross in the center of the view
that defines it, so manipulating the virtual world is tightly coupled
to the routine pan and zoom gestures. As an example, in our proto-
type’s calendar application, scheduled appointments can easily be
moved to some hours later within a day by pointing to it, pressing a
button, dragging to the right time and releasing the button (just like
in most PC applications). In contrast to other applications, through
the panning and zooming, exactly the same method can be used
to move it also to the next day, week, month or a completely new
day all across the whole year. Similarly, other interactions known
from traditional UIs can be directly utilized in Spotlight Naviga-
tion, like pressing a button, moving a lever, adding a check mark to
an accomplished task and so on.
Scribbling. For text input, Spotlight utilizes a style similar to pen
input, called scribbling. Similar to writing on a wall with a laser
pointer, virtual ink is written in the projected area, following the
hand’s tiny rotational movements. It is comparably easy to write
that way, more difficult than when using a pen directly, but seem-
ingly simpler than writing with a mouse, as informal experimenta-
tion with a group of users has shown. This might be attributed to
a tremor cancellation component that was developed that smoothes
out the movements, so that the writing becomes less jittery and
more readable and appealing. While this form of input is not suit-
able for continued text input, it is sufficient to take small notes,
annotations or take down an appointment in the calendar, or write a
small message. While in the Spotlight Navigation prototypes, just
the strokes of the scribbled input are stored, it would also be possi-
ble to utilize a handwriting recognition component like with PDAs.
Text selection widget Another common user interface task is the
selection among options, for example a file from a directory, or
the selection of a country name from a drop-down-box in an ad-
dress form. The latter is quite common, although list- or drop-down
boxes are not the first choice when implementing a selection among
so many entries such as all the countries of the world. A hierarchi-
cal grouping along continents may be preferable for instance. But
the example illustrates, that developers tend to think along func-
tional and not necessarily along usability lines. The goal for a text
selection widget was therefore, to offer a single solution that scales
well so that usability is included ’by design’, to work reasonably
well for a handful, several hundred or even thousands of entries,
with the same fundamental mechanism. The solution is inspired by
a text selection widget presented by Toshiyuki Masui et al. [2] but
is adapted for usage with the Spotlight Navigation gestures. The so-
lution takes advantage of the projected zooming interface in that it
initially renders only a part of the entries, while hiding the majority
in ’folds’ that are also visualized graphically. Then, using zoom-
ing, more and more items can be made visible by the users, while
they pan to the fold or folds of interest. Thus, instead of scrolling
through an overly long list linearly, access to the wanted item is
much faster: it is more like a logarithmic than a linear access (de-
pending on the nature of the folds). For this to work efficiently, the
way of ’folding’ is important, and, that there is an ordering of the
items such as the alphabetic order. Traditionally, in a long list, the
first few entries are presented to a user, while all others are hidden,
or, in the terminology used here, are in one large ’fold’ at the end.
Like in Masui’s work, and in contrast to the common way, in Spot-
light’s text selection widget, we start with several folds, ideally of
comparable size. So, if we are to show the list of countries, instead
of showing the first 10 Afghanistan, Albania, . . . down to Arme-
nia, we show for instance Bolivia, Cyprus, Georgia, . . . down to
Zimbabwe, with stylized folds in between. If you want to look up
China, say, pan to the fold between Bolivia and Cyprus and zoom
in. After only a few zoom steps, China will become visible. Now,
either click on it or pan over it and zoom out until there is only
China left and thus selected. An obvious improvement is to not use
blindly every nth entry, but items of high probability, if you have
the additional knowledge. In our example, this may be population
count (leading to Bangladesh, Brazil, China, India, . . . ), number of
internet users in that country, or a distribution of existing customers
etc. Also the component could be made adaptive in that it keeps a
statistic on which items are selected by users, to make more fre-
quent choices appear as early as possible, while maintaining also a
quick access to infrequent cases (by balancing the size of the folds).
User notifications. Sun rays. Ripples. With the continued work-
ing in a partial view of the data space, there is sometimes the need
to notify the users of events that are currently not in their view
space, or locationless in nature. If Spotlight Navigation users send
off messages by dropping a scribbled note over a business card,
they are informed that the message has been sent by a temporary
semitransparent overlay message in the screen center. In contrast
to alert boxes, as they do not need a confirming button press and
due to the semitransparency, it is possible to see through the mes-
sage and continue working, but the message is rather obtrusive,
which is OK for fundamental, infrequent process information or
confirmation of an achieved user goal. It is equally suitable for re-
ally important events, such as energy running out in a few seconds.
For a more unobtrusive notification, the information can be placed
in the periphery (and in line with the trend in traditional desktop
UIs). In Spotlight Navigation also two further ideas were devel-
oped, sun rays and water ripples, to notify the user of off-screen
events (i.e. that also have an origin in the data space). An example
is an alarm for an upcoming event in the user’s calendar, the source
of which can be located at the place of the respective calendar en-
try. Users can be notified in a scalable manner, from unobtrusive
to highly visible, and also the location of the event is communi-
cated. With sun rays, rays are cast from the origin that illuminate
the current view in stripes (making the content under a stripe a little
bit brighter). Through the size of the angle between stripes it can
be directly inferred how far away the source of disturbance is and
in which direction it lies. The notification can be scaled in inten-
sity (how much the stripes are lightening up the content below, or
by a ’glistening’ pattern where neighboring stripes rapidly change
in intensity). The idea that was finally implemented in the Spot-
light Navigation prototypes is ripples. Here, the generating event
literally makes waves that spread over the data space. Similar to
Halos [1], from the diameter of the wave front users can judge the
distance of the disturbance. Of course also the propagating direc-
tion gives a direct cue to the direction, and this is why ripples were
preferred over sun rays. Technically the effect is implemented by
rendering the current view on a polygon mesh that is distorted by
some time dependent sinusoidal functions. Even though that wa-
ter ripples on a wall do not have any real world correspondence,
the visualization is sufficiently convincing, so that the metaphor
works also on non-horizontal surfaces. Similarly to sun rays, the
effect can be scaled from almost unobtrusive to a point where no
reasonable working except tracing the origin of the disturbance is
possible. This is done by simply adjusting the amplitude of the
sinusoidal function and the decay / the number of wave fronts.
5. PROTOTYPES
There exist two hardware prototypes of Spotlight Navigation, man-
ufactured in summer and autumn 2003 respectively. They utilize
a monochrome XGA resolution projection engine developed by
Sony’s optical research group. It is using transmissive LCDs and a
high power LED light source the light of which is collimated using
a light pipe. The 2x2 LED array is passively cooled by means of
a heat pipe. The achievable light output is comparable to current
pico projector devices, about 10 Lumen, at about 5–7W electrical
power. The prototypes are wired prototypes: the handheld projec-
tor is wired to a PC running the software. Also the signal condi-
tioning for the LCD is done with an evaluation board outside the
handheld case. Apart from the projection engine and some controls
(the scroll wheel and two additional buttons taken from a mouse
and also interfaced like that), the mobile case hosts an inertial nav-
igation module (MT9 resp. MT9B from xsens). While the second
prototype is designed for small appearance, the first prototype is
component-based so that individual blocks for projection, orienta-
tion sensing, distance sensing, camera tracking, interaction controls
can be snapped together in different configurations. The software
is implemented in C++ using OpenGL for real time graphics gen-
eration. A standard (for 2003, now outdated) 3 GHz Pentium 4
hyper-threaded PC with Radeon 9800PRO graphics runs the soft-
ware at full 60fps. Individual modules such as the orientation track-
ing, the tremor cancellation and the main visualization component
communicate via a UDP socket. The software allows placements of
images, has a calendar application with almost fully implemented
functionality reading PDA calendar files directly, a clock and a
VNC client through which web browsers or PC-desktops can be
accessed. A rudimentary note pad and contact manager are also im-
plemented to show drag and drop of scribbled messages. All other
features mentioned above were also implemented, only the sun rays
were left out in favour of the ripples, and not all possibilities in the
text selection widget have been implemented, such as the adaptive
behavior, or a sophisticated balancing of the folds. The software
can also simultaneously show an overview of the whole data space
and the currently chosen view, for instance for explanation or de-
bugging purposes, or as a feedback to the user.
6. CONCLUSIONS
Spotlight Navigation is a fascinating and intuitive user interface
that demonstrates that pico projectors can be used for more than
just showing videos, pictures or slide presentations. With Spotlight
Navigation it is efficient to access information from a vast infor-
mation space quickly, and to manipulate data without the bound-
aries of a limited screen space. While legacy point-and-click ap-
plications using standard interface widgets can be integrated into
Spotlight Navigation easily, it is beneficial to rethink applications
ground-up from a user perspective, to use natural representations
or metaphors, a layout taking advantage of panning and zooming,
natural gesture and specifically designed widgets such as those de-
scribed in the paper. This will in general lead to a simpler and
cleaner yet more powerful design. A consequence of holding the
device in the hand is, that there are only limited possibilities to en-
ter textual information. While the device is probably unsuited to
enter extended texts, the proposed scribbling is sufficient for small
annotations or editing tasks as well as composing short messages
or to use it for sketching. The strengths of the device are in explor-
ing and manipulating vast data spaces such as access to structured
information, maps and plans, or large media archives. Currently,
the PC-based prototype is ported to a mobile platform [3].
7. CHRONOLOGY
Spotlight Navigation has been invented on 18th November 2002 by
Dr. Stefan Rapp. It was presented to a group of co-workers, all
user interface professionals at Sony research labs Stuttgart the next
day. After the management has been convinced to support practi-
cal investigation further, a small project team of 4 researchers was
put together, that started implementation work from April 2003. A
first prototype was presented in September 2003 to managers from
across the world inside Sony during a lab evaluation in Stuttgart,
and in December 2003 to hundreds of visitors including the Sony
top management at the Sony technology and exchange fair (STEF)
in Tokyo. The first presentation to the general public was at Perva-
sive 2004 in Vienna.
8. ACKNOWLEDGEMENTS
The author wants to acknowledge the core team members Martin
Barbisch, Ronan Bohan and Zica Valsan for contributing to the
Spotlight Navigation prototype software, Martin Osen for concep-
tual work as well as the interaction and industrial design of the
prototypes, Jason Williams for usability advice, Georg Michelitsch
for conceptual work and discussions, Martin Emele for discussions
and organization, Haruyoshi Suzuki for taking the funding deci-
sion, supporting us at Sony and enabling the spin-out, Manuel Tei-
jido and Frederic Ludley for the smooth collaboration on the opti-
cal engine, and the people at Sony Alsace Factory for helping with
producing the prototypes in time.
9. REFERENCES
[1] P. Baudisch and R. Rosenholtz. Halo: a technique for
visualizing off-screen objects. In CHI ’03: Proceedings of the
SIGCHI conference on Human factors in computing systems,
pages 481–488, New York, NY, USA, 2003. ACM.
[2] T. Masui, M. Minakuchi, G. R. Borden IV, and K. Kashiwagi.
Multiple-view approach for smooth information retrieval. In
ACM Symposium on User Interface Software and Technology,
pages 199–206, 1995.
[3] S. Rapp. Spotlight Navigation and LumEnActive: Ubiquitous
projection user interfaces. In Pervasive 2010 Adjunct
Proceedings (demo sessions), Helsinki, May 2010.
[4] S. Rapp, G. Michelitsch, M. Osen, J. Williams, M. Barbisch,
R. Bohan, Z. Valsan, and M. Emele. Spotlight navigation:
Interaction with a handheld projection device. In Advances in
Pervasive Computing: A collection of Contributions Presented
at PERVASIVE 2004, pages 397–400. Oesterreichische
Computer Gesellschaft, April 2004.
[5] J. Raskin. The Humane Interface: New Directions for
Designing Interactive Systems. Addison-Wesley Professional,
2000.
