ArchInSite: Augmented (Reality) Architecture

Fig. 1 Diagram of ArchInSite operation.
ArchInSite is an augmented reality application that
combines three-dimensional modelling, video compositing
and a global positioning system into a handheld
device. This system allows a user to move through the
landscape, while using a mobile device to view virtual
models accurately placed in space. ArchInSite is unique
in its combination of a hand held device with GPS and
video compositing. This system can be useful for visualizing
architectural designs, sculpture and augmented
environments directly on real site, while simultaneously
interacting with and making design decisions within
the virtual environment. Further, ArchInSite offers the
possibility of creating a virtual world woven into real
space, both as a means of displaying architectural information
and exploring design alternatives within the
existing conditions. At the conceptual level, privileging
the perspectival view challenges the conventional place
of orthographic projections as the media of architectural
design.
Architectural Background
While the vast majority of the architectural community
is concerned with scaled, planometric drawings that
represented spaces from a God’s eye view, we are interested
in exploring a technology that would allow us to
visualize buildings in a real-time, interactive first-person
perspective. Our reasoning is simple enough—if one
experiences the world in a first-person perspective, why
not design buildings in a first-person perspective?  It appears
to us that too often architects are referencing nonexperiential
spatial concepts, such as a geometrical relationship
between adjacent rooms that is only evident in
plan, instead of the direct experiential nature of inhabiting
a building.  We consider first-person perspectives of
a building to be indubitable knowledge for architectural
experience. Likewise, we consider an overall, spatial
plan of a building to be comparable to the philosopher
Edmund Husserl’s description of a belief.   Therefore,
a progression through a building, in which one is able
to take in several first-person perspectives (indubitable
knowledge), allows one to begin making some basic
assumptions (beliefs) about the overall spatial plan of
the building. So, while the experience of a building
progresses from first-person perspectives to an overall,
spatial plan, the standard practice for designing a building
progresses in reverse, from planometric drawing to
perspective. We see this problem as an opportunity to
create a device that would provide the kinds of real-time,
perspectival visualizations necessary to design in a firstperson
perspective.
It is our contention that the analysis of these first-person
perspectives, not as our own perspectives, but as an ideal,
archetypal perspective, should be the driving force of
an architectural design. Like Husserl, we believe that a
further consideration of things in and of themselves can
provide a foundation for more conceptual and abstract
thoughts.
Background/ Related Work
It is possible to trace the concept of computer generated
reality to Ivan Sutherland’s pioneering work with head
mounted displays and his idea of a window into a virtual
world (Sutherland). Much of this early work focused on
the creation of a world separate from our ‘normal’world,
which became known as virtual reality. The term ‘augmented
reality’ was first introduced by two Boeing engineers (Caudell), who created the Wire Bundle Assembly
Project, which combined a heads-up, see-through, headmounted
display that registered computer produced diagrams
superimposed on real world objects using half
silvered mirrors and position tracking. Steven Feiner’s
work on ‘A Touring Machine’ (Feiner) combined bulky
GPS equipment with optical overlay heads-up display
to geo-spatially register and display text information
about campus building interactively. Ongoing work by
Behzadam and Kamat (Behzadam) at the University of
Michigan is using video compositing and GPS data to
feed a head mounted video display for possible use in
construction.
Research Issues and Implementation
Augmenting the sensible world with additional information
is, on it’s face, a simple and reasonable goal. Because
we never just look without intention, the ability to
enrich the visual context with information makes possible
new forms of understanding and interaction.
Augmented reality has raised important ideas about
both hardware and software. The hardware issues have
largely revolved around issues of display and registration
(by contrast, generating 3d models and placing them
on still photographic backgrounds is well understood).
Display choices are largely focused on heads-up display
technologies, with choices between see-through displays
and video compositing. Registration of the model image
with the real world is a difficult and on-going problem,
requiring specialized equipment (head tracking accelerometers)
and limited environments (position tracking).
While these efforts continue to show promise, their current
state is still largely fragile and more suitable to research
labs than to wide spread use.
We have developed a prototype for a system that incorporates
three-dimensional modelling, GPS tracking, an
internal camera and video compositing to produce a device
that creates a hybrid condition between the virtual
and real worlds. The current implementation consists of
a Sony UX micro pc running Maya modelling software,
a GPS data parser and video capture software. The system
utilizes a MEL script that enables the built in camera
and the data streamed from the GPS to reorient the model
in virtual space based upon the position of the user. A
system such as this is feasible for virtually all camera/
GPS enabled phones (currently estimated at 400 million
units/year worldwide); this makes their use potentially
ubiquitous.
Results
We have successfully tested the use of ArchInSite for
viewing of proposed architectural models in the landscape.
The system is readily understood and embraced
by users, and while the registration problems endemic
with head mounted systems do not completely disappear,
separating the view port from direct attachment
to users (by substituting hand held display for heads-up
display) provides much more flexibility for the users to
comprehend the registration of the virtual and the real.
We began with preselected models that were placed at
specific points in the landscape; we then implemented
(using Maya’s default modelling tool set) the ability to
make changes to the model while concurrently viewing
virtual architecture integrated into its site.
Fig. 3 Screenshots showing change of user position
and corresponding change in model and composited
background.
We have three major findings.
1. The system that we have implemented utilizes easily obtainable handheld devices rather than headsup
displays, and side steps the difficulty of solving
registration problems. Our users have found this
handheld device approach to be a convincing window
that they can connect directly and viscerally to
the context.
2. We believe that ArchInSite can prove valuable as a
design tool for architects and designers. The ability
to see work with a model on site is useful because it
gives you a richer and more fulsome understanding
of the context than is possible when pursued through
a distance by sketches, renderings and/or drawings.
Coupling this visual understanding with the ability
to alter the computerized model allows the designer
to sensitively recalibrate and redesign the model in
the actual setting.
3. We think that significant opportunity exists to construe
a virtual world that is threaded inside the real
world and supplements it in interesting ways. Architecture
was often used as a classical rhetorical
device for memorizing facts and telling stories and
we see the potential for its re-emergence threaded
through the real world.
Further Work
We are currently developing a new modelling and compositing
implementation that will use custom software
rather than off-the-shelf applications that often have high
levels of system overhead. Our further goal is to move
this application, or a variation of it, over to a more compact
and ubiquitous device, such as the iPhone, a device
that currently contains the hardware necessary to drive
the software.
We will also be teaching an architectural studio course
that will use ArchInSite as the primary means of design,
seeking to challenge the hegemony of the orthographic.
References
Behzadan A.H., and Kamat V.R. (2005), Visualization
of Construction Graphics in Outdoor Augmented
Reality, Proceedings of the 2005 Winter Simulation
Conference, Orlando, FL
Caudell, Thomas P., and Mizell, David W. (1992). Augmented
Reality: An Application of heads-Up Display
Technology to Manual Manufacturing Processes, 1968
Fall Joint Computer Conference, AFIPS Conference
Proceeding 33, 757-764 (1992).
S. Feiner et al., “A Touring Machine: Prototyping 3D
Mobile Augmented Reality Systems for Exploring the
Urban Environment,” Proc. 1st Int”l Symp. Wearable
Computers (ISWC “97), IEEE CS Press, Los
Alamitos, Calif., 1997, pp. 74-81.
Sutherland, Ivan E. (1968). A Head Mounted Three Dimensional
Display, 1968 Fall Joint Computer Conference,
AFIPS Conference Proceeding 33, 757-764 (1968).