At the same time as a changing society pushes architecture by posing a new set of characteristic problems, the new digital design media pull architecture into an uncharted territory of opportunity. The key question here is whether the exploration of the new creative opportunities can be directed towards offering effective architectural resources that can help to answer the problems posed today.
The AA Design Research Lab  is trying to address this question. We set ourselves the task to contribute to the retooling of the discipline of architecture in the face of what we perceive to be two fundamental conditions:
The challenge presented by the new level of dynamic complexity in the metropolitan life processes and the opportunity presented by the development of new powerful digital design tools.
Therefore we start from two rather general premises as points of departure of our work:
1. The task to develop complex spatial systems that could serve to organize and articulate the new social complexities.
2. The task to utilize and adapt various advanced digital methods of designing, modelling and simulation.
Innovation is always suspended between two poles: the investigation of a domain of problems and the expansion of the domain of potential solutions. Within the discipline of architecture this polarity of innovation has often been an occasion for a productive division of labour between the analysis of new societal/programmatic demands on the one side and the proliferation of new spatial repertoires on the other side. Embodied by the Dutch avant-garde and the US avant-garde respectively, both aspects have been pursuit independent from each other. The independent elaboration of the two domains makes sense, as a division of labour or specialization. However, this led to two opposing ideologies, programme versus form, equally one-sided, posing the question of synthesis. The synthesis requires the oscillation between the two domains and is itself an act of creative intelligence. There are no one-to-one correspondences between "problems" and "solutions". Solutions can go in search of problems as well as problems in search of solutions. What we call design research is the attempt to systematise this oscillation within a well circumscribed frame that narrows down both the realm of problems and the realm of solutions.
Which social arena would best allow us to explore the architectural opportunities afforded by the new tools and the new forms?
Spatializing Corporate Organisation
The initial mission statement of the AADRL promised such an innovative synthesis. The task was to give a socio-economic/programmatic interpretation to an intense wave of formal proliferation that has been advancing for a considerable period without any specific problematic that could offer friction and resistance to select, shape and substantiate the work.
The following strategic problem domain was identified by the DRL: The paradigm shift within the theory and practise of business organisation, from clear-cut corporate hierarchies toward open, self-organising networks of collaboration.
For four years our design research focussed on "Corporate Fields", systematically instrumentalising a selected set of recent formal/conceptual repertoires (primarily single surfaces and swarms – augmented with kinetic capacities!) for the spatial organisation and articulation of an equally well-constraint set of recent concepts, tendencies and practises within the emerging post-fordist enterprise culture.
The premise of this choice of agenda was the general compatibility of concepts of social organisation with concepts of spatial organisation, and the specific convergence of terms that could be observed when comparing recent architectural discourse with recent managerial discourse. Principles of spatial organisation like the super-imposition of multiple spatial reference systems, or the concept of a continuously differentiated field, are well suited to articulate the new corporate organigrammes like the matrix organisation, or the idea of open networks, with shifting/gradient centres of authority and blurred lines of responsibility.
The managerial problematic of self-organisation, together with the insatiable need for flexibility and permanent reconfiguration, encountered in the new business culture, inspired the as yet uncharted utilisation of animation software for the design of kinetic, self-organising environments.
Our current research programme of "Responsive Environments" was abstracted from the latest instantiations of "Corporate Fields" where the dynamic of the work-flow and team reconfigurations was reflected in the scripted responses of a kinetically adaptive office-scape. The work was re-focussed to foreground this new capacity to design spaces that actively engage with their users. Here the design task goes beyond the delineation of form and demands the creation of complex behavioral systems. This in turn requires research.
The agenda of responsive environments opens up a whole new domain of design research. It announces a paradigm shift from the design of inert spatial form to the design of systems of behaviour: the design of spatial systems that are capable of interaction by means of real time reconfiguration in response to users via embedded electronic intelligence.
This research programme is founded upon two technological presuppositions:
1. Various sensor as well as actuator technologies become readily available. At the same time processing power is becoming ever more ubiquitous and cheap.
2. The tools to design and simulate responsive systems are readily available in the form of animation software like 3ds max and Maya.
These software packages offer modelling and simulation tools that allow the designer to set up complex systems of dynamic interaction. Any parameter of any object might be dynamically correlated with any parameter of any other object within the model. This means that the designer might craft an artificial “universe”, with its own peculiar “ontology” and “laws of nature”.
The designed environments are augmented with electronic “sensitivity” and “responsiveness”, and even “spontaneity”, engendering artificial life-processes that would be symbiotic with human life-processes. In order to effectively speculate about these symbiotic life-processes the dynamic modelling always includes the modelling of the movement/behavioral patterns of the human actors. “Agents” – both human and non-human - are programmed and the designer observes the unfolding life-processes with the view to calibrate the agents’ profiles.
We are trying to emphasise the possible social dimension of responsive environments, focussing on the capacity to actively stimulate and facilitate new forms of social communication. The task of responsive design does not have to be restricted to the efficient processing of movement. Although the engineering of such efficiencies is no mean endeavor, the full power of the new paradigm of responsive architecture can only be brought to fruition if the brief challenges architecture in its capacity to construct and choreograph communicative situations - a longstanding ambition of architecture.
The ability to elaborate and animate narrative scenarios - “a typical day in the life of …” - is supported by 4D simulation capabilities that include people animators. Finally it becomes possible to devise and work on complex social scenarios and to speculate in concrete detail about prospective behavioural patterns unfolding within (and in response to) the proposed spatial configurations. The animation is there as a handy artefact to be crafted and refined element by element, move by move, scene by scene. This new technique of design speculation affords a leap in our ability to innovate beyond the mere adaptation or reconfiguration of familiar types. Only on the basis of such elaborately crafted 4D scenario-productions can a radically unfamiliar environment be made socially plausible.
Within the Responsive Environments agenda we have explored a whole series of programmatic briefs set within various social contexts: corporate headquarters, residential complexes, retail malls, arts & entertainment centres, airports. These briefs furnish the material to construct the social scenarios. But what could be the source of inspiration for the forms and behaviours of our architectural agents?
In order to expand our repertoire of geometries, structures, and kinetic systems the design-work is preceded by a bio-mimetic research effort. We are foraging into the world of organic life just like robotic researchers have done for many years. Obviously robotics itself is our most immediate inspiration. But we are also going directly to the source: We are exploring various organic systems - individual organisms, collective organisms, subsystems of organisms - as source domains for the analogical transference of principles into the domain of responsive architecture. From the various biological models we are extracting geometric systems, envelope systems, structural systems, kinetic systems, sensory systems, systems of aggregation, and systems of communication. Each of these systems might be taken from one and the same organism. The transference of the models and principles leads to the construction of new architectural systems that eventually function in the context of the project scenario. It is important to note that this analogical transference from organic life into architectural artefact is mediated by the science of biology. Thus we are furnished with concepts, classifications, and measures that can be transferred alongside the organic model. The science of biology has a rich and nuanced conceptual apparatus for the description of complex morphologies in their relation to functional capacities. Distinctions like homology vs analogy, parasite vs symbiosis, genotype vs phenotype, and autopoeisis vs allpoeisis might contribute to the emerging conceptual framework of a responsive architecture. Architecture has as much to learn from biology’s conceptual sophistication as from the organic world itself.
To exemplify our work I would like to introduce one of our most recent projects. The project is based on a brief calling for the redesign of Heathrow Terminal 4 as a Responsive Environment.
Project : Heathrow.comm
Michele Pasca di Magliano, Cynthia Morales, Viviana Musceottola, Nick Puckett
The project set itself the task of catalysing communicative clusters within the domain of the airport. Various biological models (as well as their scientific elaboration and computer simulation) were studied: Slime-mold, ant colonies, and flocking of various species of birds.
These models led to the ambition to achieve the desired formation of communities by allowing spatial arrangements to emerge from the responsive interaction between programmed agents (architectural components) instead of delineating the supposed community spaces in advance.
Heathrow.comm constructs an electronically augmented fitness landscape catalysing the competitive formation of casual communities. It seeks to facilitate a new level of social communication within the airport by amplifying its existing social dynamic through the wireless connection of its inhabitants and architectural systems, thus merging the potentials of the airport’s diverse population with the methodologies and interests of today’s internet communities. The responsive architectural systems (interactive walls and ceiling panels) facilitate the clumping of the crowd in patterns defined by its dispersed movement and exchange of information. The project draws on the latent shared interests that are hidden within the anonymous crowd. Various virtual attractors compete for pulling people together: e.g. info-exchange between people heading for the same destination versus speed-dating or game-playing. The spatial field is registering and amplifying insipient cluster-formation by means of (more or less) subtle territorilizations.
In it’s ultimate form this is achieved by organizing the field as a composite lattice of cellular spaces that function to both register and prompt the users. Simulations were scripted with Alias Maya and BioGraphic Tech’s AI implant plugin.
Electronically enhanced space populated by agents
The first version of Heatrow.comm existed a scripted network of 3 initially independent systems which had different types of speed and response. The ceiling system represented the first level of response as it was the most dynamic and rapidly changing system. The rotating panels served to inscribe an orienting organizational pattern into the space that extrapolated a distinct order from the insipient formation of clusters. The interface walls responded to the level of activity of the community users and expanded to provide an interface and more definite boundaries as the community became more established. The floor system responded much slower to the inputs and created paths of the general crowd flow over time.
The simulation generated by the script operates exclusively by reading the individual pieces of information stored on each A.I. agent. Each community unit (1interface, 1 6mX6m ceiling unit) has the ability to function as an autonomous cell or as a larger cluster (4 interfaces, 4 6mX6m ceiling units). This larger cell of 4 ceiling/interface units corresponds to 1 floor unit.
The first level of system response is determined by the information gathered by each interface. The sensor in each interface processes the participation level of each person within the 5m radius to determine the community’s activity level. The activity level controls the rate of growth for the expanding interface wall. When the interface wall grows large enough that it stops rotating, the community has its birthday. Also at this point the interface orients the ceiling panel to correspond with the rotation of the interface. The information processed by each interface within each 4 unit cluster is then passed to a region infoPoint which displays the total number of people, the activity level, and the age of the cluster/community. These totals inform the response of the floor unit.
The augmented environment creates a new hybrid nature leading to the disappearance of traditional space-time relations.
There is a continuous exchange between the architectural agent’s rules of response and the character’s rules of navigation and interaction. This move from drawing to scripting allows the digital model to move from the arena of representation into the arena of a working prototype.
The final version of the simulation engine exists as a complex network of multiple scripts, referenced files, and a database of information all linked by a control interface. The flight data from Terminal 4 are downloaded from the BAA.com website and stored in an Access database. This information loads the appropriate number of passengers for the flight and tells the AI characters when to get to the airport, the gate number, and time of departure. For each flight you can also customize what percentage of passengers are members of a chat-forum, or game community. This information is imprinted on the characters to determine the way in which they influence their emerging community space.
In terms of geometry and material the research started with simple folded surfaces
and evolved into a highly complex 3d lattice. The aim was to construct a space-filling medium that could expand and contract. In the final version of the design both walls and ceilings were conceived in this way to allow for the responsive sculpting of the space.
The term auxetic refers to any material which has an inverse poisson ratio. Most all natural
Materials, when stretched in one direction, will shrink along the perpendicular axis. Auxetic materials on the other hand will enlarge on the perpendicular axis. Up until now this class of materials has not been used as a dynamic material, but rather explored for it’s acoustic and insulation benefits. For our studies we began by simply constructing a model of the molecular structure of these materials at an observable scale. These models were build up from simple base modules that captured the capacity of bi-axial expansion. Each module can expand to 1.5 the normal scale. Those modules were then arrayed into lattices where the deformation of each component induces a gradual deformation in the adjacent components, which in turn generates an overall effect in the field.
The first operable prototype was created as a mesh of 48 nodes (4X6X2). The second version consisted of 64 nodes (4X4X4) in 4 layers. It was constructed of laser cut polypropolene and polystyrene and used 16 dc motors to operate the different nodes.
Small non-geared dc motors were used to operate the delamination action of the layers
The user interaction with the prototype is registered by 4 sensors which then feed into a
controller pc via an IcubeX digitizer. The values from the sensors are processed by a custom
script created in Max/Msp which interprets these values and then sends the appropriate outputs to the microController. The output from max (which occurs every 20 milliseconds) instructs each motor whether it should be ON or OFF, and whether it should be moving UP or DOWN. The direction of the movement is determined by the proximity of the user to the sensor and is achieved by sending signal to a bank of relays on the control board which opens/closes the appropriate circuit.
Digital kinematic model
After observing the behaviour of the 3d physical models and defining it more specifically with the dynamic testing, the auxetic behaviour could be distilled as a geometric equation which related the rotation of each individual hinged strip to the expansion of the whole module. This logic was then implemented through kinematics to provide an operative digital version of the observed behaviour. One limitation of the physical models was the inability to apply very specific forces to the material and the ability to quickly create variations of the geometry itself. In a first series of studies Maya rigid body dynamics were used to investigate these areas. To construct these simulations the same geometry is built from individual pieces which are given a specific mass. These are then joined at the corners by hinges which can also be assigned certain limits and resistances. After creating the structure, forces are applied, and the series of physical tests is extrapolated by many more digital variants.
The next phase for the digital model was to both derive a way for variable input and to automate the production of models of different sizes. To do this each unit’s behaviour had to be understood in terms of its neighbours. Due to the basic nature of the geometry, when one node is compressed it must effect the rotation and the translation of its neighbours. The difficulty lies in the fact that this influence can go beyond a unit’s 4 immediate neighbours, depending on the specific situation. After an exhaustive trigonometric and conditional study a single script was developed that could create an auxetic mesh of any size which retained the properties of the material and the ability for non-regular input.
The first step was to use the Maya dynamics engine to create the geometry and replicate the behaviour of the physical model. This process yielded very accurate results, but do to the current limits of processing power, could not be utilized at a larger, airport scale. Thus,
the digital models were analysed to extract the basic behavioural qualities. These qualities
were coupled with a new knowledge base of MEL scripting which allowed it to be converted to a geometric expression. This new knowledge base also gave the ability to create a new form of input to the behaviour through sound. This was created by linking Maya to an interface in Flash which was calculating the pitch of the voice on a microphone in real-time. This new input was able to be joined with the auxetic behaviour of the walls to test the possible outcomes at a large scale. This voice activation represents a real possibility with respect to making the architectural systems responsive (positive feedback) to the level of communication density in contradistinction to the mere density of bodies.
Our methods of digital production are evolving away from the traditional mode of producing digital models. We are moving from drawing to scripting. This of course was in direct relationship to the evolving requirements of the research agenda. Scripted production takes a level of abstraction which produces a machine to make the result rather than the result itself, thus affording the ability to exponential increase output for our experimental series.