Filter Mediated Design: Generating Coherence in Collaborative Design
Filter Mediated Design (FMD) is an electronic mediating device that allows users engaged in collaborative architectural design to interact with computational agents to generate coherence out of local, domain specific contributions. Agents and users adopt different strategies for producing solutions and interacting with each other, facilitating the creative process. FDM proposes a method for negotiating architectural design across domains, by examining issues of ontology, perception, generation and evaluation. A prototype has been built in which these mechanisms are embodied using computational agents for achieving coherence in a "toy 3D building game" Â in remote collaborative design. The prototype is implemented on top of ITA's "Open Community", a platform for sharing data over the internet. FMD provides an "object to think with" Â to explore the dynamics and strategies at play in constructing intelligent designs and design intelligence.
Background & Objective: Architectural design involves the integration of diverse, sometimes conflicting, concepts and requirements into a coherent single composition. The emergence of Distribute Virtual Environments (DVEs), such as ITA's Open Community, provide an opportunity to rethink the ways in which teams have traditionally collaborated, and to offer the mediations needed to augment group negotiation, and foster collective intelligence. Researchers from the fields of Architecture, Artificial Intelligence, and Cognitive Psychology are investigating how DVEs can enhance group design, and how the requirements of collaborative design may help shape the requirements of DVEs. Recent developments in artificial life and human learning suggest that the process of generating coherence, or coordinating multiple perspectives, is decentralized, dialogic in nature, and analogous within and across individuals.
Technical Discussion: In both collaborative learning and design, coherence is achieved through iterative cycles of productions and evaluations. At each step, generations and tests are reconsidered in light of each other, and gauged according to some fitness functions. In ill-defined, loosely constrained tasks such as the ones studied here, the search space itself evolves as a function of the form under consideration. The task is to find a representation of the data to be negotiated, and means for accessing and manipulating this shared data, in ways that are sensitive to the dynamics of design. The project has a common geometric and topological database, from which multiple semantic models are constructed by Filters. These semantic models are then negotiated and changes are proposed through the central, semantically sparse database. Coherence emerges when multiple semantic models find satisfaction in the artifact. The project also contributes to the research on Computer Human Interaction. The mechanism supports the distribution of the cognitive load, allowing the computer and human users to focus on particular aspects of the design process. One scenario allows users to freely generate, relegating to the system the task of keeping track of the constraints. In another scenario, users may let the system generate alternatives, exploiting the human ability to appreciate as a lever to sustain the creative process.
Technology Areas:
Net Services
Artificial Intelligence
Modification Date: September 12, 2007