Information, Communication and Collaboration System for the Swiss AEC Industry
Objectives Results Cooperation Publications PerspectivesWe are developing a prototype interface for an information architecture in order to support building processes such as planning, construction, facility management and life-cycle design and management. Through using the World-Wide Web, a database provides for secure collaboration and information management. We have extended the spatial modeling application Sculptor to improve design collaboration on a project developed simultaneously on several workstations.
We are also concentrating on research in areas of information exchange and design decision support through constraint solving. Fundamental research results in the area of constraint solving have been applied to improve collaboration through avoiding conflicts and futile negotiation. This resulted in a framework for collaborative application of consistency techniques to constraints on continuous variables. Our industry partner, Zwahlen & Mayr has provided examples for full-scale testing in an industrial context.
A database provides for secure collaboration and information management within the WWW environment. Research on internet database agents aims to realize a database that 'reacts' through initiative, assistance and actively taking over tasks. The following examples of a document management system illustrate this cooperative character:· Documents are selected and retrieved with relation to the user, without the need for sophisticated queries. This will improve the user's familiarity with the system and reduce the need to make choices.
· Relationships between documents, such as versions and references, are presented as navigable links to the user.
· The user's access rights are presented for every document in a way that allows the user to evaluate at a glance which procedures are available.
With regard to implementation, we have adopted a 3-tier approach with a relational database on the server side and an object-oriented data model in the middleware.
In an ongoing development, we have explored two interfaces for supporting information management using a database. Each interface reflected on the latest technologies available at that moment. We tested these interfaces as the information and communication platform for our project partners. Their feedback was very important to us. The latest version presents a unified interface to manage all data types, e.g., images, documents, messages and URL's. All data entities are organized into a 3-dimensional structure of area, time and type. Further categorization is supported through the specification of labels. The user has the ability to specify readers and writers for collaborative documents, with upload, download and lock functionalities. Data entities can be collected into groups, messages can be attached to other entities, and other links can freely be defined between entities. Using 'push' technology, the interface is automatically updated following any alteration in the database. The interface's functionality can be dynamically extended through the inclusion of agents. We believe that this latest version not only addresses most of the issues and requirements brought up, but also provides us with a uniquely appropriate environment to further this project into the development of a tool set for the virtual AEC company. The development borrowed from other research and teaching projects (e.g., ZIP Bau and Phase(x)) and from our own research into delegates.
This prototype is an example of electronic commerce, providing access to a subset of the SIA standards.
We define a delegate as a process that has a particular function within the environment and is only concerned with this function. Delegates are organized into a network. We distinguish four important types of functionalities in order to support communication and information exchange effectively. These are the specification of (information) requests, the visualization of the corresponding results, the internal communication, and facilitating information exchange. We consider four types of delegates to support these functionalities: representatives, presenters, mediators and facilitators. Together, these specify the functional interface to the user and support communication and information exchange between users and delegates. Delegates serve to compose and modularize the system. They also enable to extend the system without affecting the other components.
Sorts serve to improve communication and collaboration, currently hampered by the existence of different data formats and representations. Sorts are defined as abstractions of representations and as general forms for exchanging data. Sorts closely relate to product modeling approaches. Product models seek to integrate all aspects of an object in an attempt to provide a unique and consistent model to all partners during the life-cycle of the object. Sorts, on the other hand, take a minimal approach by focusing on the components and compositional relationships that constitute a representation. Together, these define a descriptional language that allows us to relate alternative representations and support translation between different representations.
We have developed an exemplar implementation of sorts for visualizing database information in a user-centered manner. Whereas the information is stored in the database in tables, simply by category, the visualization places the core design information at the heart of the presentational structure.
A construction monitoring web site exists for the new chemical building of the ETH Zürich (Synergy with the project "Zeitrafferaufnahmen" headed by E. van der Mark at the Chair for Architecture and CAAD). Single photographs and continuous videos with time intervals of 30 minutes and 2 minutes for special actions (e.g., demolition) can be recalled.
SpaceSolver, the constraint-satisfaction framework developed within the ICCS project, was further developed . Many improvements made it stable enough to offer public accessiblity. It was also extended by user-management facilities which allow password-protected access and collaboration on CSPs. Collaborating users define constraints on possibly shared variables. All constraints are then collected into one CSP which is transmitted to a centralized solver. The connection to the project database allows collaborators to add comments and explanations in a free format to any variable and any constraint. Visualization of constraints and solution-spaces is now performed in VRML2.0. The entire framework is implemented based on the Internet as a collection of HTML-pages, CGI-scripts and a server application which calls several C++-Modules.The symbolic representation of the CSPs as it is given by the users are now transformed by SpaceSolver. Constants and certain unnecessary auxiliary variables are removed before constraints are transformed into a spatial representation. This is a critical step because the complexity of the consistency algorithm is O(n^5) where n is the number of variables. The consistency algorithms inserted into this framework now also include a improvement of arc-consistency as defined by Faltings and Gelle in 1997, path-consistency and 3,2-relational consistency. All these algorithms are implemented with significantly improved cost in space (up to a factor of 20) and time (up to a factor of 10).
In order to further validate the suggested framework, a second more compact industrial example was analyzed. This allowed testing of higher degrees of consistency on a real-world situation. It was confirmed that lower degrees of consistency are useful to detect conflicts in large examples. Even the 50-parameter examples can be analyzed within a few seconds using arc-consistency. On the other hand 3,2-relational consistency can reveal hidden dependencies between variables which a priori are not related. This type of consistency is also able to support decisions related to trade-offs when several objectives are pursued. However, the performance of 3,2-relational consistency is limited to interelated requirements that involve less than 20 variables.
We implemented, evaluated and adopted a collaborative design prototype in a close-to-practice environment. The spatial modeling program Sculptor was extended to enable simultaneous collaborative modeling in 3D on a single model, in a distributed environment. Additionally, an Internet based collaboration site enables the monitoring of ongoing collaborative sessions. This globally usable service manages all Sculptor sessions and allows everybody to view ongoing sessions, look at all current models and start a new or join an existing modeling session.This collaboration aspect, together with other features developed within this project, made it possible to use Sculptor in almost all courses at the CAAD chair, from the first year introductory courses up to the elective courses in the final years. Sculptor especially demonstrated its strenghts in collaborative design during a one-week seminar titled 'Multiplying time' held simultaneously at universities in Zurich, Hong Kong, Seattle and Vancouver in November 1997. Both the process and the results proved to be very exciting. Novel approaches were used for accessing an Internet based database in connection with Sculptor for 3D design. It provided an excellent test bed for the tools developed at the CAAD chair. Students working around the clock on a single design showed both the possibilities and potential pitfalls of our collaborative design approaches.
Research partners
Industrial partners
- Zwahlen & Mayr SA, Aigle (A. Flückiger)
- Zoelly Rüegger Holenstein Architekten AGG, Zurich (W. Rüegger)
- Amstein + Walthert, Zurich (R. Walthert, T. Baumann)
- Allemand Jeanneret Schmid SA, Neuchatel (J. Jeanneret)
- ABB Installationen AG, Zurich (R. Herlig)
Contributing partners
- Netconsult AG, Bern (B. Buess, E. Peter)
- Rosenthaler + Partner AG, Zurich (C. Rosenthaler, H. Notter, C. Gehr)
Outside of the project
- Dept. of Architecture, University of Hong Kong (B. Kolarevic)
- Faculty of Architecture, TU Delft (S. Sariyildiz)
Publications
Applications
- new methodologies, algorithms
· constraint solving
· sorts: a language for representational flexibilityZwahlen & Mayr SA working
in development
- prototypes, demonstrators
· e-commerce prototype
· information architecture prototypeSIA working
in development
- lectures
- parts of lectures
A first focus will be on the information architecture, that is, the representation, presentation and visualization of information in a networked environment, in particular the world-wide web, in order to support building processes such as planning, construction, facility management and life-cycle design and management. Building on the teaching experiences of our research team we will develop ways of visualizing information that can guide the user to zones or nodes of interest, highlight problems or issues that need consideration, or determine activity centers.
A second focus will be on the continued development of constraint solving mechanisms. Two aspects will be further developed. On one hand the distributed computation of solution spaces will be investigated. Until now, information was collected and constraints were solved centrally. In the follow-up, we target a distributed constraint solving, although still aiming for global consistency. On the other hand support to search for solutions will be provided. Working with solution spaces, as we suggest it up to now, is an advantage in preliminary phases by eventually a single solution has to be determined and used for construction. This late phase of design will be a second goal in the next step of the project.
A third focus will be on sorts as a language for supporting information exchange between different representations. So far, we have used sorts primarily as a means of visualizing data base information in a user-centered manner. Subsequently, we intend to develop practical examples to study the exchange of information between different application formats.