IIMS 94 contents
[ IIMS 94 contents ]

Interactive multimedia based intelligent systems in architecture

Hari Murthy and Linley Lutton
The University of Western Australia
Interactive multimedia and artificial intelligence are emerging computer based technologies which have been used in a number of disciplines independently. This paper brings together the developments taking place in these areas to assist the designer in making decisions in the conceptual architectural design process. A prototypical example is presented to highlight the features of this approach.


Introduction

Interactive multimedia is an emerging computer technology that combines text, images, computer aided design models, animations, audio, and video information for the development of a document. It has been used in a number of disciplines for education, training, public information, etc. Begoray (1990) provides an introduction and identifies some areas of application for multimedia technology. This technology provides a means to visualise, read, and hear, in combination, about any subject matter that is under consideration using a computer workstation.

New approaches to computing based on the developments in knowledge based systems, a sub-field of artificial intelligence, are beginning to have a significant impact on the use of computers in a number of disciplines. In the domain of architecture their applications have been discussed by Coyne et al (1990), and Murthy and Lutton (1993).

The rest of this paper describes issues involved in integrating these two technologies for architectural design with the help of an example.

Architectural design

The architectural design of buildings is a complex task which brings together a large number of disciplines and integrates these to the achievement of a single task. It is concerned with product and process. The product (usually a building as yet unbuilt) has particular requirements and constraints that necessarily, are also reflected in the process. The process includes the strategies and procedures used by the designer in preparing the design, and it progresses through a number of stages including conceptual

design, intermediate design and detailed design. The conceptual design stage is perhaps the most critical. At this stage design information is often incomplete, contradictory or inconsistent. The designer is also confronted by what seems at first to be a universe of possible design solutions. Decisions made at this stage will determine the framework for the final design solution. The overall objective of the designer, at the conceptual design stage, is to derive a buildable, integrated design that optimises the solution within the constraints and resources constituted by:

  1. the physical context (eg, location, specific site);
  2. the function (eg, the activities to be accommodated within and around the building);
  3. the legal context;
  4. any specific requirements of the owner;
  5. any requirements relevant to the users;
  6. the resources available (eg, for structure, materials, finance, etc);
  7. any constraints imposed upon the design such as, specific concerns about appearance, relationship of building to site including site coverage and so on.
The final design, which is developed from the design concept, should reflect the satisfaction of and between the many constraints and limitations imposed on the design.

Design involves deciding between alternative courses of action. Not all decisions are critical and the designer needs to ascertain which ones are critical according to the state of development of the design, ie, concept, intermediate, detail. Because design solutions unfold in stages, decisions are often made tentatively and are subsequently reassessed in light of further developments.

While there is usually some initial "hopping" between concept and detail levels as architects generate and test ideas in the process of making design decisions, the initial aim is to establish a conceptual outline or framework that remains fixed for the duration of the design. Further, the activities involved in design are often relatively routine and common to most buildings of the same type. Routine activities include overall siting and planning of the building; selecting suitable structural and construction systems; determining the nature and extent of electrical and mechanical servicing; determining overall costs and time constraints, etc.

Recent design tradition seems to treat each building design problem de novo, as though it had never occurred before. In fact the majority of buildings that are built are answers to problems and needs that are not significantly different from the problems and needs that both contemporary and earlier buildings have already solved and served. There are some design problems that are entirely new, that need to be solved in creative or inspirational ways but design of the bulk of building types, while still benefiting from creative thought, benefits most from accumulated knowledge and experience (Manning 1984).

Architectural designers recognise similarities between design solutions and use "design precedent" as a powerful design strategy. They often draw upon the whole or parts of previous designs, of their own or others, in the search for solutions to new design problems. In-house information incorporating previous drawings and specifications has been observed to be an important source of information drawn upon during design in architectural practices (Mackinder & Marvin, 1982).

Developments in knowledge based expert systems and multimedia technologies provide powerful means to acquire and incorporate knowledge from various sources for assisting in design decision making. These are discussed in the following sections.

Intelligent systems

Knowledge Based Expert Systems are a type of intelligent system that have the ability to perform human-like reasoning. KBES are interactive computer programs that incorporate the knowledge of experts and process it to provide recommendations to the user of the system. Though considerable intelligence goes into the design of conventional computer programs for various tasks, they differ from KBES in the following manner:
  1. the user of an expert system can ask why it needs particular facts and it can justify its conclusions;

  2. KBES can be developed over a period of time in an incremental manner without changing their structure;

  3. the emphasis in traditional CAD has been on quantitative aspects rather qualitative aspects of design; and

  4. in conventional programs the domain specific knowledge is interwoven within the program and is not able to be separated.
Conventional programs often tend to be complex and difficult for anyone other than their developers to understand and modify. A designer with considerable experience in (say) hospital design will have a great deal of difficulty incorporating his/her expertise in a traditional computer program.

KBES have a number of components. The basic ones are: a knowledge base, where the domain specific knowledge is stored in the form of facts and heuristics; an inference mechanism which controls the reasoning and search operations of the system. In addition to these components a KBES may also include: explanation facilities; knowledge acquisition facilities; and a natural language interface. Figure 1 shows the relationships that exists among the essential components of a KBES. A detailed discussion concerning KBES for design is presented by Coyne et al (1990).

Figure 1: Essential components of a KBES

Figure 1: Essential components of a KBES

KBES have been developed using special programming languages such as PROLOG and LISP, and by using expert system shells or programming environments. Conventional programming languages such as PASCAL and FORTRAN are usually better suited to perform algorithmic computations. Users of programming languages intending to develop expert systems have to start from scratch but the use of commercially available expert systems shells can be more convenient as they already have built in expert system components. For this research use of HyperCard, a software development tool made available by Apple Computer Inc with the purchase of its computers, has been used.

Multimedia based intelligent systems model

The knowledge and information required for architectural design comes from a number of different sources which influence the process at some stage. Some of these sources include monographs, journals, codes of practices, site investigations, and contributions of well known designers and experts in the field.

They could be in different formats such as text, photographs, videos, tape recordings and scanned images. Integration and presentation of information of such variety in a meaningful manner is a major challenge. Thus the development of interactive multimedia based intelligent systems requires a great deal of knowledge in putting together the access structure and the information relevant to the subject matter under consideration. Figure 2 gives an overview of the interaction that can take place between the designer and a KBES incorporating multimedia features.

Figure 2: A KBES for architectural design with multimedia facilities

Figure 2: A KBES for architectural design with multimedia facilities

Here the designer can access multimedia features under two situations. In the first instance the designer may find that he/she is unable to provide sufficient information for the KBES to propose design alternatives. In such a situation the designer is provided with additional information about by-laws, building products, performance requirements, physical context, design objectives, and so on. This information may take the form of text, diagrams, photographs (colour or black and white), voice and video clips. In the second instance information to assist the designer to choose between alternative design solutions is offered by the KBES. This would include demonstrations of how each solution has been used by other designers. This information may again take the form of video clips, images, voice commentary on the strengths and limitations to be considered and so on. We feel that providing this type of information at these two stages in the use of a KBES will greatly assist the designer to make more informed decisions more expediently.

Implementation details

The intelligent systems model has been implemented on an Apple Macintosh computer using HyperCard. This software was used due to its simplicity in programming and the facilities available to incorporate multimedia features. It can be described by the analogy of cards grouped into stacks for organising the storage, retrieval and association of both graphical and textual data. The basic elements in HyperCard are cards and stacks (of cards). Stacks can be partitioned into groups of cards that are called backgrounds. A HyperCard stack may be thought of as a 'book': the backgrounds represent the 'chapters' and cards represent the 'pages'. The surface of cards can be subdivided into fields each of which may be needed for storing particular kinds of data. Messages can be sent by the user, by simply clicking on words or 'buttons', to instruct HyperCard to undertake activities such as searching for, collecting and manipulating data stored in these fields. HyperCard's programming language is HyperTalk, a simple computer language whose lines of code can be written in an easy to read sentence like manner. By making use of HyperCard an expert system shell HYPEREX was developed. It consists of the expert system components as identified above, multimedia features, and the access structure.

Example

To illustrate the approach, we have taken a design problem that is common to buildings of all types ie, the proper positioning of a building on the site. In the design of large industrial production buildings, for example, building area is usually considerably less than site area and the number of possible positioning options are considerable, as shown in Figure 3.

Figure 3: Nine possibilities for positioning an industrial building on a large site

Figure 3: Nine possibilities for positioning an industrial building on a large site

Even an inexperienced designer could identify the possible positions shown in Figure 3, but it takes the heuristic knowledge gained from experience to be able to confidently, logically and expediently exclude some of the possibilities.

Drury (1981) identifies some of the critical issues that an expert may consider when determining the positioning of industrial buildings. These include the following:

  1. Is there a requirement for future expansion and if so in what direction is the expansion most likely to occur and how much area will be required?

  2. Is there any aspect of the site topography that would logically exclude that position from consideration?

  3. Is the energy supply for production equipment to be provided on site or by outside means?

  4. What is the method used for delivery to and dispatch from the site, eg, large semi-articulated vehicles, small vans, rail, etc?

  5. Is there a need to isolate buildings from neighbouring sites due to environmental issues such as noise, vibration, etc?
By asking the above questions a designer, experienced in the design of industrial buildings, may be able to reduce the set of nine possible options. There are obviously many other questions that could be asked but certain critical issue questions, such as those mentioned above, have priority over less critical ones.

Figure 4 describes how a designer interacts with HYPEREX to arrive at a suitable set of options for positioning an industrial building on a large site. Here the designer selects the "building position" item under the "decisions to make" menu. This menu is a "card" in a HyperCard stack, and is the starting point of system.

Figure 4: An example from architectural design using an interactive multimedia based intelligent system

Figure 4: An example from architectural design using an interactive
multimedia based intelligent system

It should be noted that the type of advice given by a KBES at the conceptual stage of design should allow the individuality of the designer to be expressed. Therefore, in this example of building positioning, solutions are not expressed by the system as (say) cartesian coordinates but as types of positions thus still offering the designer considerable flexibility.

Through the use of appropriately designed KBES for design, a designer is able to achieve a significant reduction in the time usually involved in analysing design problems. They are also able to establish a conceptual framework that avoids costly redesigns at later more detailed stages in the design process.

Discussion

Intelligent systems technology incorporating interactive multimedia and knowledge based systems are still in early stages of development. They require considerable resources such as powerful computers, peripheral equipment, and people with a strong background in domain specific knowledge as well as an ability to use these technologies imaginatively.

Appropriate material collection and its presentation is a very time consuming and complex process. One could also encounter problems with regard to well known designers not being able to spend time to provide details of their works and decision making strategies. One also needs to take account of copyright issues related to the use of materials from a number of different sources. The developments taking place in computer hardware technology in conjunction with the techniques being developed by the authors should alleviate some of the problems.

Conclusions

We have presented an approach to developing interactive multimedia based intelligent systems in architectural design. This approach is well suited to architectural design as interactive multimedia technology provides such powerful design visualisation facilities and it can be in combination with design heuristics. This is of considerable importance for design decision making as so much of design involves 'visual thinking'. We have presented an example that is directed at the conceptual stage of architectural design. We feel that a system based on such an approach could grow over time and create a rich design environment. This would aid the decision making needs of the designer in a much better manner than it is currently possible.

References

Begoray, J. (1990). An introduction to hypermedia issues, systems and application areas. Int. J. Man-Machine Studies, Vol. 33.

Coyne, R., Rosenman, M., Radford, A., Balachandran, B., & Gero, J. (1990). Knowledge based design systems. Addison Wesley, New York.

Drury, J. (1981). Factories planning, design and modernisation. The Architectural Press.

Mackinder, M. & Marvin, H. (1982). Design decision making in architectural practice: The roles of information, experience, and other influences during the design process. Institute of Advanced Architectural Studies, University of York, UK.

Manning, P. (1984). A strategy for multi-professional design of buildings. Can. J. of Civ. Eng., Vol 11.

Murthy, H. & Lutton, L. (1993). The place of knowledge based expert systems in the CAD of buildings. J. The Architect, Spring issue.

Authors: Dr Hari Murthy, School of Architecture, The University of Western Australia, Nedlands; WA 6009. Tel. 09 380 2587. Fax. 09 380 1082. Email. hari@arch.uwa.edu.au

Mr Linley Lutton B. Arch. ARAIA, The Building Management Authority, 2 Havelock Street, West Perth, Western Australia. Tel: 09 222 5555 Fax: 09 222 5055

Please cite as: Murthy, H. and Lutton, L. (1994). Interactive multimedia based intelligent systems in architecture. In C. McBeath and R. Atkinson (Eds), Proceedings of the Second International Interactive Multimedia Symposium, . Perth, Western Australia, 23-28 January. Promaco Conventions. http://www.aset.org.au/confs/iims/1994/km/murthy.html


[ IIMS 94 contents ] [ IIMS Main ] [ ASET home ]
This URL: http://www.aset.org.au/confs/iims/1994/km/murthy.html
© 1994 Promaco Conventions. Reproduced by permission. Last revision: 8 Feb 2004. Editor: Roger Atkinson
Previous URL 28 July 2000 to 30 Sep 2002: http://cleo.murdoch.edu.au/gen/aset/confs/iims/94/km/murthy.html