Collaborative Knowledge Sharing in the Inter-disciplinary Project Team

Introduction

 Over the past decade a succession of strategic reports (Latham, 1994; Egan, 1998; UK National Audit Office, 2001) reported on the inability of the UK construction industry to deliver a high quality product to its clients at acceptable cost and within an acceptable time. In the USA (Wright, et al 1995) also suggested that greater value should be offered to the clients of construction services. Their concerns centered on an industry that is under-achieving, recommending that substantial improvements in quality and efficiency were required. Central to the challenge is finding a better way in which all the key participants could work together with the client being core to the process. The improvements targets set for both the USA and UK construction industries are indicated in table 1:

Table 1 - Construction Sector Performance Improvement Targets for the USA and UK

The source of this information is: USA – Wright Rosenfield Fowell; UK - The Engineering and Physical Science Research Council’s Innovative Manufacturing Initiative Programme.

To achieve these improvements efficiencies will have to be found throughout the supply chain.

The Design Process

To understand how to build an effective decision support environment we need to first understand how architects, engineers, construction managers and all the other inter-related construction disciplines carry out their work. What are their processes? How dependent is one disciplines process to another or how reliant should it be to get to the best solution? How can they be best supported? What do they need?

Building design is a complex group problem solving process that Whitney (1990) describes as involving the simultaneous evolution of both the requirements and the artifact specification. However design-in-practice consists of many additional problems such as requirements analysis, negotiation, communication and conflict resolution.

Fundamentally the process of design is a complex activity involving a number of tasks that are generally broken into sub-tasks, with a number of alternative methods potentially available for each sub-task. Those design tasks are driven by certain input parameters, e.g. goals, preconditions, to produce some output parameters, e.g. layout, resources, constraints, etc.  Chandrasekaren (1989) proposed that design be defined as a hierarchy of sub-tasks that can be solved by conducting a task analysis. A task-structure is then developed that lays out the relationship between tasks, applicable methods for solving the task, the knowledge requirements for the methods, and the sub tasks generated.

How to break the design activity down into tasks forms a key area of research in task-oriented methodologies especially for knowledge-based systems and can be referenced in Pohl (1993) work who concluded that the architectural design process could be characterized by five functional elements:

It is within these five areas that characterize the process used to develop a solution for a building or a structure. It is within these five areas that the partnership between machine and human agents could be one that each is complimented by the strengths of the other in an intelligent way. Humans use complex cognitive skills whereas machines are indefatigable in their mechanistic search for information that they bring to the collaborative environment.

Intelligence in the context of this work implies that the design system has some means that allows it to anticipate the data needs, information needs or knowledge needs of the human designer. The system would act as an intelligent assistant to the evolving design, aiding the designer and freeing them from being overwhelmed with untimely knowledge. Providing such assistance to all problem solvers in the design environment requires an understanding of the various participants’ knowledge, factors that constrain their decisions and criteria they work under. Pohl (1993) called this an Intelligent Computer Assisted Design System (ICADS).

The ICADS approach is supported in several working models (ICADS-DEMO1 (Pohl, 1989), ICADS-DEM0 2 (Pohl, 1991), AEDOT (Pohl, 1992). These have provided computer scientists with a useful test bed for the development of a body of knowledge relating to software and hardware computer architecture, theoretical concepts and technical implementation issues. By linking the design objects to information they represent (e.g. functions, relationship to other objects, cost) the information value of drawings can be significantly increased.  This information could be contained as attribute data in relational databases. Advances in the object-oriented modeling paradigm advanced this concept. Having this ability to view the artifacts used in the design model as a series of objects, which have implicit attributes and features, gives scope to analyze the design with regard to such aspects as manufacture, constructability, cost, quality, safety, etc., an almost unlimited definition of machine agents could be specified that is the caretaker of knowledge pertaining to most of the constraints and criteria related to a new building project.

Team Problem Solving

Team problem solving is characterized by more than two people being involved in attempting to reach a collective decision, each with their own perceptions, expertise and commitment towards that problem which they all recognize in varying degrees. However, many problems solved in construction are resolved by a single domain rather than an inter-disciplinary team. Architects and engineers are used to working in relative isolation when proposing a solution. Construction Managers on the other hand make their decisions based on a team approach. This is well documented and construction industry research efforts over the past twenty five years have sought to find ways of effectively integrating project decision making into a team approach with timely knowledge support across all the building professions associated with a project from start to completion. ‘Buildability’ and ‘Constructability’ are terms used in earlier attempts, then came a wave of expert system shells, then intelligent computer agents. The one common purpose was to find ways to share domain knowledge thereby developing a more cost effective solution; solutions that are developed concurrently across multi-disciplinary teams rather than sequentially by isolated domains.

Some of this early research; Bui (1984) determined that group decision making can be undertaken in either: [i] a co-operative environment where there is equality and respect for other opinions; [ii] in the non co-operative situation which is characterized by conflict and competition; [iii] in the group decision making situation where ultimately one particular manager makes the decision and assumes responsibility. Espejo (1991) considered that an individual expert working in a coordinated team to solve a defined problem would not always succeed if there were failure to communicate with other team participants. His evidence from observing designers at work brought him to the conclusion that there is a tendency for designers to concentrate on their local solution and ignore how their decisions will impact on others. One reason for this is human activities in problem solving are often influenced by individuals finding stability in interpersonal interactions rather than the single focus of achieving certain goals. There is overwhelming research evidence that to arrive at the ‘best’ solutions entail creating the conditions and environment for sharing experiences and developing an understanding of other domains viewpoint and criteria. Gallegher (1990) calls this the concept of “intellectual teamwork”. He observes that information system designers working to create technologies that will help groups perform more effectively require not only technical expertise, but also an understanding of the social and behavioral processes that the technology is designed to support. Therefore, a collaborative computer-based design environment must not only rely on the hard system that selects stored  knowledge at the appropriate time but must also reflect the soft systems approaches that lead to design excellence.

The research carried out by the author, Jones (2004, 2003, 2002, 1998, 1995), conceptualized such an environment. In it the inter-disciplinary team could be supported by computer agents to work towards building solutions in a collaborative way. The human/machine environment would be used to fully consider issues that effect bringing the best solution based on best practice to the client; issues such as production, quality, safety, cost, environmental factors, life-cycle effectiveness, performance, etc.

Concurrent Engineering

Concurrent Engineering refers to the practice of incorporating various values of the product into the design at its early stages of development. These values address the entire supply chain of the product and include not only its primary function but also the manufacturing, marketing and operational stages of the production process. It is imperative that one of the goals of any machine/human partnership in building design must incorporate a concurrent engineering approach in the inter-disciplinary building team.

Howard (1989), found that the vertical fragmentation between project phases, (e.g. conceptual design, detailed design, bidding, construction), etc., and the horizontal fragmentation between participants, (e.g. architects, structural engineers, planners, estimators, builder, etc.), is unparalleled in any other manufacturing sector. Syan (1994) listed the weaknesses in the sequential method of product development as leading to:

Results reported from research, Turino (1994), indicated companies saw significant benefits in using a CE approach including 70% of participants reported a shorter time for their product to reach its market; 59% saw improved communications; 56% saw improved product quality; 33% saw reduced development costs and better management; 48% saw reduced design changes and 30% saw increased profits.

Collaborative Agent Partnerships

The advances in the concept of an object as a high-level information source led to the paradigm of object-oriented modeling and the development of object-oriented computer languages.  The premise is that a crucial element in the decision making process that human designers utilize to solve problems is the reliance they place on their ability to identify, understand and manipulate objects, e.g. architects develop solutions by reasoning about location, sites, buildings, floors, spaces, walls, windows, doors, and so on; the contractor does likewise.

Each of these objects encapsulate knowledge about its own nature, its relationships with other objects, its behavior within a given environment, what it requires to meet its own performance objectives and how it might be manipulated by the designer within a given design problem scenario. This knowledge is contained in the various representational forms of the object (e.g. factual data, algorithms, rules, etc.).

Within the computer agent environment proposed by the author; problem solving is seen as a co-operative process with mutual sharing of information across an inter-disciplinary project team to produce a solution. The resulting project solution is seen as an assembly of construction objects, e.g. bricks, walls, floors, windows, etc., to satisfy project specific criteria, e.g. quality, environmental, cost, safety, etc.

Whereas, objects are information entities only, computer agents are active and have knowledge of their own nature, needs and global goals. Objects are therefore accessible by agents but cannot take action. However, for the system to interact effectively between the interactive project team there has to be a full description of the objects. This description should resemble as closely as possible the designer's real world by including the objects physical appearance, attributes, context and relationship to other objects.

Within the computer environment the “agents” also have the ability to communicate and take action. Typically, each agent is represented at a level of detail sufficient to facilitate the project team’s decision making. The frames in such a project model could represent geometric, physical and administrative attributes of a project's components together with their topological structure. All of this information about the structure of a project and the local values of its component attributes are then available in a representation easily accessible by computer tools for solving or assisting with design and production tasks.  

Construction Computer Agents

There is an inevitable need for interaction between all the participants who input to complete the final project. Pohl (2000) suggested that the computer system should reflect the more realistic situation of a team that interacts by co-operation and persuasion. The concurrent engineering concepts apply here. Therefore, complete families of computer-agents that represent a particular domain could be built e.g. architect, interior designer, civil engineer, landscape architect, safety manager, quality manager, environmental manager, mechanical and electrical engineer, construction manager, project manager, etc. and within each family specific agents would monitor and offer assistance regarding criteria and constraints imposed in the areas of environmental, quality, safety, cost, production time, etc. For instance there could be a ‘Quality’ agent residing in a number of domains i.e. Architect, Construction manager, Project Manager, Quality manager, each would be representing the criteria and constraints of that domain.

It must be stressed that project solution development assisted by computer agents is not intended to automate the design process. Agents would initially assist the designer in the partnership by acting as co-operative search agents having the ability to liaise with knowledge bases in the search for alternative solutions. They are evaluators and solution proposers acting as system agents who operate in a defined domain. They exist to first express opinions about the current state of the design solution. The intention is to change incrementally the current state of the design through the interaction among the various agents within the environment. As pointed out previously this environment would include representation from all the built environment disciplines, agencies and client. This interaction enriches the environment with information about the current design state and how it relates to the project requirements. It should support the project team by providing adequate information about the current design state, its design objects (i.e., data-objects and object-agents), their relationships and how they satisfied the various project criteria and constraints. 

Each agent would provide two kinds of support; intermittent foreground responsiveness to requests for information initiated directly by the designer and other members of the project team, and continuous background monitoring that evaluates the evolving design/project solution.

The human agent’s role in such an environment is seen as:

Operating in such an environment would be computer agent types (Pohl, 1994) including:

System-agents resident in the design environment proposed would be:

The Proposed Decision Support Environment

In such an environment the facilitator's role would be one of searching, evaluating and modifying initially the current design state with the support of different domain computer agent families (Jones, 1994).  In this process the facilitator would direct and guide the efforts of all computer agents to advance the current state of the project solution towards a best design that is acceptable to all domains computer agents’ and the inter-disciplinary project team. The role of the designer or project leader would be that of principal long term or strategic planner while agents would focus mainly on short-term activities.  The characteristics such computer agents would possess are:

Conclusions

The integrated partnership environment proposed is one that fully utilizes the strengths of a multi-agent collaborative computer environment and the human domain built environment experts. A total project decision making and problem solving environment, where the knowledge and intelligence of all domain-contributing agents can be used to create better opportunities to arrive at the project solution. All contributors to the project are collaboratively drawn into initially the design process and then through all phases of the project life-cycle. Time is saved because a concurrent problem solving approach is adopted rather than a sequential problem solving approach. Experts can still be geographically or functionally distributed taking advantage of recent advances in technology in communication systems (co-operative distributed, broad band, etc.). Importantly, the environment proposed could be extended to continually monitor and assist decision making throughout the life cycle of the project.

References

Bui, X.T. and M. Jarke (1984), A Decision support system for Cooperative Multiple Decision Making, Proceedings of the 6th International Conference on Information Systems, Tucson, Arizona,

Chadrasekaran, B (1989) A Framework for Design Problem Solving, Research in Engineering Design, vol 1, no 2, pp 75-86.

Egan, Sir John, (1998) Rethinking Construction, Report of the Construction Task Force, DOE, UK.

Espejo, R (1991) Strategy, Structure and information Management, in Espejo and Schwaninger eds. Organizational Fitness: Corporate Effectiveness through Management Cybernetics, Frankfurt/N Y. Feigenbaum, E. A. and P. McCorduck, The Fifth Generation, Addison-Wesley, Reading, MA, 1983.

Gallegher, J and R.E. Kraut (eds) (1990) Intellectual Teamwork, Lawrence Erlbrum Asso. Pubs.

Jones, B.K. and M.J. Riley (1994) Construction Problem Solving in a Co-operative Distributed Agent Centered Environment, 1st congress on Computing in C.E. Washington D.C.

Jones, B.K. and M.J. Riley (1994) Co-operative Distributive Construction Problem Solving, 7th Intl. Conference on Systems Research, Informatics and Cybernetics, Baden-Baden, Germany.

Jones.B.K. and M.J.Riley (1995) Collaborative Construction Agents, ASCE, 2nd International Congress on Computing in CE, Atlanta, USA, June 1995, pp 1316-1323. International Conference, Technical paper and presentation.

Jones.B.K. and  M.J.Riley (1995) Autonomous Construction Agents in an ICADS environment, ASCE,6th International Conference in Civil and Building Engineering, Berlin, Germany, July 1995, pp 407-412. Technical paper and presentation.

Jones, Barry K (1998) A Model for Collaborative Engineering in the Construction Industry, PhD Thesis, University of Southampton, UK

Jones, B.K. (2002) Collaborative Problem Solving on Construction Projects, ASC, The Second International Conference on Information Systems in Engineering and Construction Cocoa Beach, Florida (ISEC 2002); June 14^th 2002

Jones, B.K (2003) “Human-Machine Collaboration to bring essential Knowledge to the Inter-disciplinary construction team”, International Conference– Construction Project Management Systems: The Challenge of Integration (CIB W99), San Paulo, Brazil 25-28 March 2003 – Professional development fund award

Jones, B.K.(2003) “Culture and Trust in Construction” ASCE Civil Engineering Conference and Exposition, Nashville, Tennessee, November 12-15, 2003, Speaker and technical session coordinator with invited speakers from the UK, India and the USA.

Jones, B.K. (2004) advisory paper “The implications of Latham, Egan and Urban Task Force reports for Interdisciplinary working”

Jones, B.K (2004) “Collaborative Decision Support Systems, 16^th International Conference on Systems Research, Informatics and Cybernetics, Baden-Baden, Germany, July 29^th 2004

Latham, Sir M (1994) Construction the Team, Final report of the government/industry review of procurement and contractural arrangements in the UK construction industry, HMSO

National Audit Office, UK , Modernising Construction,(2001)

Pohl, J., L. Myers, A. Chapman, and J. Cotton (1989). "ICADS: Working Model Version 1," Technical Report, C,ADRU-03-89, CAD Research Unit, Design Institute, School of Architecture and Environmental Design, Cal Poly, San Luis Obispo, California.

Pohl, J., L. Myers, A. Chapman, J. Snyder, H. Chauvet, J. Cotton, C. Johnson and D. Johnson (1991). "ICADS Working Model Version 2 and Future Directions," Technical Report, CADRU-05-91, CAD Research Center, Design Institute, College of Architecture and Environmental Design, Cal Poly, San Luis Obispo, California.

Pohl, J., L. Myers, J. Cotton, A. Chapman, J. Snyder, H. Chauvet, K. Pohl and J. La Porta (1992). "A Computer-Based Design Environment: Implemented and Planned Extensions of the ICADS Model," Technicol Report, CADRU-06-92, CAD Research Center, Design and Construction Institute, College of Architecture and Environmental Design, Cal Poly, San Luis Obispo, California.

Pohl,J and L. Myers (1993) A Distributed Cooperative Model for Architectural Design, CAD Research Centre, Cal. Poly, San Luis Obispo, CA

Pohl, J., Myers, L and A. Chapman (1994) Thoughts on the Evolution of Computer-Assisted Design, Technical Report, CADRU-09-94, CAD Research Center, Design Institute, College of Architecture and Environmental Design, Cal Poly, San Luis Obispo, California.

Pohl. J (1996) Agents and their Role in Computer-Based Decision Support Systems. 8th Intl. conf. on Systems Research, Informatics and cybernetics, Baden-Baden, Germany, Aug 1996.

Pohl,J., A Chapman,. K Pohl (2000) Computer-Aided Design Systems for the 21^st Century: Some Design Guidelines, Collaborative Agent Design (CAD) Research Center, San Luis Obispo, CA.

Pohl Jens(2001) Information-Centric Decision-Support Systems: A Blueprint for ‘Interoperability’, Office of Naval Research: Workshop on Collaborative Decision-Support Systems, Quantico, VA, June 5-7, 2001

Syan, Chanan and U Menon (1994) Concurrent Engineering, Chapman and Hall, ISBN 0-412-58130-2

Turino, Jon (1992) Managing Concurrent Engineering: Buying Time to Market, Van Norstrand Reinhold, NY.

Whitney, D. (1990). Designing the design process. Research in Engineering Design, 2,  3-13.

Wright, Richard, Rosenfeld, Arthur and Andrew Fowell (1995) "Construction and building: federal research and development in support of the U.S. Construction Industry"pub United States - National Institute of Standards and Technology