This paper presents an approach to solving constraint satisfaction problems using Asynchronous Teams of autonomous agents (ATeams). The focus for the constraint satisfaction problem is derived from an effort to support spatial layout generation in a conceptual design framework. The constraint specification allows a high-level representation and manipulation of qualitative geometric information. We present a computational technique based on ATeams to instantiate solutions to the constraint satisfaction problem. The technique uses a search for a solution in numerical space. This permits us to handle both qualitative relationships and numerical constraints in a unified framework. We show that simple knowledge, about human spatial reasoning and about the nature of arithmetic operators can be hierarchically encapsulated and exploited efficiently in the search. An example illustrates the generality of the approach for conceptual design. We also present empirical studies that contrast the efficiency of ATeams with a search based on genetic algorithms. Based on these preliminary results, we argue that the ATeams approach elegantly handles arbitrary sets of constraints, is computationally efficient, and hence merits further investigation.

Much of preliminary engineering design is a constraint-driven non-monotonic exploration process. Initial decisions are made when information is incomplete and many goals are contradictory. Such conditions are present regardless of whether one or several designers contribute to designs. This paper presents an approach for supporting decisions in situations of incomplete and conflicting knowledge. In particular, we use assumptions and conflict management to achieve efficient search in contexts where little reliable information exists. A knowledge representation, containing a semantic differentiation between two types of assumptions, is used within a computational model based on the dynamic constraint satisfaction paradigm. Conflict management strategies consist of three generic mechanisms adapted to the type of constraints involved. These strategies may be refined through consideration of variable importance, context, and design inertia.

Science has developed detailed and well-founded theories for analyzing the behavior of artifacts. For example, Boeing was able to correctly verify an entirely new airplane, the Boeing 777, before any prototype was even built. However, there are few theories, and no computer systems, that would allow us to design structures with a similar degree of automation.

We believe that case-based reasoning (CBR) will prove applicable to design, at least in part, as we have seen designers making extensive use of past cases. Construction of useful systems, however, requires the resolution of many open issues. In this paper we consider three issues in particular: (1) What sort of content should be captured in a design case? (2) How should the content of a complex case be segmented into chunks for use? (3) How should the resulting chunks be indexed for retrieval? These are among the issues we are seeking to address through construction of Archie-II, a case-based aid for conceptual design in architecture. In addition to our approaches to these issues, this paper also discusses our strategic choice to build a design aiding system as opposed to a system that generates designs on its own.

Concurrent engineering is a systematic approach to the integrated, concurrent design of products and the related processes of manufacturing and support. This approach is intended to cause the developers, from the outset, to consider all elements of the product life cycle from concept through disposal, including quality, cost, schedule, and user requirements. To achieve successful concurrent-engineering design, one needs an integrated framework, a well-organized design team, and adequate design tools. The research on concurrent engineering to date has focused on developing communication infrastructure, design tools, and product data representations. Little attention has been paid to developing tools to address the organizational issues involved in concurrent engineering. The authors’ research on the Virtual Design Team (VDT) attempts to develop a computerized analysis tool to sup-port the systematic design of organization structures for concurrent engineering projects. VDT is a computer simulation system. It takes descriptions of design tasks, actors (i.e., designers and managers), and organization structure as input, and produces predicted historical records of the actors’ design and coordination behavior, project du-ration, cost, and design process quality as output. VDT has been applied to model more than ten realistic engineering projects, and the results are qualitatively consistent with the predictions from theory and project managers. The VDT framework for modeling concurrent-engineering teams is described, and examples of VDT applications are presented to demonstrate the effectiveness of the Virtual Design Team approach to modeling the organizational behavior of concurrent design teams.

A difficult task in concurrent engineering tasks is to provide appropriate manufacturability feedback to the designer in a timely manner. Many tools provide designers with cost and manufacturability evaluations, but they do not necessarily help the designer to identify what aspects of the design to change in order to improve it from a manufacturing perspective. A computer-assisted concurrent-engineering technology is described which identifies cost-critical tolerances in the design and generates cost-reducing design suggestions. The purpose of the system is to help focus the designer’s attention on the specific aspects of the design that influence manufacturing cost. This can aid the designer in optimizing the manufacturing costs of prismatic, one-off parts created on a CNC machining center. This work uses a program called the Manufacturing Evaluation Agent to produce cost-reducing design suggestions. The Manufacturing Evaluation is made up of two portions: a manufacturing planner and a suggestion generator. The manufacturing planner, P3, takes a design and generates a manufacturing constraint net that represents all manufacturing steps and their sequencing. Each constraint in the network has a label indicating the portions of the design (or manufacturing environment) that gave rise to that constraint. The Suggestion Generator analyzes the manufacturing constraint net to find the cost-critical areas, and uses the labels to find the portions of the design responsible for these cost-critical parts of the manufacturing plan. By pinpointing the cost-critical areas of the design and suggesting alternatives, the Manufacturing Evaluation Agent can help the de-signer to more quickly develop a superior design. This results in more rapid turnaround time, lower cost designs, and fewer engineering change orders once the design is sent to the factory floor.

Expert systems employing current methodologies suffer from two major problems: they are brittle and their development is time-consuming and tedious. Learning, the key to intelligent human behavior and expertise, has the potential of alleviating these difficulties. The paper reviews a number of machine learning techniques and provides a framework for their classification. The description of each technique is followed by an example taken from the domain of structural design. The applicability of machine learning techniques to expert systems is discussed, including some prototype applications and their shortcomings. Three promising research directions are outlined as a partial solution for the shortcomings.

Case-based design (CBD) systems aim to solve a design problem by tailoring previously solved design problems to the current problem. Designers' specifications are used for indexing the knowledge base of the CBD system to retrieve an appropriate design case. Menu-based systems fail to capture designers' specifications effectively due to lack of expressiveness, while natural language systems are too immature to satisfy the goal. This paper presents the development of a graphical user interface (GUI) to implement a mechanical design specification language (MDSL) (Stelling, 1994) used to facilitate indexing in case-based mechanical design. The specification language is context-free and hence computable. It represents mechanical design knowledge in a (feature):(attribute) format suitable for indexing. An augmented transition network (ATN) parser is built using the grammar of the specification language. The parser provides syntactic as well as semantic checks. It also has capabilities to expand grammar and to adapt to a specific user domain. A graphical front end to the parser assists and guides the user through the specification language syntax in entering the design specifications. Provisions have been made to expand or edit the language grammar and vocabulary. The ATN parser was implemented in Common Lisp and the graphical user interface was written using the Gold Hill Windows Toolkit. Sample user interactions with the interface and screen dumps of the GUI are included.

In cases where a domain theory can be successfully expressed in a logical formalism and can be used to formulate a task in that domain in mathematical terms, the task of building sound knowledge-based systems is greatly facilitated. However, it is not immediately obvious how the design aspects of such tasks, where these are an important feature of problem solving, can be incorporated in this approach. Design issues differ from search problems in that there may be several choices, each valid in some sense, but not (necessarily) equally good or equally appropriate in the current context. A case study is described in which a methodology is used based on the development of proof plans. The ability to conduct research according to the Popperian framework of hypothesis, validation, testing, and modification in response to empirical evidence – the hypothetico-deductive approach – seems essential to any rigorous scientific endeavor. It is believed that proof planning is a method which readily exploits this inherently incrementalist approach and could prove to be a powerful tool in designing AI systems.

Gradient-based numerical optimization of complex engineering designs offers the promise of rapidly producing better designs. However, such methods generally assume that the objective function and constraint functions are continuous, smooth, and defined everywhere. Unfortunately, realistic simulators tend to violate these assumptions. We present a rule-based technique for intelligently computing gradients in the presence of such pathologies in the simulators, and show how this gradient computation method can be used as part of a gradient-based numerical optimization system. We tested the resulting system in the domain of conceptual design of supersonic transport aircraft, and found that using rule-based gradients can decrease the cost of design space search by one or more orders of magnitude.

The synthesis of four-bar mechanisms is a well-understood, classical design problem. The original systematic work in this field began in the late 1800s and continues to be an active area of research. Limitations to the classical theory of four-bar synthesis potentially limit its application to certain real-world problems by virtue of the small number of precision points and unspecified order. This paper presents a numerical technique for four-bar mechanism synthesis based on genetic algorithms that removes this limitation by relaxing the accuracy of the precision points.

Recent knowledge-based expert systems for structural engineering design have focused on design-independent knowledge (abstract reasoning rules for designing), and while great strides have been made in that area, there is still a significant need to develop systems to take advantage of the wealth of knowledge contained in every substantial structural design. On the other hand, previous database-oriented design efforts have focused primarily on knowledge-poor databases of solutions, in which the traditional engineering handbook of solutions has simply been replaced by digital data. The challenge is to find a way to capture and apply the kind of case-based, design-dependent knowledge that structural engineers have traditionally used. The long-term results will be better structural designs and better structural designers. This paper discusses the character of the design-dependent knowledge in a structural engineering context, describes two initial applications of case-based reasoning to component design, and presents a general paradigm for a knowledge-based design system integrating rule-based and case-based reasoning.

The goal of this paper is to raise awareness and generate discussion about research methodology in engineering design. Design researchers are viewed as a single communicating community searching for scientific theories of engineering design; that is, theories that can be tested by formal methods of hypothesis testing. In the paper, the scientific method for validating theories is reviewed, and the need for operational definitions and for experiments to identify variables and meaningful abstractions is stressed. The development of a design problem taxonomy is advocated. Generating theories is viewed as guided search. Three types of design theories are described: prescriptive, cognitive descriptive, and computational. It is argued that to seek prescriptions is premature and that, unless the human and institutional variables are reduced to knowledge and control, cognitive descriptive theories will be impossibly complex. A case is made for a computational approach, though it also shown that computational and cognitive research approaches can be mutually supportive.

This paper describes current research toward automating the redesign process. In redesign, a working design is altered to meet new problem specifications. This process is complicated by interactions between different parts of the design, and many researchers have addressed these issues. An overview is given of a large design tool under development, the Circuit Designer's Apprentice. This tool integrates various techniques for reengineering existing circuits so that they meet new circuit requirements. The primary focus of the paper is one particular technique being used to reengineer circuits when they cannot be transformed to meet the new problem requirements. In these cases, a design plan is automatically generated for the circuit, and then replayed to solve all or part of the new problem. This technique is based upon the derivational analogy approach to design reuse. Derivational Analogy is a machine learning algorithm in which a design plan is saved at the time of design so that it can be replayed on a new design problem. Because design plans were not saved for the circuits available to the Circuit Designer's Apprentice, an algorithm was developed that automatically reconstructs a design plan for any circuit. This algorithm, Reconstructive Derivational Analogy, is described in detail, including a quantitative analysis of the implementation of this algorithm.