Click for next page ( 27


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 26
3 ISSUES THAT INFLUENCE COMPUTER INTEGRATION A number of issues can affect a company's willingness to pursue computer integration, the approach taken, and the likelihood of success. These issues can be categorized as technical, organizational, financial and accounting, and governmental. While many of the issues can be seen as barriers to computer-integrated manufacturing, recognizing them can provide opportunities to facilitate progress toward integration. TECHNICAL ISSUES Many problems perceived by organizations considering integration can be addressed with existing integration technology, but a number of technical barriers remain. Before computer-integrated manufacturing can reach its full potential for increasing productivity in both design and production, technical advances are needed in the following areas: data communication in a system in which both hardware and software are heterogeneous validation and consistency of data representation of integrated textual and geometrical data expert systems and artificial intelligence analytical models of manufacturing processes These advances are listed in the probable increasing order of difficulty, but their relative importance will vary among companies, depending on company size and products. As the advances are realized, the implementation of the first three must be standardized within any organization if they are to be effective. Data Communication The data communication problem in CAD/CAM exists at three levels, as in many other complex computer systems. At the first level is communication be tween di f ferent programs running on a single 26

OCR for page 26

OCR for page 26
28 The NBS and others are working on procedures to test the conformity of individual systems to the protocol standards, so that eventual purchasers can have confidence that a mixed vendor system will work. A number of computer vendors whose products are used for graphic design, analysis, and management are also considering offering these ISO standard protocols. Validation and Consistency of Data The validity and consistency of data in a manufacturing data base must be assured if it is to be a reliable and comprehensive source of manufacturing information. Unreliable and inconsistent information is probably worse than no information. Traditional data processing has had separate data files for each application program. As a result, manual or automatic data translation between files has been needed, and data were often inconsistent. It is now widely accepted that data files should be separated from particular application programs by limiting data access with a common set of data management routines that preserve accuracy and consistency. Validity generally means that all data entered into the data base obey any direct or calculated constraints imposed on them. This shifts the burden from the person or process entering the data to the formula- tion of the constraints, which must be accurate and tight enough to be effective, but not so tight as to inhibit legitimate change. Effective constraints will reflect models of product characteristics or process performance. One vendor has developed primitive automatic constraints on the designer, based on the capability of the production machinery and the material being fabricated. Consistency means that a change in one data item is accompanied by changes in related data items. It is enforced by prohibiting the entry of inconsistent data or by recalculating dependent information. Although recalculation is feasible for simple constraints, in many cases the constraint calculation may require too much time, or the nature of the constraints may not be understood well enough to be expressed in current software technology. Whether data are dependent or independent will depend on the point of view of the user. Part geometry, for example, is the independent variable from the point of view of the designer and dictates the process required to make the part; the manufacturing engineer, on the other hand, may view the available tooling as the independent variable that should constrain the part design. Most companies treat this as an organizational issue because techniques that provide for this multi- directional dependency are not yet well developed. Another consistency issue arises when changes by different people at different times are not coordinated. Many of these problems can be handled with automated versions of existing sign-off or release systems. These systems, for example, allow only one individual to change the master copy of a design file; all other copies are considered unofficial working copies used at the

OCR for page 26
29 risk that the official version may change. The IPAD program began to deal with this problem in a distributed heterogeneous environment of computers and data base software. Text/Geometry Integration Most computer-based graphic design systems originated as more efficient ways to create conventional drawings. The intent was to provide efficient and accurate storage of enough information to recreate or modify a drawing; generally, no attempt was made to make the stored information usable by other systems. As improved technology became available to analyze the functional behavior (e.g., strength, vibration, aerodynamics, heat transfer) of a mechanical part or assembly, it made sense to use the same data files for analysis that were used to generate drawings. This eliminated duplication of data entry and the possibility of inconsistency between the two data sets. A further step is to use geometric information for production more directly--to use descriptions of sheet metal geometry, for example, to calculate die designs, or use descriptions of part geometry directly to calculate tool paths for numerically controlled machine tools, or calculate drill patterns for printed circuit boards directly from the file that defines the metal patterns. The objective is to extract the required information automatically from the computerized, graphical representation of an object. If the information required by the production group is known in advance, algorithms (and therefore computer programs) might be con- structed to obtain it from the model. However, the amount of semantic information carried by the graphical representation (and potentially required for fabrication) is very large; it is impossible to prepare in advance a sufficiently complete set of programs to answer all the queries that may come from the production group. Indeed, it is diffi- cult to construct algorithms that yield answers to such simple queries as "what are the dimensions of the pockets to be milled?" or "which webs are thinner than 0.05 inch and higher than 0.5 inch?" It is quite difficult to construct a completely integrated CAD/CAM system in which the graphical representations of objects can be treated on a par with numerical or textual information whose semantics can easily be extracted as long as such data follow a prescribed format or syntax. The PDDI project is addressing the problem of making it easier to extract manufacturing process information from the design data base. Currently, queries from the production group are handled by engineers looking at a drawing. These manual operations are ~ major impediment to the development of an integrated manufacturing system. Replacing them by a computer program will require considerable research, involving pattern recognition, scene analysis, graphic modeling, and artificial intelligence.

OCR for page 26
30 Expert Sy8 tems The companies visited by the Committee viewed artificial intelli- gence as a long-term, not a current, need as they work toward CIM. Artificial intelligence is the part of computer science that is concerned with the symbol manipulation processes that produce intelli- gent action; intelligent action is goal-oriented and arrived at by an understandable chain of reasoning steps which are guided by knowledge of the world.1 The most highly developed branch of artificial intel- ligence is expert systems. An expert system captures the knowledge of an expert in a particular area and transfers it to other users. A few such systems are now commercially available, though not for manufac- turing problems. The lack of commercial products for manufacturing explains the present lack of demand for artificial intelligence. Applied research is needed in building knowledge bases for selected manufacturing processes, especially in planning. Research is needed to clarify the issues of designing for manufacturability before an expert system for that purpose can be built. The concept of geometric reason- ing is complex; it is difficult to create an expert system that can respond, for example, to the request to "find a symmetric object in the data base." A final research need for expert sys tems is to develop knowledge bases for making approximations in complex analysis prob lems . 2 Process Models More accurate mathematical models are needed for production processes, such as stamping, molding, cutting, grinding, or welding over a wide range of material or process parameters. Without them, engineers cannot perform the analysis and mathematical optimization of a process, nor attempt the synthetic construction of the process from the design data. Existing models tend to reflect the worst case. They do not indicate the real capabilities of a process under design and hence lead to far from optimum performance of the process in operation. The primary problem in modeling many production processes is the lack of knowledge about or ability to control many of the important variables. As the importance of improving productivity increases and the cost of sophisticated measurement and control systems decreases, it will be feasible to use quantitative models to plan and control processes. The major research need is for a self-augmenting data base. ORGANIZATIONAL ISSUES Factors that are primarily a function of the modus operandi of an organization can affect the integration of CAD/CAM as much as the technical issues, if not more so. These issues can be categorized as:

OCR for page 26
31 fragmental ion and isolat ion o f de per tment s support of top management implementation uncertainty education and training The history and geography of the United States and personal behavior patterns in U.S. society have a pervasive and lasting effect on the attitudes, responses, objectives, and approaches to problems and solutions that are found in manufacturing organizations. Our large domestic market and separation by oceans from other leading industrial powers has encouraged a tendency to ignore innovations in manufacturing technology in Japan, Europe, and the countries of the Committee on Mutual Economic Cooperation (COMECON), often to our disadvantage, until forced by the marketplace to acknowledge them. Our society is heterogeneous, and the social structure is based on personal freedom, competition, and rewards--both monetary and in status--based on individual effort. The result can be a tendency toward adversarial relationships, such as are observed between labor and management, government and industry, line and staff, and even between groups within departments. One source of power in that adversarial process might be an individual's control of the data that a CIM system needs to operate. Fragmentation and Isolation of Departments Manufacturing organizations, as described in Chapter 1, have been progressively divided into separate and to some extent isolated departments. Decentralization in the past has proved useful to large organizations, as it dispersed decision-making and allowed each department to concentrate on its respective contribution to the manufacturing process. At the large organizations observed by the Committee, the benefits of decentralization may now be outweighed by the barriers it creates for integration. The mass of data and the diverse data base structures, programming languages, computers, communications systems, and specialized hardware have presented groups with almost infinite numbers of alternatives. It is little wonder that these groups, seldom with integration or organi- zat iona 1 s tandards as an objective, have made selections based on their internal objectives, which are not necessarily compatible with objec- tives across the organization. The longer a group goes without integration as a major objective, the more committed it is likely to become to its often unique configuration, and the more difficult integration becomes. The more diverse and isolated the groups within the organization, the higher the probability of problems and the more difficult their resolution. Performance criteria, rewards, deadlines, and professional jargon are among the characteristics on which departments may differ. As a result, people in traditional organizations frequently exhibit loyalty to their individual disciplines and protect their turf to a degree that inhibits exchange of information and efforts to integrate. All too

OCR for page 26
32 frequently, they fail to view information, the data base, as a company resource--as valuable as money. Potentially even more damaging are groups that do recognize its value and restrict access through claims to ownership. Organizational fragmentation and adversarial relationships can hamper the introduction of a new technology or a new product. These problems could prevent the introduction of computer-integrated manufacturing because CIM affects all parts of an organization. The solution, many companies have found, is to supplement reliance on the traditional hierarchy for coordination with the creation of project teams made up of representatives of the departments that contribute to the manufacturing process. Communication and problem-solving within a group that crosses departmental boundaries is a step away from fragmentation. Some form of team management was used for computer- integration efforts by all organizations visited by the Committee, even if only on an ad hoc basis. Support of Top Management Computer-integrated manufacturing is knowledge-intensive and affects all parts of an organization; therefore, its success requires the participation of all levels of the organization. The keystone is unified data base, which may require years to realize. Without the commitment and understanding of top management and a top operations officer thoroughly committed to the concept, the coordination of changes throughout an organization cannot take place. In the company with the highest level of integration of those visited, the CEO actively supported the effort, even ordering a moratorium on new computing projects to stop proliferation of data structures while a unified data base was created. The absence of a long-term strategic plan covering technology' marketing, finance, and integration appears to inhibit the adoption of CIM in many instances. Such neglect is frequently a result of manage- ment's failure to appreciate the corporate benefits from integration. Often these benefits are difficult to quantify, probabilistic, and longer term, whereas managers tend to stress factors that are short term and easily quantified. One result of the nature of the benefits is that it is not easy to know when to begin integration. A firm that does not do the proper systems engineering in advance might end up, not with the "islands of automation" suggested earlier, but with "reefs of automation" that actually slow integration because of the reluctance to scrap investments despite difficulties in linking those investments in a system. The recommendation of several executives was that the CIM plan come from the top down, implementation from the bottom up. Implementation Uncertainty Even with the assurance provided by top management support, integrating CAD and CAM produces great uncertainty for individuals and

OCR for page 26
33 organizations. When confronted with the unexpected, companies may be tempted to abandon all efforts rather than modify their plans. One of the companies visited was an instance of this phenomenon: an all- encompassing integration plan was dropped after the early milestones proved difficult to attain. That company is now reconsidering how tees t to approach integration. Another impediment to integration is the likelihood that the transition will be uncomfortable, perhaps j eopardiz ing the power bases of key individuals. Integration will require changes in the management and control of organizations, even to the point of restructuring, and most people prefer to avoid the ambiguity end 'uncertainly that reorga- nization will be sure to bring. An infrastructure that includes a basic understanding of systems, feedback control, computing, data base management, automated control, local area networks, and computer application systems is a prerequisite to successful integration. It is unusual in 1984 to find such an infrastructure in any but the largest or most progressive companies. A company can acquire it through training, but the breadth of training required may be costly and time-consuming. In addition, the necessary basic talents must be available. A good machine operator does not necessarily make a good NC programmer, nor are the characteristics that make a good manager of a labor-intensive job shop necessarily those required for an automated factory. Even when a company has overcome these barriers and set integration as a goal, success is not assured. Several organizations that were studied had set such a goal and embarked on the program, but then lost sight of the original goal and settled for more limited objectives. This is not to indicate that integration should be an end in itself; it is a route to a more effective manufacturing system. Furthermore, it is a continuous variable, not binary or absolute; a company should attain the degree of integration that it can assimilate and not strive for 100 percent integration all at once. Supply of Labor The "factory" in this country is often perceived by managers and by the general public as a scene of labor-intensive activity, performed under dirty, noisy, sometimes dangerous conditions, and best avoided if at all possible. It is too often assumed that if finance, marketing, and product engineering perform their functions, production will take care of itself with little or no attention from top management: that production really doesn't require many important decisions and return- on-investment calculations will adequately cover them. Skinner has pointed out how inaccurate these impressions can be,3 and entire industries, such as consumer electronics and appliances, cameras, and motorcycles, have already suffered from reliance on such calculations. Long-held beliefs have lives of their own, however, and the image of production still limits the creativity and manpower applied to the solution of production problems. Students who might otherwise be interested generally prefer the prestige and challenge of a career in

OCR for page 26
34 engineering or management, with the result that few U.S. universities offer degree programs in manufacturing. Not long ago, more students of manufacturing were enrolled at the University of Stuttgart, or Dresden, or Karl-Marx-Stadt, than in the entire U.S. higher education system. People in many countries see manufacturing as a prime source of improvement in their standard of living, and careers in manufacturing as prestigious and satisfying. Society's main concern about automation is the threat of unemployment. However, while the level of automation has increased significantly over the past 200 years, it has increased incrementally. Unemployment levels averaged over a period of years have stayed at about Six percent. If the technological changes now under way raise unemployment to a higher base, the problem must be faced by society in both human and economic terms.4 Manufacturing has always required a variety of skills for the production of any item; further, the amount of effort expended varies from one skill category to another. As manufacturing progressed from a handcraft to today's high technology, these skill categories have changed and their relative proportions have varied. The past century has seen the virtual elimination of tasks requiring mere physical strength; during the past half century, automation has greatly reduced the tasks requiring full-time manual control by a worker. As a result, the number of people employed in "direct" labor has been declining. Those who could not learn any other skills became unemployed. The displacement has been incremental over the half century, however, and society has accepted, and partially adapted to, this erosion of the number of manual workers. Direct labor is now such a small percentage of the total cost of production that it is a small target for further automation, relative to other possibilities. The past decade has seen the transfer of tasks requiring mental skills for the control of production machines, or for the management of work flow in the factory, to data processing equipment. Again, the number of those employing mental skills in the direct control of equipment has diminished. However, the number of those employed in preparing the body of knowledge for execution by the data processing system has increased substantially. Similarly, the skill required to supervise the operation of a computerized system, and the number so employed, have also increased. The experience of this past decade suggests that most workers using mental skills could be retrained to prepare and control computerized equipment or to perform some other function requiring mental skills. Because of their better educational foundation, they are easier to retrain. The transition to CIM technology, therefore, should not in the future have a profound effect on the level of employment in an enterprise . It is hoped that the enhancement of productivity, quality, and speed of response to the market's demands will halt the emigration of industrial production to other countries, thus sustaining or improving the overall level of industrial employment in this country.

OCR for page 26
35 This optimistic view will be realized only with nationwide acceptance of this fact: Workers in the industry of the future must enter industry with a higher educational level, and with different skills, than today. They must accept the certain knowledge that skill requirements will continue to change at an ever increasing pace. The applicability of an industrial entry-level education is today less than half the expected working life of the entering worker. Educational updating of those employed in manufacturing will be an essential routine throughout a worker's employment. FINANCIAL AND ACCOUNTING ISSUES The United States emerged from World War II relatively unscarred and in a strong economic and military position. We had a large backlog of domestic consumer demand and technology, as well as readily available labor. The subsequent years were an ideal time for firms to grow and prosper. Anything that couldn't be sold domestically found ready world markets. Many began to believe that the U.S. industrial base was indestructible and that its fruits should be broadly shared. Political circumstances dictated a shift from military and investment spending to social and consumer spending. Taxes and labor costs increased. Many manufacturers attempted to maximize profits by offshore production and emphasis on short-term payoff from investment, often using cash payback or return-on-investment (ROI) calculations as the sole criterion. In a financially and technologically stable environment, emphasis on short-term criteria may lead to the highest profits, and firms that took this approach flourished for some time. During this period foreign competitors, sometimes fueled by national consensus and government support, invested aggressively in manufacturing facilities and technology. In the process, they took advantage of the dramatic decline in the cost of computer hardware relative to performance. In this economic context, the need for new financial models is clear. By models we mean the quantification of the appropriate variables and establishment of relationships among them to describe a situation and facilitate making decisions about it. The area that particularly demands improvement is evaluation of investments, particularly in new technology. In the past, ROI has been effectively used almost as a standard procedure by much of industry. Companies undoubtedly face situations, such as the comparison of projects for capital investment, in which ROI and other strictly financial measures are still useful. ROI, however, at least as presently used, is clearly not the appropriate method for making decisions on investments in new technology. Until a generally acceptable method is available to measure the long-run effects of investments in technology, the introduction of flexible automation and integration will suffer. ROI methodology assumes stability in the economy, technology, labor, and, most important, the marketplace behavior of competitors-- assumptions that have proven time and again to be false. In addition,

OCR for page 26
36 it stresses short-run returns rather than long-run strategy. No adequate effort has been made as yet in this country to remedy the situation. If the true costs and benefits could be included, ROI would be appropriate for determining the value of investments, even those as complex as computer integration of factory information. The difficulty lies in the disparity between the apparent ease of quantifying costs and the difficulty of quantifying benefits. The result has all too frequently been the following scenario. A new technology becomes available, but its costs (including education of workers) and the uncertainty of the benefits result in an unacceptable ROI. As refinements and enhancements of the technology become available, the investment is successively less cost-effective because of the higher costs associated with overcoming the ever-widening infrastructural gap. For some years, the company's failure to invest in the technology poses no significant problem because most of its competitors are following the same path. Then, other firms, perhaps foreign, begin implementing the new technology, even incurring early losses to do so. At some point these firms are able to reduce their manufacturing costs, which, coupled with stable market prices, enables them to generate sufficient funds for technological innovation and expansion, each stage built on a sound prior base. The innovator is often able to cut prices and increase market share, sometimes driving its competitors--with their outdated technology and facilities-- entirely out of the market. Unfortunately, the emergence of a gap in manufacturing technology is not a single, epoch-making event, easily noticed, but a continual process that slowly erodes the foundation on which the f irm operates. While ultimate collapse may be avoided, the U.S. manufacturing sector has not only lost once-viable firms through this process, but is on the verge of losing entire industries as well. The challenge for research on financial and accounting analysis techniques is to accommodate risky investments as a cost of long-term survival. Prof. Robert Kaplan reports5 that a theory now being explored would support investment on the basis of a combination of financial measures and such nonfinancial indicators as quality, inventory levels, productivity, delivery lead times, new product launch times, new product characteristics, employee training, employee morale and promotions, and customer and supplier satisfaction. While these factors are difficult to measure and, therefore, are seldom if ever included in traditional analyses, they are essential to the strategy of companies aiming to be world-class competitors. The Committee found in its interviews that the most common driving force for integration of information was actual or threatened loss of market share. If loss of market is the signal to begin improving manufacturing technology, however, the response frequently begins too late for the firm to recover. The message is clear: companies that face competition, particularly foreign competition, must have current manufacturing technology or risk cataclysmic consequences. The United States, like other industrial societies, has gone through phases of abilities fundamental to its well-being. Years ago, the successful firms were those that could make products. Later, the

OCR for page 26
37 ability to invent or design products became more important. Next, the important consideration became the ability to aggregate demand--to expand the market--so as to achieve economies of scale in production. The emphasis then shifted to finance and conglomerates. The widespread use of the digital computer in industry marked the dawn of the information phase, where the factors important to success include the ability to generate, transmit, maintain, and use informa- tion in operational and control activities. This phase influences every aspect of a company's operations. Individuals within organiza- tions may have to adapt to quite different functions and responsibili- ties. If technological changes are evolutionary, the costs will be small and incurred over a long time. If they are revolutionary, the time will be short and the costs may be high enough to force some companies from the marketplace. To remain competitive, companies must invest in training, a factor that is rarely included in financial analyses, but has an obvious impact on a firm's viability and long-term options. Many companies have compensation schemes that do not adequately reward individuals for integration, cooperative activities, and projects with long-range payoffs. In fact, most corporate incentive plans are oriented toward short-term profits rather than attainment of strategic goals. Clearly, either a more comprehensive method must be developed for guiding investment decisions, or decision-makers must be found who have both sufficient strategic knowledge of their firms and the backbone to break the viselike grip that ROI calculations currently have on management. The Japanese do not have the same commitment to ROI. It was reported recently, for instance, that the Seibu group has built a robotic grocery store in Nomidai, which the company president admits "will not be profitable by itself." It is the prototype for a fully automated store he plans for 1985 in the Tsuk~ba Science City north of Tokyo.6 A corporation has to invest in technology before reaping its benefits, and ROI calculations do not capture all those benefits. It is noteworthy that none of the firms the Committee interviewed mentioned ROI calculations as significant in its decision to undertake integration. GOVERNMENT ISSUES The federal government is not only a large purchaser of manufac- tured goods, but also has a vested interest in the benefits of a strong manufacturing sector throughout the economy. The goods-producing sector of the economy creates the basic, tangible wealth of a nation, the wealth that ultimately supports the entire economy.7 The services sector contributes to the standard of living and quality of life, but is itself dependent on the goods-producing sector. In the United States, manufacturing accounts for two-thirds of the goods-producing sector of the economy, with the other one-third divided

OCR for page 26
38 between the extractive industries (agriculture, fishing, and mining) and construction. Thus, improving the cost-effectiveness of manufac- turing has tremendous potential for improving economic and social well-being in this country. Computer-integrated manufacturing has demonstrated its ability to improve co~t-effectiveness, even at its early stages of development. Yet the government in its own activities has not promoted CIM or required its contractors to use it, either through contract specifica- tions or by establishing industrial standards. The government has recognized the need for improved manufacturing technology for military systems, as demonstrated by its Manufacturing Technology (ManTech) program. Commercial manufacturing has similar needs, but not such a beneficial program. We believe that government should encourage-- rather than discourage--cooperative efforts in integration techniques in commercial as well as military manufacturing. The government is the nation's largest purchaser of manufactured goods and in this role can provide leadership through purchasing specifications (as was done with numerical control of machine tools), lending capital equipment, requiring technology transfer on projects, and assisting industry in the development of standards. There is always a trade-off in the establishment of standards: if it is done too early, performance will be restricted and the full benefits of the technology never realized; if it is done too late, changeover costs will be so high that standardization will be impracticable. The U.S. manufacturing community generally believes that standards should be set as a result of usage, and that is it too early to consider standards for CAD/CAM or factory communications. The FEDD (For Early Domestic Dissemination) clause limits the use of the results of research or developmental projects funded by the DOD. It permits the distribution of the results of the projects within the United States, but prohibits their export to foreign parties. This provision precludes the use of the material in teaching or collegiate research where foreign students might attend a course or work on a research project. Nor can a professor publish any work that incorpo- rates material subject to the FEDD clause. Thus, the diffusion of new technology from DOD projects carrying FEDD clauses is substantially curtailed. While the establishment of the FEDD clause was motivated by concern for national security, we believe that the DOD should consider dropping the clause to improve the technology of its industrial base. Finally, the government can play a role by reducing the risks of technological innovation in manufacturing. This can be done indirectly by creating an economic, legal, and social environment that is conducive to and stimulates risk-taking; or it can be done directly through government programs that support technological innovation or the direct financing of high-risk or generic research. NOTES 1. D.R. Brown et al, R&D Plan for Army Applications of AI/Robotics (SRI International, 1982~.

OCR for page 26
39 2 . Committee on Science, Engineering, and Public Policy, "Report of the Research Briefing Panel on Computers in Design and Manufacturing" (Washington, D.C.: National Academy Press, 1983) , p.60. 3. Wickham Skinner, Manufacturing in the Corporate Strategy (New York: John Wiley & Sons, 1978~. 4. "The Process of Technological Innovation: Reviewing the Literaturet' (National Science Foundation, May 1983~. 5. Robert S. Kaplan, "Accounting and Control Systems for the New Industrial Competition" (Working Paper, Harvard Business School, November 1983) . 6. Geoffrey Murray, "High-tech bean-stockers invade Japanese supermarket." Christian Science Monitor (April 5, 1984), p.9. "The National Role and Importance of Manufacturing Engineering and Advanced Manufacturing Technology" (Position Paper by the Society of Manufacturing Engineers, Dearborn, Michigan, May 1978~.