TECHNOLOGY TRANSFER Fall 1990

49

BOOK REVIEW David A. T a n s i k Book Review E d i t o r

I n t e l l i g e n t M a n u f a c t u r i n g S y s t e m s . Andrew Kusiak. Englewood Cliffs, New Jersey: Prentice Hall, 1990. This book is written to provide the reader with a very detailed and advanced-level introduction to how knowledge-based software (read artificial intelligence or expert systems) has been developed for manufacturing systems. The book begins with an intended discussion of the overall structure and components of a computer-integrated manufacturing (CIM) system. While the chapter fails to provide such a structure, it does give brief hints at some issues relevant to computer-based manufacturing. The reader is encouraged not to look too carefully at the details of the first chapter since many of its statements are of arguable validity. For example, CIM implementation is not increasingly rapid as the author argues; nor does the existence of computeraided quality-control systems imply that the human involvement in quality control is greatly reduced (changed, yes; reduced, no). A second chapter on machining and assembly systems (which is the core production process for manufacturing systems) is followed by two chapters on what intelligence means to the author: knowledgebased systems and machine learning. These four are "fundamental" chapters. The next 10 describe the details of knowledge-based systems that have attempted to address fairly traditional manufact u r i n g problems, such as part design, process planning, equipment selection, and machine layouts. There is no concluding chapter. This book is reviewed for the Journal of Technology Transfer because Kusiak contends that it "deals with the design and management issues in an automated manufacturing environment," He further says: "lit is his] belief that the functional separation into 'design' and 'management' should not take place in modern manufacturing systems. The design of a part, product, or system may have a significant impact on the complexity of the system management phase. Any design activity should consider

its implications on the system operations and management aspect." However, only if Kusiak narrowly defines manufacturing management to be the making of very specific decisions about the actual design of manufacturing systems (decisions that, in most companies, are made by manufacturing engineers and not by managers) can the book be construed to be concerned with management. Similarly and unfortunately, technology-transfer issues are all but ignored, except for a few mentions here and there about the difficulties of finding data to support certain intelligent systems. One of the fimdamental problems the book has in addressing a managerial audience is the nature of what Kusiak appears to think is involved in management, and the degree of detailed knowledge that a manufacturing manager must have to manage an intelligent manufacturing system. He appears to assume that if managers are to manage intelligent manufacturing systems they must have a working knowledge of: LISP, knowledge representation such as production rules and semantic networks, knowledge-acquisition methods such as protocol analysis, and machine-learning techniques. In addition, he believes that managers should have a detailed understanding of a variety of knowledgebased software systems that perform such functions as design of mechanical parts, process planning, equipment selection, group technology, machine layout, and scheduling. This detailed knowledge also should include feature extraction/recognition, specific rules and algorithms used, optimization models, and the programming codes. While I am not philosophically opposed to expecting managers or technology-transfer agents to know a great deal about the specific technical aspects of the problems and decisions they confront, my reading of this book is that, for the practicing manager, it loses the forest for the trees. That is, the overall situation into which intelligent systems are fitted is rarely, if ever, mentioned. For example, Kusiak usually does not adequately explain why a knowledge-based system is better suited to a problem than existing systems. The reader is also often

50

T E C H N O L O GTRANSFER Y Fall 1990

left with virtually no examples of actual applications of knowledge-based systems to actual problems, including follow-up on how the use of a system actually met or could potentially meet a company's business objectives. All the knowledge-based systems described are applied to theoretical manufacturing problems. In a similar vein, the reader is often asked to accept without explicit mention or explanation the criteria for judging that a solution has been found. Any manufacturing manager knows that decisions made in a manufacturing environment often must optimize multiple, sometimes conflicthag, goals. While, in the preface, Kusiak warns the reader that the knowledge-based systems are primarily effective for solving "well-structured problems" (and it can be argued that not many actual manufacturing problems fall within this band-width), the reader at least deserves to be informed of the criteria by which any particular problem (such as a machine-layout problem) is judged to be well-structured and solved. Is the layout problem solved when the most efficient solution is found; if so, how is efficiency measured? Moreover, what happened to criteria of flexibility, quality, etc.? The book is also not written in a manner that allows for an easy understanding of the issues. Nomenclature that is not known to the average manufacturing manager (even those with some rudimentary knowledge of intelligent manufacturing systems) is simply presented in the book without explanation. For example, this reviewer called a few of her manufacturing-modernization friends in industry to ask if they knew what "state-space search" and "activation decay" meant and all indicated ignorance. At the very least, a glossary of terms would have helped; more appropriately, an effort to reduce the jargon would have been better. Several chapters begin with no introductory discussion. Another source of consternation is when the author makes unexplained assertions. For example, at the beginning of the chapter on "Aggregate Scheduling of Machining and Assembly Systems," he says, "One of the most difficult problems arising in FMSs is the scheduling problem." (p.338) The reader must apparently simply believe this statement since no elucidation is provided.

Given that the current reviewer comes from a broader background than intelligent manufacturing systems, I was also quite uncomfortable about the numerous assertions that are clearly unfounded when examined from a broader perspective. For example, Kusiak states that the layout of machines in a flexible-manufacturing system (FMS) is determined solely by the type of material-handling equipment used. What happened to other considerations, such as the human need for information sharing or task completion, or the most efficient sequence for producing parts to keep variations in quality to a minimum, or likely breakdowns and capacity needs to avoid bottlenecks? Because of its writing style and detail, the book is much better suited to a manufacturing engineer interested in getting very involved in the design of intelligent systems than a technology-transfer agent or a manager responsible for design of that system. This is quite unfortunate since the book leaves the impression, based on the preface, that if the reader is able to understand its excruciating detail, then he/she will be able to "understand how the design of a part, product, or system may have a significant impact on the complexity of the system-management phase" (p.xv). That impression is quite wrong. The book does not discuss management implications of the design of any part, product, or system. The entire literature on managing system implementation, the technology-transfer process, and managing intelligent-manufacturing systems once they're up and running has been completely ignored. In fact, it is the reviewer's impression that if the author had not described the intention of the book as marrying design and management, the book would be most usefully applied in a manufacturing-engineering graduate course on intelligent-manufacturing systems. With this book as a foundation in the field, others could be used to help the student marry the design of intelligent systems with their management.

Ann Majchrzak Associate Professor University of Southern California

T E C H N O L O G Y T R A N S F E R Fall 1990

51

Continued from Page 4 Iranians died under [Hussein's] artillery, half the casualties ofthe...Iran-Iraq war"). In the late 1970s, the US Army rejected Bull's advanced howitzer designs, which were based on research he did "under Pentagon contract." Then "US officials quietly sent South Africa, Iraq, and China to Bull's door, allowing them to equip their forces with the long-range cannons the Pentagon wouldn't buy." Bull later was jailed for smuggling shells to South Africa after he had obtained a "letter of waiver from the State Department's Office of Munitions Control...to export shell forgings without a license." After his release from prison, "infuriated" at the US, he moved to Brussels from where he sold his design for a towed 155mm howitzer to Austria's Voest Alpine steel works. In 1982, Voest Alpine sold 200 of the cannons to Jordan, knowing they were destined for Iraq. 9 R i c h m o n d Times-Dispatch, J a n u a r y 1991. Iraq may be capable of using a fuel-air explosive (FAE) possibly 10 times more powerful than conventional explosives. The bomb, which releases an expansive cloud of volatile fuel that is ignited by a second explosion, is said to be able to level an area about a mile around and create a fire storm. Much of the technology was developed in the US, according to Herbert Krosney in the Jerusalem Report. He said, "Senate investigators believe [a] German firm might have transferred that.., technology to Iraq." In December 1990, "the Los Angeles Times reported that [Honeywell] had sold design data for the FAE to the Iraqis." 9 The New Yorker, Feb. 18, 1991. "Chemical companies in West Germany, France, Belgium, and Italy are reliably reported to have supplied Iraq...with precursor materials for the manufacture of chemical weapons." US companies apparently were not involved in such trafficking. Although federal controls may have influenced this restraint, the American chemical companies may have been more responsive to the prospect of subjecting"themselves yet again to hostile campaigns by peace groups, environmental groups, and just plain citizens." The business and political entanglements involved in these technology transfers are beyond the scope of this editorial. But the ethics are not. Is technology transfer simply a "mechanical" process? Most of what's written and discussed by professionals in the discipline has to do with the transfer process: What are the most effective ways of moving a technology from here to there? Between which providers and users. Under what circumstances of user capability? Through which intermediaries? With what kind of attention to patents and licensing? These are the mechanical elements of transfer. There is, however, more for us to think about and discuss. In the case of Saddam Hussein's procurements, the horrendous potential of the transfers must have been known to most participants. But what may a mid-level practitioner or high-level technology-transfer manager, with the sharpest political and military prescience and hair-trigger virtue, do when faced with transferring sure pain and death to thousands of h u m a n beings and chaos to nations? He or she may be overwhelmed by the appalling vision and morally, but not professionally, paralyzed. Or the prospect may be an abstraction that raises no more than a twinge before it is suppressed. Or, aware of the consequences of a transfer, a conscience-stricken professional may sidle over to a supervisor, murmur his or her concern, and be subjected to impotent sympathy or rebuke. Or a moral giant may go as high as the boardroom where the response may vary from dire to '%Ve'll consider your points." I think that most professionalsmwith standard awareness and principlemwould be inwardly concerned in a variety of ways, dependingin part on their political beliefs, and would do their jobs without a fuss. I've raised this issue because technology-transfer people are important 9 contributors to local, national, and global development. We are not mindless. We are more than cogs in a big, dumb machine. We are capable of considering and discussing the value and impact of our work and work relationships. Your thoughts and suggestions are welcome. RS