Gerovitch Automation

8/7/2019 Gerovitch Automation

1/5

122 AUTOMATION

Planware has several types of knowledge, all encodedthrough parameterized heories. The first s knowledgeof the scheduling domain, including the constraintson use of the different types of resources, such as reus-able or sharable resources. Another type of knowledgeis algorithm nowledge, such as generate-and-test,branch-and-bound, divide-and-conquer, dynamic pro-gramming, and hill-climbing( see ALGORITHMS,DESIGNAN D CLASSIFICATION F) . By codifying them as para-meterized theories, algorithms can be automaticallyderived or a givenvery-high-level problem specifi-cation, given appropriate domain axioms.A third typeof knowledge s implementa t ionknowledge, whichdefines how higher-level onstructs such as sets can beencoded as more implementation-level constructs suchas lists or bit-vectors.All of these tools use advanced knowledge representa-tion and automated reasoning capabilities. Althoughresearch tools today, they represent thedegree ofprogramming automation that may become commer-cially available within a decade.

Bibliogvaphy1956. Newell, A , , and Simon, H . A.The Logic Theory

Machine, IR E Transactions on In format ion Theory ,IT-2, 3(March), 61-79.

1963. Simon, H . A. Experiments with a Heuristic Compiler,1982. Martin, J . Application Development without1986. Green, C., Luckham, D., Balzer, R. , Cheatham, T., and

Journa l of the ACM, 10, 493-506.Programmers. Upper Saddle River, NJ: Prentice Hall.Rich, C. Report on a Knowledge-based Software Assistant,in Readings in Artificial Intelligence and SoftwareEngineering (eds. C. Rich and R. C. Waters ). San Francisco:Morgan Kaufmann.

1990. Rich, C., andWaters, R. C. The Programmers Apprentice.Reading, MA: Addison-Wesley.

1991. Lowry, M. R., and McCartney, R. D. (eds.) AutomatingSoftware Design. Cambridge, MA: MIT Press.1993. Kant,E. Synthesis of Mathematical Modeling Software,IEEE Software, 10 , 3 (May), 30-41.1995. Flener, P. Logic Program Synthesis f ro m IncompleteInformation. Boston: Kluwer Academic Publishers.1996. Smith, D. R., Parra, E. A , , and Westfold, S. J. Synthesis

of Planning and Scheduling Software, in Advanced PlanningTechnology (ed. A . Tate) , 226-234. Menlo Park, CA:AAAIPress.1997. Browne, T . , Davila, D., Rugaber, S., and Stirewalt, K.Using Declarative Descriptions to Model User Interfaces withMASTERMIND, in Formal Methods in Hu ma n ComputerInteraction (eds. F. Paterno and P. Palanque). New York:Springer-Verlag.for A pplications. Boston: Kluwer Academic Press.1998. Bibel, W., and Schmitt, P. Automated Deduction. A Basis

WebsitesAmphion,http://ic-www.arc.nasa.gov/ic/projects/Mastermind. http://ww.cc.gatech.edu/gvu/

amphion/.user-interfaces/Mastermind/.

Planware,http://www.kestrel.edu/HTML/projects/SciNapse. http://www.scicomp.com/about/

arpa-plan2/ .technology.h tm1.

Michael R. Lowry

AUTOMATION

A u t o m a t i o n is the conversion of a work process, a pro-cedure, or equipment to automatic rather than humanoperation or control. Automationdoesnot simplytransfer human functions to machines, but involves adeep reorganization of the workprocess, during whichboth the human ndhemachine functions areredefined. Early automation relied on mechanical andelectromechanical control devices; during the last 40years, however, the computer gradually became theleading vehicle of automation. Modern automation isusually associated with computerization.This article examines the major phases of historicaldevelopment and social and economic aspects of in-dustrial automation, focusing on the computeriza-tion of production, engineering, and managerial asks.Other areas of computer-basedautomation includeadministrative applications (4 .v.),communication viaelectronic mail (q.v.), anking applications, medicalapplications (q.v.),nd library automation ( see DIGITALLIBRARIES).Phase I: Mechanization andRationalization of LaborThe mechanization of machine tools forproduction be-gan during the Industrial Revolution at the endof the18th century with the introduction of the Watt steamengine, the Jacquard loom, the lathe, and the screwmachine. Mechanization replaced human or animalpower with machine power; those mechanisms, how-ever, were not automaticbut controlled by factoryworkers. The factory system, with its large-volume,standardized production, and division of labor, re-placed the old work organization, where broadlyskilled craftsmen and artisans produced small quan-tities of diverse products. In he late 19th centuryFrederick W. Taylor rationalized the factory system byintroducing the principles of scientific management.Heviewed the body of each worker as a machinewhose movements had o beoptimized n order ominimize time required to complete eachask and thusincrease overall productivity. Scientificmanagementstrictly separated mental work frommanual labor:


2/5

AUTOMATION 123

workerswerenot o think but o follow detailedinstructionsprepared for them by managers. Therationalized factory system gave birth to a new man-agerial class and large clerical bureaucracies. TheTaylorist principles served as a basis for Henry Fordssystem of mass production. In 1913 the Ford MotorCompany introduced a moving assembly line, drasti-cally cutting assembly time. The assembly ine imposeda strict order onproduction by forcing workers tokeeppacewith the motion of the conveyor belt. Massproduction relied on he standardization of compo-nents and final products and routinization of manu-facturing and assembly jobs. The Ford assembly linebecame a symbol of efficiency of American manufac-turing; for workersand social critics, however, itepitomized the monotony and relentless pressure ofmechanized work.Phase II: Automation of ProductionIn 1947 the Ford Company brought the term auto-mation into wide circulation by establishing the firstAutomation Department,harged with designingelectromechanical, hydraulic, andpneumaticparts-handling, work-feeding, and work-removing mechan-isms to connect standalone machines and ncrease therate of production. In 1950 Fordput into operationhefirst automated engine plant. Although early auto-mation was hard,or fixed in the hardware, and didnot involve automatic feedback control, this conceptprovoked great public enthusiasm for unmanned fac-tories controlled by buttons that push hemselves,as well as causing growingconcern about therospectsof mass unemployment.To meet U S Air Force demands for a high-performancefighter aircraft whosecomplex structuralmemberscould not be manufactured by traditional machiningmethods, a technology of Numerical Control (NC) ofmachine tools was developed in the early 1950s. NClaid foundation for programmable, or soft,automa-tion, in which the sequence of processing operationswas not fixed ut could be changed or each ew prod-uct style. Commercial NC machines for batch produc-tion appeared in the mid-1950s. Designed to militaryspecifications, early NC equipment proved too com-plex and therefore unreliable, as well as prohibitivelyexpensive, and was applied mostly in the state-sub-sidized aircraft industry.The abstract, formal approach of NC, based on mathe-matical modeling of the machining process, supersededthe record-playback technique of direct machine imita-tion of workers actions. While the record-playbackapproach relied on the skill and discretion of theworker, NC technology allowed engineers and man-agers to exercise greater control over the productionprocess.

Phase Ill: Computer-Aided Manufacturing(CAM)The first industrial applications of digital computersoccurred in the electrical power, dairy, chemical, andpetroleum refinery industries for automatic processcontrol. In 1959, TRW installed the first digital com-puter designed specifically for plant process control atTexacos Port Arthur refinery. Early applications wereopen-loop control systems: gathering data from mea-suring devices and sensors throughout the plant, thecomputers monitored technological processes, per-formed calculations, and printed out operator guides;subsequent adjustments were made by human opera-tors. In the 1960s closed-loopfeedback control systemsappeared. These computers were connected irectly toservo-control valves and made adjustments automati-cally (see CYBERNETICS).In he ate 1960s, with the development of timesharing ( 4 . v . ) on large mainframe computers ( q . ~ . ) ,standalone NC machines were brought under DirectNumerical Control (DNC)of a central computer.DNCsystems proved vulnerable to frequent failures due tomalfunctioning of the central computer and the nter-ference of factory power cables with the data trans-mission cables of the DNC system.With the introduction of microprocessors ( q . v . ) n the1970s, centralized DNC systems in manufacturingwere largely replaced by Computer Numerical Con-trol (CNC) systems with distributed control, in whicheach NC machine was controlled by its own micro-computer. This blending of information and produc-tion technologies produced a new breed of machinist-programmer who could operate CNC equipment bygenerating and debugging NC programs, thus break-ing down he traditional distinction between white-collar and blue-collar jobs.Robotics combined the techniques of NC and remotecontrol to replace human workers with numericallycontrolled mechanical manipulators. The first com-mercial robots appeared in the early 1960s.Robotsproved very efficient in performing specialized tasksthat demanded high precision or had to be done inhazardous environments. To approach the humanevelof flexibility, robots were supplied with sophisticatedtechniques of feedback, vision and tactile sensors, rea-soning capabilities, and adaptive control. In the 1980sindustrial applications of robots slowed down, as theirincreasing complexity resulted in growing costs andinsufficient reliability.Hierarchical NumericalControlSystems ombinedDNC and CN C features: they linked each standalonecomputer controller to a central computer that main-tained a large library of CNC programs andmonitored


3/5

124 AUTOMATION

production. This approach aspired to replace thehuman operators xpertise by engineering knowledgeformalized in CNC programs. In suchsystems, humanoperators generally no longer programmed CNCequipmenton heshop floor, andproductionwasbrought under remoteupervision of a central manage-ment-controlled computer.Flexible Manufacturing ystems (FMS) combinedD NC equipment with machinesor automated loading,unloading, and transfer of workpieces. These systemspermitted varying process routes andsequences ofoperations, allowing automatic machining of differentproducts insmall batches in the same system. Cen-tralized FMS have often proved too complex, however,and they are increasingly ubdivided into smallerflexible manufacturing cells (FMC) hat include severalCNC machines, robots, and transfer devices controlledby a single computer, the cell controller.Phase IV: Automated EngineeringIn the 1960s large aerospace manufacturers, such asMcDonnell-Douglas and Boeing, developedproprietarycomputer-aided design (CAD) systems, which providedcomputer graphics (q.v.) tools for drafting, analyzing,and modifying aircraft designs. In 1970 Computer-Vision Corporation introduced the irst complete turn-key commercial CA D system for industrial designers,which provided all the necessary hardware and soft-ware in one package. In the 1970s, combined CAD/CAM systems emerged which used the parameters ofa geometrical model created with the help of CAD togenerate programs for CNC machine tools and developmanufacturing plans and schedules. While CADsystems are often packaged and standardized, CAM(Computer-Aided Manufacturing) pplications tend tobe industry-specificand proprietary. With the introduc-tion of Computer-Aided EngineeringCAE) systems forstandard techniquesof engineering analysis, the wholerange of engineering tasks-from conceptual design toanalysis to detailed design to drafting and documenta-tion to manufacturing esign-became automated. Thedistinction between blue-collar and white-collar jobswas further blurred, as engineers, clerks, and managersbecame integrated in an automatedoffice.Phase V: Automated ManagementAmong the earliest applications of information tech-nology was the automation of information-processingtasks. The first stored-program digital computer pur-chased by a nongovernment customer was UNIVAC( q . v . ) ,installed by GE in 1954 to automate basic trans-action processing: payroll, inventory control and mate-rial scheduling, billing and order service, and generalcost accounting. Large clerical bureaucracies, which

processed huge amounts of data generated in massproduction and mass marketing, became arimary tar-get of automation and job reduction in the 1960s and1970s. By 1970 the profession of bookkeeperwasalmost completely eliminated in the USA. In the mid-1960s the first management-information systems (MIS)appeared, providing management withdata, models ofanalysis, and algorithms for decision-making; even-tually they became a standard tool for operation con-trol, management control, and strategic planning.Phase VI: Computer-IntegratedManufacturing (CIM )In the late 1980s an integration of the automated factoryand the electronic office ( 4 . v . ) began. CIM combinesflexible automation robots, numerically controlledmachines, and flexible manufacturing systems), CAD/CA M systems, and management-information systemsto build integrated production systems that cover thecomplete operations of a manufacturingirm, includingpurchasing, logistics, maintenance, engineering, andbusiness operations. CIM emphasizes horizontal linksbetween different organizational units of a firm andprovides the possibility of sharing data and computingresources, making it possible to break the traditionalinstitutional barriers between departments andcreateflexible functional groups to performasks more speed-ily and efficiently.Social and Economic Dimensionsof AutomationViews of automation range between two extremes-unabashed optimism and utmost pessimism. The opti-mists believe n a technological utopia, an imaginedbright future in which machines will relieve people ofall hard work and bring prosperity to humankind. Thepessimists view machines as instruments of subjuga-tion and control by a ruling elite, argue that automa-tion leads to the degradation of human beings, anddepict the future as a grim technological dystopia. Bothsidesview automatic technologyas an autonomousforce determining he direction of human history.Automation itself, however, is a social process shapedby various social and economic forces. This processmay take various directions and mayhavediverseconsequencesdependingon hesocioeconomicandorganizational choices made during automation.The Productivity ParadoxWhile productivity in major industries in the USA rosesharply duringproductionautomation in the 1950sand 60s, its growth has slowed significantly since the1970s, precisely at the time of widespread computer-ization of the factory and the office. The link between


4/5

AUTOMATION 125

computerizationand productivity remainsproblem-atic. The advantages mostommonly associatedwith computer-aided manufacturingnclude increasedproduction rates,better product quality, more efficientuse of materials, shorter lead times, reduced workhours, and improved work safety-all factors leadingto higher productivity. Among its main disadvantages,analysts usually cite the high cost of designing, build-ing, and maintaining computerized equipment;ulner-ability to downtime; elatively low flexibility omparedwith humans; and worker displacement and emotionalstress-all leading to lower productivity. It is particu-larly difficult to compare directly productivity beforeand after computerization, since it brings with it notmerely technological, but also organizational changewhich transforms the entire nature of production andbrings with it the most benefits and losses.As manufacturerswhontroduced omputer-aidedmanufacturing systemsffirm, the largest payofffrom computerizationcomesnotfrom speeding upold operations but rommaking work organizationmore flexible and efficient. On theotherhand, ifcomputers are used to conserve old inefficient organi-zation, computerization can only accelerate negativetrends. As John Bessant has remarked, Whenou puta computer nto a chaotic factory the only thing you getis computerized chaos (quoted n Ayres, 199 1-1 992,Vol. 4 , p. 94). Most successful manufacturers stream-line operations before computerization, following thedictum, Simplify, then automate! Efficient compu-terization takes far more han merely nstalling acomputer: it requires changes in the entire workstyle.

Worker Displacemen t, Skill, and WorkingConditionsA leading concern among workers, labor leaders, andsocial critics has been the issue of worker displace-ment-a loss of work, transfer to a different job, orgeographic dislocation-due to automation. Suchate-gories as welders, carpenters, insulators, machinists,and clerical staff have been most heavily affected. Atthe same time, automation creates new highly-skilledjobs in programming, operating, and maintaining com-puterized production machinery. Workers needxten-sive retraining programs, however, to prepareor suchjobs.Another risk ishe dangerof employees losing essentialworking skills as work becomes increasingly mediatedby the computer. With automation, the worker hasgone through a series of transformations-from a di-rect producer of goods and services to the operator ofproduction quipmentoheprogrammer of thecomputer that operates and controls that equipment.Engineering changed romhands-on tinkering with

machinery to the use of standard design and analysisprocedures that tell the computer how to design andbuild a needed par t. Management evolved from directsupervision of labor to management by numbers,based on numerical ata reports and pre-programmedcomputer algorithms for decision-making. Whenoperators must step in and take control in case of anemergency at an automatically controlled nuclearpower plant, would they possess the necessary skillsif their training and daily experience mainly concernedwork with a computerized ontrol system?Because of the high cost of downtime, efficient main-tenance and fast repairs become crucial in automatedproduction, which places a great burden of responsi-bility and tight time constraints on maintenance andrepair crews. Computerized equipment can besed toenhance the flexibility of work organization, leavingone in charge of planning ones work time, but it mayalso be used to mpose a strict and inflexibleworkregimeon factory and office workers byloselymonitoring their performance. As a result, automationcan make work either easier or more exhausting andstressful, depending on he type of work organization.

Technocentric vs. Human-CenteredApproachesHistorically thepredominantapproach oautoma-tion has been echnocentric: a goal of automation is toreduce and ultimately entirely eliminate human par-ticipation in production and eventually arrive at anunmanned factory. From this standpoint, workers areseen as a source of potential errors, disturbance, andunreliability; on the other hand, automatic machineryisviewedas inherently more precise, reliable, andcontrollable. The technocentric approach extends theprinciples of Taylorist work organization to moderninformation-processing and production systems. It isbased on further subdivision of labor, with more com-plex and intelligent tasks trusted to flexible computersystems and simpler tasks left to low-skilled workerswho assume aesidual role. Skill gradually passes rompeople tomachines,and control functions are alsotransferred in the same direction.The technocentric approach faces a fundamental para-dox: it aspires to replace human skill with highly flex-ible computerizedmachinery,but this machineryrequires even more human skill to operate, maintain,and repair it . Instead of freeing production fromthe human element, automation only increases theimportance of highly qualified, versatile, and motivatedworkers. Accidents at the nuclear power plants at ThreeMile Island and Chernobyl testify hat automationdoesnot eliminate the possibility of human error; it onlymakes this error morecostly.


5/5

126 AUTOMATION

The Taylorist logic of seeking productivity by accel-erating the pace of work may not apply in a compu-terized workplace. With computerization, companiesdo not simply automate, but informate their opera-tions. Computer-based control of production becomesan information-processing task; workers urn intoanalyzers of information rather than simple machineminders. Improving thequality of this analysis, insteadof speeding up workers movements, becomescrucialproblem of automation.An alternative approach aspires to change the work-force from being part of the manufacturing probleminto part of the solution. Instead of taking skills,responsibility, and control away from the worker andabsorbing hem into themachine,human-centeredCIM systems mobilize the intellectual resources of allemployees. Leading Japanese ompanies,uch asMatsushita and Toyota, achieved much greater pro-ductivity gains from automation than their Americancompetitors by decentralizing control and reorganizingthe factory layout into production islands controlled bysemi-autonomous multi-skilled teams responsible forall operations. Reversing the Taylorist trend of subdivi-sion of labor, the human-centered approachntegratesfunctions and skills in flexible teams, where workerscan rotate obs and choose the optimal order and paceof work. Instead of being forced to follow instructionshanded to them rom above, workers are motivated toplay a greater role in decision-makingby programmingCNC equipment on the shoploor. In the ate 1960s andearly 1970s only a handful of American companies,such as Procter & Gamble,Cummins Engine, andGaines Foods, realized that greater productivity did notcome automatically withmore sophisticated equip-ment but required profound organizational change.In1974 Volvo built a highly productive plant atKalmar, Sweden, which implemented the sociotech-

nical systems approach, elaborated in Britain. Basedon group assembly instead of a conventional assemblyline, this newesignave workersmore initia-tive, flexibility, and control over product quality. In the1980s major American manufacturers began experi-menting with workernvolvement in decision-making,a recent example being GMs Saturn project. Thehuman-centered approachfinds a source of productiv-ity inmore efficient utilization f human abilities, ratherthan in the utopian efforts to eliminate people fromproduction.

Bibliography1967. Bright, J. R. The Development of Automation, inTechnology in We stern Civilization, Vol. I1 (eds.M . Kranzberg and C. W . Pursell, Jr.), 635-654. New York:

Oxford University Press.1984. Noble, D. F. Forces of P roduction: A Social History ofIndustrial Automation. New York: Knopf/Random House.1988. Zuboff, S. In the Age of the Smar t Ma chin e:The Future of Work and Power. New York: Basic Books.1989. Forester, T. (ed.) Computers in the Hum an Context:Inform ation T echnology, Productivity, and People.

Cambridge, MA: MIT Press.1992. Ayres, R. U., Haywood, W ., and Tchijov, I . (eds.)Computer Integrated Manufacturing. 4 Vols. London:

Chapman & Hall.1994. Allen, T. J., and Scott Morton, M . S. (eds.) InformationTechnology and the Corporation of the 1990s: ResearchStudies. New York: Oxford University Press.

Value Conflicts and Social Choices, 2nd Ed. San Diego, CA:Academic Press.Consequences of C omputer ization. Princeton, NJ: PrincetonUniversity Press.

1996. Kling, R. ed.) Computerization and Controversy:

1997. Rochlin, G. I. Trapped in the Net: The Unanticipated

Slava Gerovitch

AUXILIARY MEMORYSee MEMORY:AUXILIARY.

Documents

Gerovitch Automation