Robotsfor

nuclearpowerplantsby Taylor Moore

Odex, a walking robot with multiple potential uses. (Credit: Odetics).

In the USA, robotics technology used at the TMI-2 cleanup and at other nuclear plantshas prompted interest and shaped research on how robots might best be used

Once confined to the- pages of sciencefiction, robots h.ive dramatical ly capturedthe attention of the public and the indus-t r i a l business community in recent years.Many observers view robots as a h a l l -mark of neoindustrialization, breathingrenewed economic vigor and com-petitiveness into depressed industriesthrough improved productivi ty andreduced labor costs.

At the same time, however, workersoften respond wi th apprehension to themental image ot a robot performing atask that formerly required a human.The social implications of the robot-i/ation of American industry will surelybecome of more concern to workers,managers, and policymakers alike asmore robots enter the industrial work-place.

According to the Robotics IndustriesAssociation, only f>3()() robots had beendelivered in the United States by theend of 1983; most of those had been in-

stalled since 1976. But the force of tech-nologic change and the pressure olinternational economic competitionpromise an accelerated pace of robotdeployment in the years ahead. Someexperts predict that as many as 10(1 000robots may be at work in this countryby 1990—one-tenth of the total numberprojected worldwide.

For most industries in which robotshave been or are expected to be appliedin significant numbers, such as auto-mobile production, metalworking, andmachinery manufacture, the incentivesto robotize relate directly to preservingor recapturing competitive advantagethrough lowered uni t costs of produc-tion and improved product quality. Butfor some industries, the attraction of ro-bots is their potential to work in ha/ard-ous environments, thereby reducing thehuman risks associated with the work.

The electric uti l i ty industry is onesuch industry. Although ut i l i t ies are not

viewed by most industrial robot manu-facturers as a significant potential mar-ket, special-application robots are underdevelopment for performing inspectionand maintenance tasks inside nuclearpower plants, where radiation levels,heat, and humid i ty either rule outthe presence of human workers orseverely l imi t their abi l i ty to work. Formany of these tasks in a nuclear plant ,robots would be a welcome addit ionto the workforce, freeing humans fromsome of the more onerous and discom-forting jobs and, possibly, permittingcertain tasks to be performed while aplant remains on-line, thus avoidingcostly plant downtime tor inspection ormaintenance.

Some of the robots under develop-ment for u t i l i ty applications representthe state of the art of robotics engineer-ing, and the related research effortscould pioneer advances that have broadapplication to other industries. EPKI

Mr Moore is on the staff of the EPRI Journal, from which this article is reprinted. EPRI is the Electric Power Research Ins t i tu te in theUSA, P.O. Box 10412, Palo Alto, California 94303. EPRI technical contributors to the article were Floyd Gelhaus. Michael Kolar.Thomas Law, Adrian Roberts, and R.K. Winkleblack.

IAEA BULLETIN, AUTUMN 1985 31

Nuclear power and electronics

has several current projects aimed atevaluating the technical and economicpotential for robot applications in utilityoperations and at translating the under-standing gained from these efforts tothe utility professionals who have workaplenty waiting for robots that provereliable and cost-effective.

Such research is necessarily longrange. The robotics industry, fewer than20 years old by the broadest definition,remains in its infancy, awaiting sub-stantial technical advances in visionsystems, miniaturization, and computercontrols before truly economic, versa-tile, and powerful robots are common-place items of commerce. But R&Dsuccess with robots in recent years sug-gests that such machines will emergefrom the laboratories and enter thecommercial market before this decadeis over. EPRl's research in robotic appli-cations, at least in part, is intended toensure that when that day arrives, util-ities will have a clear understanding ofthe work robots can do for them andwhether it makes economic sense toput them to work.

Robots for nuclear plants

The use of remotely operated androbotlike equipment to protect nuclearworkers in high-radiation areas is notnew. John Taylor, an EPRI vice presi-dent and director of the Nuclear PowerDivision, divides robotic equipment innuclear applications into two broad cat-egories: single-purpose devices withlimited ability to perform different op-erations, and reprogrammable, multi-purpose robots with some degree ofcomputer-based artificial intelligence.

"I think the first category has reacheda reasonable level of maturity," saysTaylor At EPRI's Nondestructive Evalu-ation (NDE) Center and among reactormanufacturers, nuclear service con-tractors, and some utilities, these typesof devices are in use today for suchtasks as pipe cutting, welding, steamgenerator tube inspection and repair,and ultrasonic scanning of pipe sectionsfor crack detection. "These deviceshave proved to be absolutely essential;we simply could not get some jobsdone without them," adds Taylor.

Robots in the second category, thosewith sufficient computer-based intel-ligence to support a variety of applica-tions, "have a long way to go," inTaylor's words, before they can dem-onstrate significant practical benefit innuclear plant operations. But, as Taylor

adds, such robots are under develop-ment, and their initial trials are ex-pected to provide valuable insight totheir ultimate potential.

Soon after remote manipulator armswere developed for use in hot cells andfuel reprocessing activities, an armmounted on a transporter with camerasand lights made its debut in the 1950sat the government's Hanford nuclearfacility in Washington State. Developedby Westinghouse Hanford Co., the re-motely controlled transporter vehiclewas dubbed Louie after a technicianscrawled the nickname on the robot'sarm. Louie has proved to be a versatileand long-lived workhorse and is still inuse today.

Some fundamental aspects of howthis equipment is applied distinguishrobotic equipment for nuclear plantapplications from the more widely fa-miliar industrial robots—those fixeddevices that typically are employed forpick-and-place operations or otherhighly repetitive tasks.

In many industrial applications ofrobots, the objective is to replace humanworkers with machines that are moreproductive, efficient, and accurate. Butfor nuclear applications, the objective isnot so much to replace workers as it isto extend their presence—for example,to project their reach into areas of a nu-clear plant where the thermal or radia-tion environment prohibits or limits ahuman presence.

"In contrast to most robotic appli-cations, we want to keep man in theloop, rather than replace him, to ob-serve the work, make decisions, andcontrol the robot," according to R. K.Winkleblack, an EPRI project managerin the Nuclear Power Division. "Strictlyspeaking, the devices we are looking atare remote-controlled equipment, nottrue robots," adds Winkleblack

Improving availability

The economic motivation to use robotsfor nuclear plant inspection and main-tenance is centered on their potentialfor improving plant availability; a by-product is the potential for reducingthe occupational radiation exposure(ORE) of plant personnel.

Many inspection and maintenancetasks can only be done when the re-actor is shut down because radiationlevels under operating conditions wouldbe too high even for humans fully out-fitted in protective clothing. These jobsare usually deferred until scheduled re-

fueling outages to minimize plant down-time. They thus can become part of thecritical path of activity needed to bringthe plant back into service.

Delays are critical to plant availabil-ity, as well as costly Purchased replace-ment power to substitute for the outputof a 1000-MW|e reactor costs an aver-age of $500 000 a day. Robots poten-tially could contribute to improvedplant availability by avoiding delays inscheduled outages and handling sometasks while the reactor is operating.

Nuclear workers are currently limitedby federal regulation to no more than 3rem per quarter-year or an annual totalof 5 rem* This means that for manyroutine jobs large numbers of workersmust be assigned a small portion ofthe work because each will quicklyreach the ORE limit and must then berestricted to nonradiation areas untilthe next quarter. Consequently, utilitiesare forced to employ significant num-bers of transient workers, or so-calledjumpers—temporary personnel whomove on to other jobs after receivingthe ORE limit.

According to a Nuclear RegulatoryCommission (NRC) estimate, everyman-rem of personnel exposure has avalue to utilities of $1000, althoughsome utilities assign a value of as muchas $5000 a man-rem. Some types ofwork, such as health physics surveysand inspection of primary reactor cool-ing systems, can involve radiationfields of several hundred rads an hour.

Utilities may face even tougher ORElimits in the future. In addition to guide-lines that call on utilities to reduceOREs to levels "as low as reasonablyachievable," NRC for several years hasbeen studying proposals to reduce theORE standards; such a developmentcould have a multiplicative effect onutility costs for personnel exposure.

Feasibility studied

EPRI and NRC have both sponsoredpreliminary assessments of the poten-tial for applying robotics in nuclearpower plants. NRC, motivated primar-ily by the objective of reducing person-nel radiation doses, looked mainly atsurveillance and inspection tasks in astudy performed by Remote Technology

* In international usage, the rem has beenreplaced by the sievert in accordance withrecommendations of the InternationalOrganization for Standardization. Onesievert corresponds to 100 rem.

32 IAEA BULLETIN, AUTUMN 1985

Nuclear power and electronics

Corp. EPRI's analysis, conducted byBattelle, Columbus Laboratories, fo-cused on maintenance activities andattempted to identify potential avail-ability improvements, as well as oppor-tunities to reduce radiation exposure.

Each study attempted to quantify thecost in ORE and man-hours of a varietyof jobs that a robot system might be ca-pable of performing; the costs werethen compared with those of the robotand its associated support systems andpersonnel

Surveillance and inspection tasksevaluated in the NRC study range fromdetection of steam or water leaks, verifi-cation of valve positions, and readingof gages to measurement of radiationlevels in components and various meth-ods of sampling to detect contamina-tion. The EPRI study surveyed 22 tasksthat are performed routinely or duringrefueling, including control rod drivemaintenance, steam generator tuberepair, and repair or replacement ofvarious pumps and valves.

Although the scope of activities ana-lyzed were different, both studies con-cluded there were potentially signi-ficant net positive economic benefits ofapplying robots in nuclear plants TheNRC study, based on application of acost-benefit methodology to two exist-ing plants, concluded that commerciallyavailable robotic technology can beretrofitted into existing plants and willreduce both radiation exposure toworkers and plant operating costs.

The NRC study cautioned, however,that benefits can differ significantlyamong plants because of dissimilar de-sign factors and operating histories.The report encourages utilities to per-form plant-specific cost-benefit anal-yses, including consideration of allcosts of personnel entry into radiationareas, to determine whether roboticapplications for such inspections areeconomical.

In the Battelle study for EPRI, poten-tial maintenance applications werescreened to identify candidate taskscommon to many nuclear plants thataccount for a significant share of main-tenance costs and are amenable to per-formance within the limits of currentrobotic technology. Follow-on cost-benefit analyses were performed for theapplication of robots to reactor cavitycleaning, health physics surveys, andflange unbolting/rebolting. Despite theamenability of these tasks to currentrobotic technology, it was concluded

that none could be performed robot-ically without further technology de-velopment.

Using the net present value method,Battelle researchers found that robotsfor reactor cavity cleanup and main-tenance bolting activities would payback in less than one year, while healthphysics survey applications would payback in about three years. The resultswere then tested with a range of valuesfor outage time and radiation exposurecosts.

Even with the lowest values ($700per man-rem exposure and $300 000 aday outage time costs), robotization ofmaintenance bolting would pay backin slightly more than one year, whilehealth physics survey tasks would re-quire less than four years to pay back,the study found. Overall, the study in-dicated cost savings ranging from$100 000 to $1 million in net presentvalue per robot, with the purchaseprice for each robot projected at under$200 000.

An important caveat noted by theBattelle researchers, however, is thelimited availability of commercial ro-botic equipment geared specifically tonuclear applications. Because the nu-clear industry has not been a majormarket for robot manufacturers, thebusiness has generally been left tosmaller entrepreneurial firms that canadapt robotic equipment for low-volumeapplications.

The nuclear industry thus needssome way to fund these developmentsor to attract entrepreneurs who arewilling to financially shelter the tech-nology during its demonstration phase,the study points out. This is in contrastto the situation in Japan, where a coop-erative relationship between utilitiesand vendors has led to a more unifiedapproach

Prototype development

The core of EPRI's research in roboticsis its participation in the developmentand testing of several prototype robotsystems that could be forerunners ofcommercially available machines Someof these robots could be used as trans-port vehicles to carry other roboticequipment, such as a flange unbolteror a steam generator tube-repair robot,into a high-radiation area, set up thesmaller device at work, and then mon-itor its activity. Some, on the otherhand, may be less capable of doing de-manding labor, but could be used as in-

telligent master robots, controlling thework of stronger drones.

Several robot prototypes are makingtheir debut in the recovery and cleanupof the damaged Three Mile Island Unit2 nuclear plant in Pennsylvania, the siteof a March 1979 loss-of-coolant accidentthat destroyed much of the reactor coreand left large areas of the reactor con-tainment building inaccessible to hu-mans. Remote inspection has shownradiation fields as high as 3000 rad/h insome areas of the containment.*

According to Adrian Roberts, a seniorprogram manager in EPRI's NuclearPower Division and manager of itsTMI-2 information and examinationprogram, the TMI cleanup effort hasbecome a particularly strong spur to ro-botic equipment development. "At TMIwe have a challenge for robotics that ishere and now, some of the jobs simplycan't be done other than remotely. Andbecause we can't wait for the ultimaterobot, we're taking advantage of workfrom a number of areas to developrobots that will get the jobs done. Ifrobots are shown to be feasible for cer-tain jobs at TMI, they can be applied atother nuclear plants."

Robots, in fact, have been tried atvarious times at TMI since the accident.In August 1982, a 25-lb (11-kg), remotely-controlled, tracked, tanklike vehiclesupplied by DOE and called SISI (forsystem in-service inspection) was usedto photograph and obtain radiationreadings in areas surrounding theplant's water makeup and purificationsystem. The water system's filters arehighly contaminated with fission prod-ucts from the primary core cooling sys-tem. The following spring, a six-wheelremotely-controlled device dubbed Fredwas outfitted with a high-pressurewater spray and used to decontaminatethe walls and floor of a pump cubicle inthe auxiliary building basement. Fredweighs in at 400 Ib (181 kg); its mechan-ical arm can lift 150 Ib (68 kg) and ex-tend to a height of 6 ft (1.8 m).

The venerable Louie from Westing-house Hanford has been brought toTMI to perform radiologic character-ization during decontamination of thewater purification system. Officiallyknown as the remotely controlled trans-porter vehicle, Louie will be used tomonitor radiation levels as the demin-

* In international usage, the rad has beenreplaced by the gray. One gray correspondsto 100 rad.

IAEA BULLETIN, AUTUMN 1985 33

Robots atTMI-2

Cleanup and recovery work at the damaged TMI 2 reactor in Pennsylvania presents a unique challenge for the application ofrobotics technology. Two remotely operated manipulators called Fred and SISI have already seen service in surveillance anddecontamination tasks. The RRV, nicknamed Rover, has been assigned the job of inspecting the contaminated basement ofthe reactor containment building. A remote scabbling machine has been developed to remove contaminated layers fromconcrete floors. Louie, specially modified for the TMI work, is slated to monitor radiation levels as the plant demineralizertank is decontaminated. Rosa, a versatile remote manipulator arm, has been proposed to lend a hand in defuelling the TMI 2reactor core.

Rosa(core detuelingi

Remote scabbier(decontamination)

Fred(decontamination!

34 IAEA BULLETIN, AUTUMN 1985

Nuclear power and electronics

erali/er rvsins in the water system an1

flushed nut. I hough the robot's nearlyKXXl-lb (454-kg) lifting strength willnot be needed in this operation, itsradiation-hardened television cameraswill get a workout near the demineral-i/er tank, which has a contact readingot 3000 rad/h.

Perhaps the most ambitious effort todate to applv robotics in the I'Ml clean-up has been the EPRI-SUpported devel-opment by Carnegie-Mellon Univer-sity's (CMU's) Civil Fngineering and(. onstruction Robotics Laboratory ot theremote reconnaissance vehicle (KRV) to

probe the basement ol the reactor con-tainment building. The basement level,where no human has entered in overfive years, remains highly contami-nated with the radioactive sludge lefttrom some 600,000 gallons (2270 m1)of water, including primary coolingwater, most ol which has since beenpumped out.

The RKV, nicknamed Rover by GI'UNuclear Corp., the operating utility atI'M I. has been assigned the task of en-tering the dark and damp basement bvcrane hoist, inspecting the scene withits three television cameras, and sur-vwing the area radiologically withseveral on-board detection instruments.

The six-wheel, 1000-lb KRV, devel-oped in a cooperative effort involv-ing EPRI, CMU, C,PU Nuclear, DOE,and the Hen l-'ranklin Partnership inPennsylvania, was designed by CMU'sWilliam Whittaker, an assistant pro-fessor of civil engineering and directorof the robotics laboratory. It features aninnovative on-board umbilical reelingsystem designed to permit the vehicleto negotiate obstacles without draggingthe umbilical. A stainless steel framemounted atop the transporter basecarries the umbilical reel, cameras,monitoring instruments, and controlsystems. The vehicle has also been de-signed for quick decontamination withwater sprav after it is removed tromwork areas.

A two-person crew controls the RKVfrom a ionsole equipped with televisionmonitors that is located a safe distanceaway from the ha/ardous area (at TMI,

this distance is over 500 ft. or 150 m);one person steers the craft and manipu-lates the cameras while the other oper-ates the umbilical reel learns of oper-ators practiced maneuvering the RKVfor several months along an improvisedobstacle course in the adjacent turbinebuilding (the staging area for much

IRIS - for Industrial Remote Inspection system — is a general-purpose robot forhazardous environments. (Credit: EPRI)

of the cleanup work) in preparationfor lowering it into the containmentbasement.

The RRV is the first of three similarremote vehicles to be developed underthe joint TMI recovery program. An im-portant feature of the design is that theframe mounted on the chassis can beremoved and other equipment addedto the transporter. The second RRVbase vehicle, modified by 1'entek, Inc.,HPKI's site contractor at TMI, is out-f i t ted with a pneumatically poweredscabbling machine and vacuum systemfor removing the contaminated toplayer of concrete from floors in partsot the reactor building.

A third RRV remains at CMU's robot-ics laboratory for future developmentefforts. Other tasks proposed for futuremodifications of the prototype RRV in-clude collection of liquid and sludgesamples from the containment base-ment, collection of concrete core sam-ples from the floor and walls, and someminor structural dismantling.

"At TMI the interest is in workingvehicles with high strength, reliability,and mobility," explains Whittaker, theRRV's designer. "The challenges at TMIare very physical and active, and theequipment that will meet those chal-lenges will be similarly physical andactive. But there is certainly no one

machine that will do it all, so we arelooking at the evolution of a family ofthese things. One mode might be afully configured RRV to supervise theactivity of a drone that would earn'tools only. Another possibility is a min-iature version of the RRV that wouldoperate radio-remote from the mothership."

Clearly, robotic equipment is provingto be a valuable tool in the TMI recov-ery effort. Other applications of robotsat the site are also planned. A manipu-lator arm built bv Westinghouse ElectricCo. and known as Rosa (for remotelyoperated service arm) has been pro-posed for use in the defueling of the1 Ml reactor core, tentatively plannedfor next year. Rosa, which can alsooperate underwater, is already knownamong some utilities operating pressur-ized water reactors for its ability to auto-matically inspect and repair steam gen-erator tubes after it is mounted on thesteam generator by service personnel.

Waiting in the wings

In addition to the robots that have beendeployed at TMI, EPRI is evaluatingtwo other prototype devices that couldprove useful in nuclear plant environ-ments. These machines could becomecousins of the TMI machines in the ro-bot family that Whittaker envisions.

IAEA BULLETIN, AUTUMN 1985 35

Nuclear power and electronics

One of these, produced by AdvancedResource Development (ARD) Corp.,is known as an industrial remote in-spection system (IRIS). Designed as ageneral-purpose surveillance and inspec-tion robot for hazardous environments,IRIS is a relatively small (compared withthe RRV) battery-powered, trackedtransporter that can be equipped withoptical, audio, and environmental sen-sors; manipulators; and communica-tions and control subsystems.

The 200-lb (91-kg) IRIS features aunique high-frequency wireless com-munication system, specifically de-signed to operate in an environmentcluttered with physical barriers, as wellas with signal interference, which al-lows it greater mobility and range thanmost robots developed to date. A tele-scoping arm and a three-dimensionaltelevision system with zoom lens andmicrophones mounted on a pan-tilt padbring the current payload to 70 Ib (32kg). Eventually, IRIS will contain somelimited on-board intelligence, enablingit to retrace its steps backward evenif normal control signals are lost orblocked by interference.

According to Floyd Gelhaus, an EPRIprogram manager who is evaluatingIRIS and other robots for potential nu-clear applications, the current ARDdevice has been designed strictly asa remote surveillance vehicle. "Its abil-ity to do robust tasks is limited," saysGelhaus, "but the mobility and unteth-ered configuration of the transporter,with its ability to carry various pay-loads, make it a valuable member of arobotics staff."

Gelhaus plans to have technicians atEPRI's NDE Center put IRIS through itspaces before taking the robot into a re-cently constructed, nonradioactive plantenvironment. Duke Power Co. has agreedto host the testing activities at its newCatawba nuclear unit. The final step willbe to test and evaluate the device in anoperating plant.

Gelhaus is also considering possibleapplications for what is probably themost advanced robot developed sofar—a six-legged, free-walking machineknown as Odex. The Odex prototype,built by Odetics, Inc., "represents a re-markable breakthrough in its strength-to-weight ratio," comments Gelhaus,as it can l i f t more than five and a halftimes its 370-lb (168-kg) weight. Almostany other robot can heft little morethan one-twentieth of its weight. "Withthat kind of power, there are a lot of

potential applications," adds Gelhaus.Odetics has demonstrated Odex onvideotape around the country, includ-ing scenes of it lifting the end of a com-pact pickup truck.

Because each of Odex's articulators,or legs, uses its own 'microprocessor,with a seventh computer coordinatingoverall movement, complex maneuversare possible under the control of eitheran operator or a remote computer. Themachine can pirouette 360° while simul-taneously advancing in any direction.Its jointed legs permit it to assume sixdistinct profiles, ranging from a narrowstance for negotiating tight doors to alow squat. Odex is outfitted with twinTV cameras for visual transmission.

"Odex is a breakthrough in the stateof the art," says Gelhaus, "but it willtake some careful research to defineapplications for it in a nuclear plant."EPRI's work with Odetics has led toconceptual design modifications thatwill enable Odex to negotiate a powerplant's internal obstacle course.

Future development

Technologically, Odex may be close tothe fully autonomous, intelligent robotthat researchers say would representthe ultimate marriage between machineautomation and the developing field ofartificial intelligence Its ability to ma-neuver around or over obstacles underthe guidance of a remote operator ap-proaches the level of computer controlintegration that will be needed if a robotis to be capable of autonomously re-sponding to a programmed set of direc-tions by referencing a self-containeddata base for its location, destination,route, and tasks.

Consummating the union betweenrobots and artificial intelligence is along-range research goal, however, be-cause the challenges involve advancingthe frontiers of computer modeling ofsolid geometry, as well as the struc-turing of large amounts of computerdata for logical access by the robot.Various military and nonmilitary re-search programs around the countryare now focusing on the mathematicaland computer science aspects that willeventually be brought to bear on thischallenge. The military programs arelargely funded by the Office of NavalResearch and the Defense AdvancedResearch Projects Agency. Others, in-cluding programs at Stanford Univer-sity, Purdue University, the Universityof Michigan, the Massachusetts Insti-

tute of Technology, and CMU, involvenonmilitary as well as military-relatedR&D.

Irving Oppenheim, an associate pro-fessor of civil engineering at CMU, isworking with EPRI on some aspects ofthe problem in a research project to as-sess the potential for applying artificialintelligence in robots for constructionand maintenance work. The Japanesealready make significant use of auto-made devices for various tasks in con-struction, but, in general, these devicesare not the smart type. Two elementsthat are needed to make robots autono-mous, according to Oppenheim, are theability to logically detect and avoidobstacles and a way of modeling thethree-dimensional work environment ofthe robot so that its "world map" canbe referenced as it proceeds on an as-signed task.

"There"are some attempts at themathematics that will permit a robot tofind a configuration that avoids an ob-stacle, and we are working with the ex-isting ones, testing them out, findingtheir shortcomings, and modifyingthem to accomplish some of the ob-jectives that these obstacle avoidancecapabilities are going to have," saysOppenheim. "For example, we're test-ing whether a control algorithm canfigure out how to command a robot toreach around two pipes, then reach inand touch a third pipe."

Progress in the second area of re-search—providing the robot with anaccurate, three-dimensional model ofits work environment—could some-day lead to a robot's direct use of theoriginal and as-built design drawingsof an entire nuclear plant ExplainsOppenheim. "There must be a datastructure, a computer program, thatstores all the plant dimensions, thewall openings, solid areas, pipes, inter-sections, and so on.

"There are two approaches to thisproblem. One is to build a robot thathas sensors all over it and simply keepsit eyes and ears open and doesn't touchanything. The other way is to somehowmake use of all the dimensional datathat have already been recorded andare on drawings and computer-aideddesign systems. We are exploring thekind of computer data structure that isbest suited to the problem."

Designing nuclear plants with robotsin mind is another area in which EPRIhas sponsored research. Many of thedifficulties involved in using a robot to-

36 IAEA BULLETIN, AUTUMN 1985

This radio controlled robot called "Kluge"was designed for surveillance and to carrydifferent types of equipment. (Credit:Cybernation Inc.)

"Herman" is a mobile manipulator at theOak Ridge, Tennessee Y 12 plant used asa stand-by system to work in toxic orradioactive environments. (Credit:Martin Marietta Energy Systems, Inc.)

Robotic evolution

ISIS, developed by Hispano Suiza, isbeing used at France's Chinon A3 reactorfor repairs. (Credit: Hispano Suiza)

Although the robotics industry itself is less than twodecades old, the technology can broadly claim old anddistant relatives — from musical statuettes to mechanicalmanipulators and programmable machines — around theworld.

Early Greeks, Egyptians, Ethiopians, and Chinese, forexample, created a variety of moving figures that werepowered by water and steam. Later, in the 18th and early19th centuries, Swiss craftsmen built life-like "automata"that could write, draw, and play musical instruments; andthe French developed mechanical looms controlled bypunched cards, introducing the f irst programmable machine.

The term "robot" itself, however, was not widelyused until 1921, when the play Rossum's Universal Robotsopened in London. Written by Czechoslovakian dramatistKarel Capek, the play popularized the derivative of theCzech robota, which means forced labourer.

Today the definition of the word "robot" il lustratesboth rapid technological advances and modern expecta-tions. In the USA, the Robotics Industries Associationdefines a robot as a "reprogrammable multifunctionalmanipulator designed to move material, parts, tools, orspecialized devices through variable programmed motionsfor the performance of a variety of tasks". In Japan,classifications are used: M1 are simple tele-operatedmanipulators; M2A are devices that can be programmedto do fixed repetitions; M2B are those that can performvariable repetitions; M3A are more sophisticated devicesthat can be taught a sequence of steps by an operator

leading them through the motions; MSB are robots thatcan be numerically controlled with a computer; and,ultimately, M4 are robots having "artificial intelligence"that are capable of fully autonomous operation.

Today the technology generally is considered to beat the M3 stage, with research and development well intothe M4 level. Computer "chips", sensors, televisioncameras, and other electronic devices are fuelling theevolution. Economics, however, will dictate the extentof future applications, experts say.

In the nuclear industry, mechanical cranes and mani-pulators used in the early days of development are amongthe forerunners of today's more advanced remotesystems and robotic technologies. One of the first robotsfor practical use was developed in 1958 by Hughes Aircraftto handle radioactive materials at nuclear faci l i t ies in the USA.

Some indications of how far the technology has comein the nuclear industry emerged at an international seminarlast year convened jointly by the IAEA and the NuclearEnergy Agency of the Organisation for Economic Co-operation and Development. More than 200 participantsexchanged information detailing the advances in electro-nics, optical/visual systems, and material technology thathave combined to enable innovations and improvements.(Proceedings of the Seminar on Remote Hand/ing inNuclear Facilities now are available from the OECD,2 rue Andre-Pascal, 75775 Paris, Cedex 16, France.)

Photos on these pages show some robotic systems inuse today.

Information for this article is drawn from "Industrial Robots on the Line", by Robert Ayres and Steve Miller, TechnologyReview (May/June 1982), and from an article by T. Moore in EPRI Journal (November 1984).

Surveyor can be used to monitor radiation,listen for steam leaks, and read gauges.(Credit: Automation Technologies, Inc.)

A robot manipulator for use in Japan'sreprocessing plants and radioactivewaste facilities. (Credit: PNC, Japan)

The MF3. developed by CMS Technologies,Inc., in the FRG, has been used at nuclearplants over the past 10 years. (Credit: EPRI)

IAEA BULLETIN, AUTUMN 1985 37

Nuclear power and electronics

day stem from the fact that the plantswere not built with such devices inmind; advanced reactor plants of thefuture will likely have special featuresspecifically to accommodate surveil-lance or maintenance robots.

Under an EPRI contract, Westing-house's Advanced Energy SystemsDivision studied the feasibility of usingrobots in a large-scale prototype breederreactor. The analysis considered variousroutine and nonroutine maintenanceand inspection tasks and outlined de-sign factors that could enhance the ap-plicability of robots These include pro-vision of adequate wp,rk and accessareas, lighting and power outlets, andlocation of equipment and other poten-tial obstructions.

As more special-purpose robots aredeveloped for-nuclear applications, thejob of technically evaluating these de-vices with utility requirements in mindwill also grow. EPRI's NDE Center maytake on expanded responsibilities inthis regard, having already participatedin the technical evaluation of IRIS.

Breaking new ground

Directed R&D efforts and the immedi-ate needs in nuclear power plants for

reduced maintenance costs and loweroccupational radiation exposure arebreaking new ground in the applicationof robots to tasks with which most peo-ple would rather not be burdened. De-spite the significant achievements todate, however, researchers caution thatmuch more progress must be made be-fore robots are seriously considered asreliable, economic tools. The entry ofrobots into the nation's nuclear plantswill not occur rapidly, but a trend inindustry thinking toward applying ro-botic equipment when and where it isfeasible is already clear.

Michael Kolar, until recently an EPRIsenior program manager who was in-volved in the Institute's study of roboticapplications since the effort began in1981, reflects the mixed viewpointsamong many researchers in the field.

"There is some robotic technologythat will let you do certain jobs, but it'snot at all clear that you'll see many ofthese machines in wide use in the nearfuture," says Kolar. "There are signi-ficant unresolved uncertainties, relatingnot only to the technology's hardwareand software but also to other issues.Will the time required to train crewsand execute a job with robots be shortenough to be practical? That's not yet

clear. NRC may decide to regulate someaspects of plant maintenance, and therole of robots in licensing issues hasnot yet been defined.

"Ultimately it will all come down toeconomics—are robots truly cost-beneficial?" asks Kolar. "Unless thecosts of robot systems come down, orsomeone offers to provide them as partof a service package, I don't think we'llsee widespread use of sophisticated ro-bots soon. For EPRI, the issue is to en-sure that good technology gets into theplants. But first, we have to find outwhat these machines can do. If we suc-ceed, robots just might make it."

Utilities are expressing increasing in-terest in robots for nuclear plant ap-plications, and as a result, the R&Dcommunity and the robot industry areresponding with a range of devices andmachine capabilities The current activ-ity represents a model of cooperativeresearch, with both large and smallcompanies, universities, government,and industry research groups workingtogether to advance the technology. Ifrecent success is any indication of thefuture, the outlook for robots to make asignificant contribution to improvedplant economics is encouraging.

For further reading . ..

Evaluation of Robotic Inspection Systems at Nuclear Power Plants, prepared for US Nuclear Regulatory Commission byRemote Technology Corp., NUREG/CR-3717 (March 1984).

Automated Nuclear Plant Maintenance, final report for RP2232-1, prepared by Battelle, Columbus Laboratories, 505 KingAvenue, Columbus, Ohio 43201 (1985).

"I ndustrial Remote I nspection System", by E.B. Silverman, Proceedings of the Robotics and Remote Handling in HostileEnvironments, National Topical Meeting, American Nuclear Society (1984).

38 IAEA BULLETIN, AUTUMN 1985