A major faci
is being establis
provide extreme o sn aw m
p nt r e s m t and standards
emulating a wide varilsty Bf rn
typlcal of a small machine job shop. The control in the decades of the 1980s and
architecture adopted is hierarchical in nature 1990s. When completed in 1986, this facility will be
and highly modular. The facility will be used for capable of full-scale emulation o f the flexible
research on interface standards and metrology machining c in the automated faczory o f the
in an automated environment. future.
Keywords: Automated Machining, Hier-
archical Control, Manufacturing Research, Purpose of the A M R F
The Automated ,Manufacturing Research Faci-
The Congressional Acr e d n g up the National i will reside in the Center for Manufacturing
Bureau of Standards charges the Bureau with; Engineering which was founded to supply to the
1. The custody, main&nce, and dcveiopmcnt mechanical manufacturing sector the servicts des-
of the national standards o f measurement, and cribed in the enabling legislation and to carry on a
the provision of means and methods for making research program to develop 'means and methods "
measurements consistent with those standards. for making the measurements that w be needed by
2 Cooperation with other government agencies this -or in the future. The Center currently
and with private organizations in the establish- provides a wide range of calibration services for
ment of standard practices, incorporated in mechanical artrfact standards such as gage blocks,
codes and sp&ifications. thread gages, and line ss as shown in Figure 1
To perform these functions, the Bureau hs a , Thcse artifact standards, many of which were
i s ld
over the yean, ntc numerous experimental developed by NBS in the first thm decades of this
facilities, including a nuclear m a r c h reactor, a century, arc idealized models o f the products to
* A c t of 22 July 19SO. W Sut. 371 (Pubiic Lav619.31 Congms) - An Act To amend seaion2 of the Act of March 3.1901 (31 Scat.
IMP). I provide borrc authority for the performance of certain funcuoru and activities of the Department of Commerc:. and for other
which they are compared. The comparisons (cali- aeu
to the labor intensive n t r , high s k i l l requirement.
brations) arc organized according to statistical and time consumption o f classic metrology.
quality control methods developed during and The first effon of NBS to meet the upsoming
immediateiy after Worid War 1. Anifacts currently
1 challenge was in 1968 when a research program was
a n the basis for the National MeYurrmenr System mounted to investigate the possiblity of automatmg
which provides nation-wide dimensional cornpati - surface plate metrology by the use o f the then new
bility by a chain of comparisons back to National computer controlled coordinate measuring machines
Standards. The system has remained virrually (CMM). A decade o f work realized a measurement
unchanged since the 194Os, except for the intro- system based on such machines where the "product -
ducion in the 1960s and 70s of the concepts o f like" artifact standards of the past were replaced
Measurement Assurance Programs (MAP;', which ,with measurement protocols based on laser inter-
emphasize the system aspects of measurement and ferometer techniques for characterizing the mea-
introduced the concepts of closed loop feedback suring system (coordinate measuring machine) itself.
into metrology management. Transfer standards were developed that permitted
Manufacturing technology, however, bas not such machine or process characterization to be
remained unchanged. The introduction o f numcri- economically realized on machines o f lesser but
a l l y controlled machines, group technology con- known precision'. The three-dimensional ball piate
cepts, and the first steps toward nexible Manufac - on the table of the CMM in Figure 2 is one of the
turing Systems (FMS) in the 1960s called attention latest. of such standards. These new measurement
The total Bureau experience has amply demon-
s t r a t c d that one Cannot learn to measure without
hs on" experience, and every attempt to attack
measurement problem on a purely theoretical basis
has proved less than satisfaczory. Therefore, in
cooperation with the Bureau of Engraving and
Printing, an NC machining center was set up in the
NBS Instnrrneat Shop to explore the measurement
problems involved in assuring pan dimensional
llceutacics by machine calibration. It was soon
shown that the calibration techniques and software
c o d o n algorithm for static e n o n developed on
coordinate measuring machines could be applied to
machine tools in a shop environment. A five-fold
increase in accuracy was demonstrated. Figure 3
i v t s the de- of comction obtained for a
The c o m a i o n of dynamic e r r o n such as
thermal distortion due to internally generated heat
or distonion due to cutting forces needs further
mearch, but appean to present no insurmountable
obstacles. Certain complex dimensional measure-
ments such as drill condition and tool setting are
needed, but modern microcomputer based tech-
nology appears adequate to the task.
The AMRFwillallow research in mdsurernent
technology to be expanded to include those system
elements at the c (multiwork station) level. The
methods are rapidly becoming the norm for certain AMRF w provide a test bed where integrated
pan familia; T a ae arc medium to large in
h t fm l
ii manufacturing system measurement research a n be
size and compiex -prismatic in nature, and hence . pdomled.
similar to the output of the first and second . The AMRF will provide a test bed for research
generation FMS, directed toward the'establishing of standard prac -
Even before this work was completed, it became tices". If flexible manufacturing systems are to
obvious that the= were many pan families that were become widely atioptcd in the discrete pans Industry
is i d to measurement by CMM. Small pans,
- t when 87% of the fi'ims employ less than 50 persons,
turned parts, and very simple parts arc all either very they must become much more modular then they
difficult or uneconomic to :measure in this manner. are today. I t must become possible for a finn t o stan
Moreover, the rapid development of FMS, with the with an NC machine, add a robot, add another
ability to reduce inventory by shoner runs, casts machine, and so on as capital is accumulated and as
doubt on the continuing usefulness of any QC the f i r m ' s business grows. Systems must also be
system which depends on statistical sampling. Along capable of being tailored to various pan mixes
with others, NBS became convinced that the QC without extensive engineering effort. However.
system of the future would increasingly depend on before this degree of flexibility can beaccomplished.
characterization of the process. monitoring of the interface standards must be adopted so equipment
machine parameters, and adaptive control rather of diverse origin c n be integrated incrementAly
than measurement of part parameters after the into the systems.
process, or a step in the process. was complete. Such n e first steps in this direction have already
a development will require NBS to provide the b e taken. Under Air Force Integrated Computer
-means and methods " o f measufement where the Aided Manufacturing (ICAM) sponsorship, the
measurements are deeply embedded in the process. NBS coordinated the efforts of a consonium of 45
pnvate firms 10 generace the Initial Graphic test bed for the development of similar intedace
Exchange Specification’ (IGES). IGES is a common standards for integrated manufacturing systems. It
public domain data format which allows geometric will allow the test and verification o f interface .
data to be exchanged between two different types of standards in an open and nonproprietary atmo-
computer aided design systems, or between a sphere.
computer aided design and a computer aided manu-
facturing system. IGES thus allows access to the
geometric data bank o f a computer aided design
system without the necessity of producing a drawing.
Description of the AMRF
IGES has recently b n n incorporated into a national
standard (ANSI Y 14.26M). The development of The Automated ,Manufacturing Research Faci -
this standard is important for its intrinsic value; but lity will supeficially resemble a FMS designed to
perhaps more important, it has demonstrated that handle the bulk of the part mix now manufactured
such Interface standards can be structured and in the N5S lnstntment Shop, T’his part mix has betn
generated in a manner which provides full protection studied using Group Technology‘ concepts and I S
for proprietary interests. The A.MRF will provide a shown t o be similar to a typical job shop. The pans
Jounul of Manufaccunng Systems
Volume I. \umber I
manufactured wiil fall within the following limita- elaborate total system inventory. At this stage we
tions: plan to use the inventory system for thcstorage of
1. Weight: Less than 50 kilogram (100 I s.
b) raw material blanks, tools and tool holders
2 Size, Prismatic: 300 mm cubes (12” x 12“ x
. (assembled), special fiitures. and finished parts or
127. paru in-process. The inventory system will be
3. Size, Rotational: 250 mm diameter x 250 rnm loaded and unloaded manually while the facility is
length ( ”x IO”). .
IO in operation.
4. Parts Run: 1 to 1 O pieces. The Materials Transport System will provide
5. Complexity: Up to 4 axes prismatic. the means of moving parts, tooting, and fixtures
6. Mate& Steel, stainless s e i , aluminum, within the facility. Two mechanisms will be used.
brass, iron, lucitc. i
One, a carousel, w also serve as the inventory
The AMRF and the mearch performed on it a
system. The second, a robot c n o r automated
wiil address only tne manufacture of individual guided vehicle (AGV), will allow great flexibility in
pans by chop forming metal removal. Hence, the layout and easy access to the machines’. Although,
unit operations will include only: fixturing, milling. the transfer system itself is not s e as a primary
drilling, reaming, tapping. boring, turning, facing, research area for NBS, the interfaces between the
threading, cicaning, debumng, and inspection. Such work stations and the transfer system will be
problems as automated assembly, welding, harden - designed to accoaunodate many different types of
ing, and finishing w i l l not be systcms as well as other options in order to maintain
The AMRF hardware is structured around the modularity.
concept of singie selfxontained work stations, e c ah The machine tools were chosen to be represen-
with a well defined set of functions which can be tative of the types of general purpose machine tools
useful as a stand-alone entity. The current plan cl a
sl in common use throughout the U. S. The chotce also
for the existence of eight such stations with varying matches the speclftc needs of the NBS Instrument
degrees of complexity of function. They arc: Shops as revealed by the Group Technology Study.
I. Horizontal Machining Station. Each of the machines will be configured into a work
2. Venical Machining Station, station with a single industrial robot.
3. Turning Station. NBS’has chosen to use standard, modem.
4. Cleaning and Debumng Station. general purpose machine tools in the construction
5. Inspection Station. .of the AMRF. This is a different strategy than that
6. Materials Inventory Statlon. taken by two other wciI known national programs in
7. Transfer System (station). automated batch manufacturing, the Bntlsh A S . P.
8. Housekeeping System (station). plan’, and the Japanese MUM or FMC plan-. Both
Items ’land 8, the Transferand the Housekeep - o f these other programs have assumed apriorr that
ing Systems arc not strictly stations since they are curnnt machine tool designs are inadequate for an
nonlocalized in the facility. From the point of view automated research facility. However, based on an
of the control system, however. they w i l l be treated international study of the state-of-the-art in machrne
as stations. tool science’, NBS has decided that this assumption
T h e Materials Inventory Station wiil be used is highly questionable. We have chosen to rely on
as a buffer to allow storage o f sufficient material for the engineering experience of a well -developed
several days o f operation and an automatic inven- industry rather than a radical new design. Should
tory for much of the raw material requirements of a problems arise in r ‘ ~ bdity, reparabllity. and chip
Job shop. Ifsuch a system were to serve sunply as a removal, we plan to subcontract any needed modi-
buffer, ony three or four days storage would be fications to the same industry.
required. that is. enough for automatic operation Cleaning and debumng was made Into a
through a long weekend. Since It is not the purpose separate function (and station) because of the
of this program to study such systems per se, one importance of this task for automatic Inspection. As
week of storage was chosen as a reasonable trade off many . debumng operations as possible will be
between the requirements o f the simple buffer (or carried out at the machining site. Neverthe!ess. there
Interface to the manual world) and a much more appears to be no way to avotd cleaning and
Jounul of Manufacturing Systcrm
Volumr I. Numkr 1
deburring as a separate operation in all cases. This system will be kept as simple as possible wlth
Studies have r e d that the cost for cieaning and little attempt to optimize for long unattended runs.
deburring in batch manufacturing is high and often Layout for the facility is shown in Figure 4.
~nfc~~gnizcd~. This configuration allows easy access to the
The Inspection Station will be a modifmi four macbines, and the transfer mechanism can be either
axes horizontal ann measuring machine tended by a the automatically guided vehicle system or the
robot. It w be very s@ilar to the machining work a n d carousel system. Coolant and cutting fluid
stations from the control pointsf -view. This arc r yd at the machine. Buffering is provided to
configuration was chosen primarily for flexibility in the m c i e through the row of "file-cabinets "
a hn s
w. which make up the carousel shown in the center o f
The Howkaping System will provide for the the model.
removal o f chips during automated operation. The three robots in the lower right are part of
Cleanliness during manufacturing and fituring, the cleaning and deburring station. The separate
and the effects of cutting fluid and chips (dust) on room at the upper right is the inspection station. The
sensors have been serious problems in many of the four machining stations each consist of an industrial
existing F M S In the AMRF, chip mnovai robot, a machine tool, a localized inventory o f tools,
is expected to be complicated by the variety of fixtures, grippers (end effectors), probes, and inter-
material, the l r e number of sensors contemplated, faces to the transfer and housekeeping systems.
and the decision to address the flexible fixwing The proposed operational scenario places
problem robotically a t the rnrchincs. A pian severe requirements upon the work station and its.
regarding chip removal is being dweloped at this
p subelenrenu. Some of these arc necessitated by the
time through both externai " ' 2 and i 1 studies. dccision not to palletize and others by the wide part
. . . . . . . . . . _. . . . . . _. _ ........ . .. .. ..... . k..
Modd of AMRF Skohng Loatioa in Insmtrnmt ShoQ.
The cmtnll. louted u r o w r l rill be used to convq rnrtni.l.
t o and finished parts. For purposas of mvironmcnrd control
the meuunng station is, as shown. in a wpance room.
mix envisioned. The tools and material amving at a Tab&I
work station will not be precisely located ia space. Functions of the Work Station Subelements
This will require advances in the state-of-the-art Robots Arm Functions
over current industrial robot capabilities. The robot .
I Part loading and unloading.
capabilities will abo be stretched by the requirement .
2 Tool loading and unloading.
o f futuring on the machine. The problems of chip 3. Rough (k 50 mil) p a n fixturing or fixture
buildup and tool wear w be aggravated by the
material and part mix contemplated. I t is our belief 4. Chip removal and control.
5. Come visual inspection of faturn and pant.
that those requirements wbe the n o m in second
l n i
6. I t l part and tool location.
generation FMS s y s t e m which wbe available in
li 7. End effector sefmion.
the 1990s. 8. Deburring and deaning (oniy as needed for next
Table 1 gives a. paniai list of the functions opcration).
q u i d of the robots and machine tools in the 9. safety.
AMRF. As c n b e seen. it is intended that the
industrial robot be able to locate pans, toob and Machine T o d Functions
fixtures, transfer these items to the machine tool,
fixture the pan. and monitor the process while 1. Machining.
2 Part locations.
machining is carried out. Thus the robot will have 3. Tool weart breakage sensing.
extended sensory c p bs the ability to precisely
a at i i
grip and position variable shapes, and considerable 5. Proctss mcnitoring (cutting).
manipulative ability to fixture the p a m upon the a Dac
. n i.
machine tool. In general, solutions to thae problems b. T m
h ar l
c. Hydraulics (ac.)
are more difficult for prismatic than for cylindrical
wor k p i c k . 7. Deburring and cfeaning (as pan of the machining
To the best of our knowledge, no one has operation).
addressed the flexible fmturing problem. in any 8. Adaptive control.
depth though some very elaborate and expensive, .
solutions have been proposed by T~ffmsamrner'~. needed to load the facility will be the larsely manual
Tool setting on machining centers is in*a similariy processes currently used for the NC station of the
undeveloped stage, as is generic tool wear/ breakage instrument shop.
sensing". Our projcct plan has been to delineate as
carefully as possible those areas requiring develop
ment, research carefully the state-of-the-art in these
areas, and if required, initiate research dirccted ~ ~~ ~~~
towards the appropriate goal(s). In order for the A.MRF to serve as a research
A t present the: Center. for- Manufacturing facility overthe next decade. it must exhibit a higher
Engineering has two projects. one in robotics, the order o f flexibility than any cuncntly available
other in precision machining. These projects are FMS. I t must not only be capable o f very wide p a n
directed towards the development of the major mix, but must also be capable o f easy reconfigura -
subsystems required for the AM RF. The integration tion to emulate work stations or small ce!ls operating
of these two programs will take place first in the in the environment o f a much larger and perhaps
Horizontal Machining Work Station which will be unmanned system. To accomplish .these goals
the tint work station to be assembled. T h e archi - requ ires a control system architecture of consider -
tecture and control system hardware for this work ab12 sophistication. The conventional Direct
station will. s e r e as a model for the other four Numerical Control (DYC) top down architecture
generically similar work stations. was judged to be unsuitable. primarily because o i
Although i t is recognized that there are the inability o f such a system to react !o feedback
imponant problems of CAD. CA,M integration to from sensors in real-time. In order for an entire
be solved, the current plans do not include work in machine shop to completeiy operate automatically,
this area. Production and process planning systems ail the machines must be equipped with sensors to
Jounul of Manufaauring Systems
Volumr 1. Numbcr 1
monitor their perfonnanct and compensate for w react to events of days or weeks duration
irregularities and uncertainties in -the work (production planning and scheduiing problems).
environment. The sensor data must be processed The levels in between these extremes will produce
and analyzed, and the r ds inuoduccd into the
u t intelligent automatic rcsponsu to many different
machine control system in real-time so that the types of shop floor conditions and situations.
response of each machine is goddirmed, relioble,
A high degree of kory -intcnretivsbehavior
ahu e t
on the part of individual m cn a a enormous
u contra1 suucturc
system control problems for an entire shop. The
problem of automatically controlling a number of
feedback driven machine tools is much bigger than On the left of Figwe 5 is an organizational
simply the s u m of the contrcl problems for the hierarchy wherein computing modules a n arranged
individual machines. The interactions among many in layers. The basic structure of the organizational
sensory -interactive machines creates a system hierarchy is a tree. The flow of command and
control problem in which complexity grows expol control is vertical. Each node in the tree represents a
nentidy with the number of individual machines raxives input commands
and sensor systems. Once there are more than a few d e (predecessor
nucfrines, each reacting to m o r data in reaI-time, to one or more
the overall system control problem can bccomc r node). There may
cornpleteiy unmanageable. This is the point at 8 sensory icputs and
which most of the early attempts at building the cing data that flow
agtomatic machine shop failed. The controi software horizontally and/ or rise from lower levels in a cross-
for such a system can bccorde enormously oomplex coupled netwark o f communication channels, but
to write and virtually impossible to debug. The the primary command and control pathways form a
classical solution to control problems of this s t r i c t hierarchical tree.
complexity is to partition the problem into modules At the top of the hierarchy is a single high-level
and introduce surne type of hierarchical command computer module. Here at the highest level, most
and control struaure. The advantage of hierarchical &ob4 goals are decided upon and long-range
control is that it allows the.contro1 problem to be strategy is formuiated. Feedback to this level is
partitioned so as to limit the complexity of any 'integrated over an extensive time period and is
module in the hierarchy to manageabie'limiu, evaluated against long-range objectives. Here long-
regardless of the complexity o f the entire StruCture. range plans arc formulated to achieve the highest
The use o f hierarchical control for industrial priority objectives. Decisions made at this highest
applications is not new. It has been employed in level commit the entire hierarchical structure to a
controlling complex industriat plants such as steel unified and coordinated course o f action which
mills, oii refineries, and giass works for years. would result in the selected goal o r goals being
However, such hierarchies are usually limited to two achlevcd. At each of the lower levels, computing
or t h m levels and; most often represent falrly modules decompose their input commands in the
stralghtforward servo control applications. The context of feedback information generated from
unique features of the control system being planned other modules at the same or lower levels, or from
for the AMRF are the number o f hierarchical levels the external environment. Sequences of subcom -
(perhaps as many as seven or eight), and the amount mands are then issued to e t s of subordinates at the
of real-time computation and sensory -interaction at next lower level. This decomposition process is
' each level. Each hierarchical levef will perform a repeated at each succtssiveiy lower hierarchical
significant amount of real-time computation and levei, until at the bottom of the hierarchy there is
will Interact dynamically with the shop environment generated a set of coordinated sequence of primitive
in many different ways. T h e plan is to build a r a- actions which drive individual actuators such as
time sensory -interactive control system which at the motors o f hydraulic pistons in generating motions
lower leveis will respond to events o f millisecond and forces in mechanical members.
duration (tight servo loops), and at the upper levels Each cham -ofcommand in the organizational
Journal of Manufacrurtng Systems
Volume 1. \i;moer I
ORGANIZATIONAL COMPUTATlONAL BEHAVIORAL
HIERARCHY HIERARCHY HIERARCHY
. hierarchy consists of a computational hierarchy of 'from the sensory processing module at :he same
the form shown in the center of Figure 5. This level) and computes an appropriate output. A
computational hiciarchy contains t h m parallel detailed description of such a system as applied to
hierarchies: (1) a task decomposition hierarchy en
robots has b e published
which decomposes high-level tasks into low level The sophisticated real-time use o f sensor data
actions, (2) a sensory .processing. hierarchy which for coping with uncenainty and recovering from
p~ccsses sensory data and extracts the information errors requires that sensory information be able to
needed by the task decomposition modules at :ach interact with the control system at many different
level and (3) a world model hierarchy which gene- levels with many different constrLints on speed and
rates expectations of what sensor data should be timing. Thris in general. sensov information at the
expected at each level based on what subtask is higher levels i s mort abstract and requires the
currently being executed a l that level. integration of data over longer time intervals.
Each level of the task decomposition hierarchy However, behavioral decisions at the higher levels
consists o f a processing unit which contains a set o f need tu be made less frequently. and therefore the
procedures, functions. or rules for decomposing greater amount of sensory processing required can
higher level input commands into a string o f lower be tolerated.
level output commands in the context of feedback Attempting to deal with this full range of
information from the sensory processing hierarchy. sensory feedback in all of its possible combinations
A t e v e y time increment each H module in the task at a single level leads to extremely complex and
decomposition hierarchy samples i t s inputs (com - inefficient prosrams. The processing o f sensor data.
mand inputs from the next higher level and feedback panicularly vision data. is inherentiy a hierarchical
Journal of Manufacturing Systems
Volume 1. Number I
process. Only if the control system is also panitioned level of the task decomposition hierarchy are
into a hierarchy can the various levels of feedback represented as points in the rnultidimenstonal
information be introduced into the appropriate "stateapace "consisting of the coordinates of all the
control levels in a simple and straightforward degrees of fmdom of the machine or robot, and
manner. these points are plotted against time, the behavioral
The world model hiexarchy contains prior hierarchy shown on the right of Figure5 results. The
howlcdge about the uqk, the pu and the work lvl
lowesteetrajcctories of the behavioral hierarchy
environment. Typically, the type of f eed campond to obscmble output behavior. All the
mation required by the task decomposition modules other trajectories constitute the structure o f condi-
at each lwei depends upon what task is k i n g tiom deep within the control programs.
performed. As conditions change, different scasom, ah
A t e c level in the behavioral hierarchy, the
different * resolutions, and different processing string of commands m k s up a program. This
algorithm may be needed. Given the mtt of the architmure implies that thm is a programming
task exccution at e c level, the world model c n
ah a language unique to each level of a hierarchial
predict what kind of sensory processing algorithms control system, and that the procedures executed by
should k applied to the incoming data. Funher- the computing modules at each level are written in a
more, sensor data c n often be predicted from the
a language unique to that level. This partitioning o f
actions k i n g exccutd by the control system. thecontrol problem into hierarchial levels limits the
The worid modei generates expectations aa to complexity of the progrnrnming language and the
what the sensor data should look like. These l v. l
programs at each ee It also generates a whole
predictions may be based on prwious experience hiefarchy of languages for programming the robots.
when a similar task was performed on a similar pan, machine toob, and inspmion systems. and for
or may be generated from a Computer Aided performing, planning and scheduling operations. I t
Design (CAD) data base which contains a geo- is to be noted that sucha hierarchy lends itself to the
metrical representation of the part. The world utilization of IGES-Cype interface standards at each
model hierarchy may contain information as to the level.
shape, dimensions, and surface features of parts and If the controi problem is further panitioned
tools and may even indicate their expected position aiongthe time axis, an additonal degree o f simplicity,
and orientation in the work environment. This can be achieved. If time is panitioned into a finite
information assists the sensory processing modules number of computational periods, each computa -
in seiecting processing algorithms appropriate to a
tional module c n be represented as a finite-state
the expected incoming sensor data. and in comiat - machine. At every time interval, each computational
ing observations against expectations. The sensory s
module samples t inputs (command and feedback)
processing system c n thereby d n m the absence of
a and computes an output. The programs resident in
expected events and measure deviations between each of the computational modules then become
what is observed and what is expccted. simple functions which can be represented by
Feedback can be used by the task dccomposi - formulae of the form P=H(S), or by a set o r
tlon hierarchy erther to modify action so as to bring production rules of the form IF<S>, THEY<P>.
sensory observanons into correspondenct with The control structure becomes a simple search. of a
world model expectations. o r to change the input to state transition table.
the world model so as to pull the expectations into Each entry in the state -table represents an
comspondcnce with observations. In either case, IF/THEN rule, sometimes called a production.
m c e a match is achieved between the two, the task This construction makes it possible to define *
decomposition hierarchy can act on information behavior of high complexity. An ideal task per-
contained in the model which cannot be obtained formance can be defined in terms of the sequence o f
from diren observation. For exampie. a robot states and state transition conditions that take place
control system may use model data to reach behind during the ideal performance. Deviations from the
an object and grasp another object which is hidden ideal can be incorporated by simply adding the
from view. deviant conditions to the Itft-hand slde of the state-
If the symbolic commands generated at each table and the appropriate action to be taken to the
Journal of Manufacturtng Systems
Volume I. \umber I
right-hand side. Any conditions not explicitly The logical structure of Figure 5 is mapped
c o v c d by the table results in an"1don't know what into the physical structure o f Figure 6. The
to do" failure routine being executed. Whenever coordinate transformations of Figure 5 a r t impie -
that occwf, the system simply stops and ask- for mented in one of the microcomputers o f Figure 6.
instructions. If the condition can be corrected, a The elemental move trajectory planning is imple-
human programmer c n enter a few more mles into
a mented in a second microcomputer of Figure 6. T h e
the state-table and the system c n continue. By this processing of visual data is accomplished in a third
means, the system gradually ers how to handle a
la n microcomputer, and the processing for force and
larger and larger range of problems. This cxtensi - touch data in a fourth microcomputer. A fifth
bility of the system to new problems is essential in a microcomputer provides communication with a
research facility which. by t very ~ t m w i l l
s , minicomputer wherein reside additional modules of
usually oprate at the very m of the arrent state
l is the control hierarchy. It is anticipated that these will
of knowledge. evennully be embedded in a sixth microcomputer.
Such a finite -state machine hierarchical control Cornrnuniation from one module to another
system has b e implemented on a microcomputer
en is accomplished through a common memory 'mail
network. This network, shown in F i g w e 6 has b een drop" system. No two microcomputers cornmuni -
under evaluation as a control system for the robots catt directly with each other. This means that
in the AMRF ". common memory contains a location assigned to
J b b
16 a BI T~P 16 B l l p P 16 a l r p
MlNlCOMPUfER 16 COMMUNICATIONS ~
COORO TR AJ.
INTER FACE XFORM CALC
COMMON 16 611 pP VISION
MEMORY VISION INTER F A C E
c I I
Journal of ManuPnunng Systems
Volume I. ?(umber I
every element in the input and output vectors of the logical modules not directly involved in the
every module in the hierarchy. No location in change.
common memory is written into by more than one Funhennore, the common memory always
computing module, but any number of modules contains a readily accessible map of the current state
may read from any locatioa. of the system. This makes it easy for a system
Time is sticed into 28 miibecond t
n mn. monitor to trace the history of any or all of the state
At the &ginning o f e c increment, uch l o g i d
ah variables, to set break points, and to reason
module reads t set of input vllucs from the
s backwards to the source of program erron or faulty
appropriate lo lOgk
computes t st of o
s The rmkompute -write -wait cycle wherein
back into the co each module is a statemachine m k s it possible to
millisccood i n t d d
es Any of the 1 stop the process at any point, to singie step through
which do not complete their computations M o r e ak
a t s and to observe n detail the performance of
the end of the 28 millisecond interval write extra- the control system. This is extremely important for
polatcd estimates of their output accompanied by a program development and verification in a sophisti -
flag indicating that the data is extrapolated. The cated. real-time, sensory -interactive system in which
process thcqe ar e,
ps t many processes arc going on in parallei at many
Each logical module is thus a state-machine different hierarchical leveh.
whose outputs depend only on t i
spresent inputs and The hierarchical control stnffure just des-
t present intrmal state. None of the l o g i d modules
s cribed is a generic concept which can be extended to
admit any intermpu. Each s t a r t s t read cycle on a
s apply to a wide variety of automated manufacturing
clock s n , computes and writes i output, a d
i l s
t n systems. NBS plans to use this conceptual frame-
g a h,
i l u
waits for the next ciock sn. T seach logicai work for the control system and data base archi-
module is a finitc-state machine with the IFITHEN, tixture for the AMRF. figure 7 is a block diagram
or P=HfS) properties o f an arithmetic function. . of the control system planned for the AMRF. The
The common memory ' a li
mdrop" communi- square boxes arranged in the hierarchical structure
cation system has a number o f advantages and in the center o f the figure represent the task
disadvantages. One disadvantage is that it takes two decomposition modules at the various levels o f
data transfers to get information from one module control.
to another. However, this is offset by the simplicity A t the lowest levei in this hierarchy are the
of the communication protocol. No modules talk to individual robots, Ni C machining centers, smart
each other so there is no handshaking required. In sensors, robot carts, conveyors, and automatic
each 28 millisecond time slice, all modules read from storage and retrieval systems, each o f which may
common memory beforeany are allowed to wnte have its own internal hierarchical control system.
their outputs back in. The bottom row of boxes represents the control
The use of common memory data transfer systems for these individual machines. The small
means that the addition o f each new state vanable subboxes labeled s and C correspond to the sensory
requires only a definltion o f where the newcomer is and command interfaces to these control systems.
to he located In common memory. This information The command input to the robot in Figure 7
is necded only by the module which generates it so corresponds to the Y3 Elemental Move Module
that It knows where to wnte it, and by the modules input in Figure 5.
which read it so that they know where to look. None The bottom row o f control modules in Figure 7
of the other modules need know, or care, when such is organized into work stations under the second
a change ts impiemented. Thus. new microcomputers row of work station control modules. A work
can easily be added, logical modules can be shlfted station may consist of a machine tool. a robot. and a
from one microcomputer to another, new functions set of smart sensors. It may also consist o f a set of
can be added. and even new sensor systems can be robot carts, or an automatic storage and retrievai
introduced with little or no effect on the rest o f the system with its associated robot. A machine work
system. As long as the bus has surplus capacity, the station control module accepts input commands o f
physical structure of the system can be reconfigured the form <MMACHINE P.ART x>. A material
wlth no changes requlred in the software resident in handling work station may accept commands o f the
Journal of Manufaccurlne S!sterns
i o l u r n e I. Vurnoer I
form <MOVE TRAY Y TO WORK STATION behavior ofthe work station to adapt to unexpected
0.The machine work station controller decom- conditions such as broken tools or defective or
poses t commands into sequences of subcommands
i missing pans.
to the machine controllen of the form <FETCH ~evera~ work station control units are organ -
P A R T X>, <INSERT X IN FIXTURE Y>, ized under and receive input commands from a cell
<EXECUTIVE CUTTING PROGRAM t>. control unit. The cell controller schedules jobs.
<CLEAR CHIPS>. etc. T h e material handling routes parts and tools t o the proper machines. ana
work station decomposes its commands into balances the workload among the work statlons
sequences o f subcommands o f the form <DIS- under its control. The cell controller makes sure that
PATCH C A R i A TO PICKUP STATION B>, each machlne has the proper tools at the proper tlme
etc. in both cases, the decomposition is performed to perform the required work on each pan.
in the context of feedback information that is passed Programs in the cell controller are also w n t t e n
through the factory status data base shown on the in state-cable form and can contain any number o f
left of Figure 7. rules for adapting to error conditions such as tool
The work station controllen may contain failures or changing pnoritles.
programs written in the form of state-tables, or Several cells could be orpnltrd under a shop
production rules. T h i s formulation w i l l allow the control unit. However, the A,MRF lnltlall> at least.
Journal of Manufacturing Syrtmu
Volume I. Number 1
w be considend as a single c, and hence only
l el queries to and from this data base enable manage -
one ci controller is planned. The possibility exists
tl ment to monitor and manage the whole factory by
for either further expansion of the AMRF or setting priorities or entering control parameters
emulation of other cells if the research task which alter the mode of operation o f the control
demands i. t hierarchy.
There are two data bases planned for the The sccond Kction of the Factory Status data
AMRF. On the right of figure 7 is a Part Data Basc base contains the status of tach machine tool and
which contains design data such as par?dimensions, robot. in the plant as well as the s t a t u s of each
d e s i d grip points for robot handling, group tech- cornputex in the control hierarchy: What program is
nology codes, and material and tooling require- ah
e c machine running? What step in the program?
ments. A second scction of the right-hand data base How long in that step? What part is being operated
contains proccss plans for routing and scheduling on?, etc.
and robot handling as well as cutter location data The third section of the Factory Status base
fa ncedcd for performing the various machining contains the s t a t u of each parr in progress. In this
operations. These process piam arc, in fact, the data base, there exists a data fiIe corresponding to
programs required at the various lcveir of the every part that gives the part name, the tray that is
control hierarchy in order to perfortu the neccssaq i
transporting it, t position and orientation in that
manufacturing operations. T s the right-hand tray or in the work station, t state of completion,
data base s in parr, a program library which
i, and a number of quality control parameters.
contains thecontrol programs needed by thecontrol All thee data bases are sewed by several
modules at the various levcis of the control 0)
Input/ Output (I/ controllers. The Factory Status
hierarchy. A third section of the righthand data ais0 has a hierarchy of feedback procrsson
base contains data related to fc+ds and speeds which cn
that s a the various leireis of the data base and
may be changed as a result of tensed conditions in extract the information needed by .[he control
the factory environment. modules at the next higher level. As in the
When a order is entered into the shop control
n microcomputer robot control network, information
module, the process plan to make that pan is d ed is passed from one level to. another, and from one
in from the right-hand data base. The pr-plan is computing module to another through the data base
hierarchically structured so that at the top there is which serves as a common memory. This makes the
only the name o f the process pian. This name is s n et System modular and defines the interface between
to the cell control. T h e cell control computer hs
modules to be the data base. T u , specification of
accesses the data base which calls in the sequence of the data base specified the principal interfaces of the
steps (i.e., the program) that is the process plan at control system. This means that as long as a robot or
the cell level. Each command ->inthis program is machine tool controller can read from and write to
passed in sequence to the next level down, which is the data base. it can be added to or deleted from a
the work station. As each cell output command system with a minimum o f impact on the other
enters the work station, it is the name o f a process components of the system. '
plan for the work station. The work station then Because the status data base will be updated at
goes to the part data base as its level and calls up the each time increment, i t will always contain a
sequence of instructions required to decompose that complete and current state description of the mire
proc.ss pian for the robot or for the machine tool. factory. T h i s will make it possible to restart the
The data that reside in the pan data base come system easily in the event bf a computer system
from an interactive design graphics system and an crash. I t will also be useful as a debugging tool.
interactive process planning system shown at the Activities of the various modules and o f the system
top right of Figure 7. variables themselves can be traced and recorded for
On the left o f Figure 7 is a second data base debugging, analysis. or optimization.
which contains dynamic Factory Status infonna - The control architecture has been described in
tion. This Factory Status data base is also divided considerable detail since it is this feature that most
into three parts. On the far left is a management clearly distinguishes the A M RF from "just another
information and control data base. Entnes or FMS". T h i s system wilI provide the modularity
Journal of Manufacturing S w e r n r
lolurne I . \umber I
needed to carry out the NBS research program in and the Depacment o f Commerce, and the support
interface standards and will eventually make FMS and encouragement of the machine tool community.
technology practical for many smaller shops. especially those who have joined in the program as
Rcscarch Associates. The support of the Depan -
ment of Defense at various points in the program
Concluding Remarks has been most helpful as has been the university
Although it w require a certain amount of
i community working with us both informally and
research to construct the AMRF and to test the under what is hoped to be an expanding grants
concepts on which it is based, the AMRF t e is not
sl program. The AMRF is truly a National effort.
considered a research projcct. As various portions
come on line, research projects, often with university
or private sector cooperation, will be started. Many Ref erencas
of these projccu will deal with n w and improved
sensors to monitor machine performance. Others I. Camema. J. M. and Huia. 0.. 'NBS Techntal Note 844'.
Nu~ond Bureau of S u n d u d t , Wuhlngton. D.C.. 1974
will deal with the problem of calibrating sensors so 2 Hacken. R Si- J.,
. . % Barchur. Lzr
Stein, P A ~ u w l s f h CIRP. V a l m 3 1977.
J.. R m e . C.. and
that the product dimensions (not the sensor
responses) arc traceable to National Standards. If
L. F Q U ~ I ~ . Kelly. J C. Kenntcoct. P .
~rld.M..Moorr.D..andWdlln~~onJ..(IGES YIJ.26M Response
Comnuua?. " gl D Repmmtatron for Communlacron of P T W U C ~
more than one machine is involved in the manufac - Definrtron Data'. Ropovd A m m a n Nauonal Standard. Engrnetnng
ture of a part so that the refnturing effecu the Dnwng a d Rebred Docruamutlon Pnctlcu. Approted 4\51
sundud. kpnaber 1981.
critical dimensions, this traceability becomes a .
4 Ham. 1 and Gonpwam. T 'Appliauon of Group Technolog
Concept for H i g h ProduclrvttyoCNBS ShopOpmtroru'. R W K tor
complex problem How both the mechanical opera- NBS. The Pennsyi- Sum L'nivrmty. licuvenrty Park. Pennsi -
tions and their supporting software a n validated van% 1981.
S For eumpk. the yvubk Mruior Manufaaunng System o i
opens v a t new areas for Measurement Assurance Cimnnau Mikron.
b Antomorion of S m d Batrh Produrrron. Vatronai Eng!ncenng
Program technology. Along with this metrology laboratory. East Kribndc. Ciugow. Scotland. 1978.
research will. go research on the detailed nature of 7. Hmhnl. M. E.. 'Worfd Trends rn Advanced L1anu:actunng
ma0 3 0
Technology'. P r ~1 Prh . I n 4 TrrJervrce Wunularrurtnp
. the data formats at each interface to determine how r
Technolo@ Confertnw. 0d.ndo. Ronda. 19ff
standards can be designed so as to neither compro - LSutton. G.,C~rman.Cfshnolog). o i Macnlae Tools "-4 Surrey
of the Succ-ot.che -An by the Machtnc Tool Task Force. Lawrence
mise proprietary interests, nor inhibit innovations L~vcrmore Laboratory. LCRL-52960. Volume 1-5, 1980
9 Gtllap~e.L. K.. Advanres m Deewrrnr. Socrny o i Manuiacunng
The A.MRF, like other Bureau facilities, will be En@- u
D br. Mlchtgra 1978.
made avallable to university and industrial groups 10. For example. the FW 5 System forengme blocks at Yawman Dtesel
has duardcd o p t r a l sno due to cast Iron dust.
for nonproprietary research in manufacturing 11. D e V n a , U F. and Agaplon. J..'A Study of Hach~nabtltt\ Data
Banks Adaptable to an 4utornattc Hanufactunng Test Facrltn'
engmetring which e funher afield than the Report for SBS. Cinlveotty o( Wlsconsm, Hadisan. Wlrconrtn. 1982.
metrology and standards of NBS. .
1 2 Dct'ncr. M. f and 4pp1on. J.. 'A Survey o[ Ch I p Brmkrnj
Tooling Adaptable to an Automattc Hanufacturrng Test FmIIt\
The AMR F is only one in a continuing series of Report for Y 8s. L'nlvenlty of Wlrconsm. Hadaan. Isconrln. 198:.
faclllties that permit NBS to fulfill i designated
t 13. Tuffensamms. K ClRP A n n u & . Volume M ? ) . 1981. pp 553
1 . Cook. S H.. Subnnnmn. K and Lsllc. 5. A.. Turvey o i :ne
role as the nation's measurement and standards Sute -of-the -An of Tool W a r Senrmg Tcchnqua ". prepared In
suppon wtn W F Cant UCI 03861. Publ. by Yaswchuwu Instttuce
laboratory. o Technology. 1975
IS. Albus. J S Barben. 4. J..anP Yagel. R. Y .'Theom and Pracrlco
of Hlcarchlul Control'. Ptorrrdrnqs. 3 r d lfff Cornpurer Socu*t
lnrrrnorlonal Conjrrmrr. Wuhmgron. D C. 19% . I
16. Albus. J S.. Barbcra. 4 J Fttrprrald. L1 L.. and hashman. J .
'Sensorv Intenctne Robots'. Anna13 0 ClRP Toronto. Canada.
The Automated Manufacturing Research 17. Albu. J S.. Barbm. 4 J . Fitrgcnld. J L.. ~ H t c r ~ r c h r c Cantto1
Facility' owes its existence to the foresighted for Senran* Intenctrve Robots'. Proczcdtngs. Iltn lnrrrrrartonol
SImpcrtui on lndwrrral Robots and Robor Erhtbrt. Tok!o. Japan.
management of the National Bureau o f Standards 1981.
'Contnbutlon of the Yarional Bureau o f Standards. Not sub~cc! I copyrrght.
Jounul of Manufanunnu Systems
Volumc I. Number I
Dr. John A. Simpson is presently Dimor of the Center for Manufacturing Engineering at the National
Bureau of Standards? (NBS), with background in the fields of dimensional metrology. electron optics.
photwpticai instrument design, and mechanics: Prior to his appointment to this position, Dr. Simpson
s e as Chief of the Mechanics Division, Deputy Chief o f the Optical Physics Division. and Chief of the
Electron Physiu Section, all at NBS, and was a Research Physicist at &high University.
Dr. Simpson is a FeiIow of the American Physical Society and has been active int Division of Electron
Physics. He has s n on the National Acldcmy of Scicnces/National Research Council Panel for NATO
Postdoctoral Fellowship. In 1975, he received the Department of Commerce Gold Medal award in
recognition for his accomplishments in modernizing the metrolow services of NBS. ln 1980, he received the
NBS Applied Rcsearch Award (jointly) for his part in the development and implemcntation of the automated,
self-correcting threc-axk coordinate meamring machine which enables rnanufmuren to characterize and
correct errors in machine tools during manufacturing processes.
Dr. Robcn J. Hocken is p m n t l y Chief of the Automated Production Technology Division at the
National Bureau of Standards, with background in the a m of critical phenomena. machine tool metrology.
thm4irnensional metrology, l s r optics, manufacturing technology, and polarimetry. His divis~on NBS at
develops and maintains competence in machine tool dynamics, precision engineering, robotics. and computer
aided manufacturing, and is the incorporation of metrology into the *
' metal working
processes, including the stan for integration of equipment up to the uring cell level.
Prior to appointment to this position, Dr. Hocken held a National Ruearch Council Postdoctoral position at
NBS. was Leader of the Dimensional Metrology Group, and Chief of the Dimensional Technology Section at
. . Dr. Hocken is a member of the American Physical Sociefy, the American Society for-Testing and
Materials, and the International Institute for Production Enginecling Research. He is a widely recognized
expert in production metrology and recipient of the Taylor Medal of CIRP for caritributions to metrology.
the Department of Commerce Silver Mcdd for thmdimensional metrology, the IR-100 Award for the
large-scale measuring machine. and the NBS Applied Research Award (jointly) for the development of the
three-axis measuring machine which enablqs manufacturers to characterize and corrcct errors in machine
tools during manufacturing processes.
Dr. James S. Altrru is presently Acting Chief o f the Industrial Systems Division and Manager o i the
Programmable Automation Section. Center for Manufacturing Enginering, National Bureau o f Standards.
He has receaved the Department of Commerce Silver Medai for his work in control theory and manipulator
design and the Industrial Research IR-100 award for his work in brain modeling and computer design. He is
the author of numerous aniclcs in technical journalsjncluding a survey article on robot systems for ScienlrJic
American (February 1976) and an entry to trcyclopedia Amerrconu on "Robors ". H e has wntten anlcles on
robotics for O M N I Magazine, lUeruf Working News. and BYTE Magazine. Dr. Albus has also been quoted in
orher national magazines. such as TIME, Forrune. Reader's Digest, NEXT, and Discover, and has appeared
in a number of TV interviews.
Before coming to the Bureau of Standards. he designed electr --optical systems for more than4 5 NASA
spacecraft. seven of which are on permanent display in the Smithsonian Air and Space Museum. For a shon
tune he served as program manager of the NASA Anificial Intelligence Program.
His latest book Bruins. Behavior and Robotics was publlshed t y McGraw -Hill in Yovember 198 I. Dr.
Albus has also written a book entitled Peoples 'Capiralism: The Economm of rhe Robot Revoluiion in which
he addresses some of the ctntral socral and econornlc issues raised by the advent oicomputer contrcllled robot
lure readmgs. and home positions. A rotational geometries to existing forgings accuracy and falls under the heading of
record is kept of each failure. and geometries, sornetlmes avoiding the deterministic metrology. Work on the
The computer -aided process planning need to retool, and cutting machining latterdeals with interfacestandards which
(CAPP) module is a generative program time if the process is near net shape. allow easy transfer of part information
based on parr codes. I t lets engineers Additional capabillties can be incor- between different manufacturers' equip-
porated into the CIM system, providing ment and easy replacement or upgrading
as-needcd flexibility. These capabilitles of design and manufacturing equipment.
may include a robotic sermetal painting Deterministic Metrology. Gage blocks
ic application of brazing alloys, and other artifact standards provide the
process t r o l
~ ~ ~ u t e ~ ~ ~ ~ parameters ~ basis of a mature technology which
and pan position in^ fur vacuum plasma assures accuracy by direct comparison of
depmition, and DNC Iaserdrilling. With dimensions. For complex parts, the
a system that has already increased computer -controlled coordinate measur-
productivity and reduced costs by as ing machine has successfully automated
much as 2% added capabilitics can only such comparisons, but it sil depends
enhance the AEBG CIM system. upon measuring the part itself.
In a welldesigned automated manu-
facturing process. if one good part is
produced and if the process parameters
(such as cutting force and temperature
distribution in the machine tool) are
controlled or corrected. then subsequent
parts cut by that same program are also
Manufacturing at NBS likely to be good. This observation under-
lies the philosophy of deterministic
metrology which conantrates on under-
standing. monitoring, and controlling
A n update on the National Bureau of Standards' the manufacturing process itself rather
than checking the part after cutting is
facility for automation research hs
finished. Tu, the standards' responsi-
bility which cl for careful custody of
master gage blocks and calibration of
other artifact blocks from thesc masters
DR. ROBERT J HOCKEN function of "tesring and calibration of is now advancing into a technology which
and standard measuring apparatus [and] the requires a fundamental understanding of
DR. PHILIP NANZETTA solution of probkms which arise in con- ways to monitor' and conttol basic
NBS Center for nection with standards. "Two major prob- cutting processes.
Manufacturing Engineering lem areas related to standards arc being Close attention to process control has
addressed in the AMRF. These concern led NBS scientists to work in the area of
I N LATE 1980, the National Bureau of interchangeability of parts and inter- software accuracy enhancement. By
Standards (NBS) made a decision to changeability of mnnufacturing units. developing an "error map" for a CNC
develop and construct an Automated Work on the former involws dimensional machine or coordinate measuring ma-
Manufacturing Research Facility(AMRn
by the m i d a s to support research on A horizontal machining center work nation within the Automated Manufacturing
machine interfacing and inspection of Research Facilit-v ofthr NBS i tendedby a Cincinnati Milacron robot.
parts produced in small batches.
The facility is designed to serve t h m
major sectors-industry, government.
and academia -in developing, testing.
funding. and implementing advances in
automated manufacturing. The A MRF is
a direct resource for members of the
manufacturing community who wish to
see demonstrations of existing and near-
future technology as an aid in their own
automation decisions. The project also
plays a crucial role in meeting the NBS'
legal commitment to leadership in stan-
dards and measurement activity, espe-
cially in a time of rapidly advancing
technology. Finally, this new facility
serves as a " test bed" for industrial
research associates, university workers,
andscientistsat theNBS whoarcprepar -
ing the way for the computer automation
technology of the next decade.
testing and Calibration
The legislation which created and gov-
erns the NBS specifically assigns to it the
08 October 1983/Mmufacturing Enginafing
chine and incorporating a correcting
algorithm Into 11s control computer. the
accuracy of standard equipment can be
Interface Standards. In the area of
interface standards. NBS scientists coor-
dinate and support work on a common
domain graphics exchange standard
known as the Initial Graphics Exchange
Standard (IG E s). This standard, the work
of an industry -wide committee, allows
the transfer of CAD data between the
systems of various manufacturers. Pre-
and postprocessors for ICES have b e en
announced or promised for all of the
major CAD systems.
American industry has reached a con-
sensus that it must increase its concentra -
tion on quality and productivity in order
to develop and hold our country's leading
position in international commerce.
Thus. the NBS has found strongindustry
support for development of the AMRF
from industry, universities. and other
agencies of the government concerned modular and designed so that it can be and vibration. High-precision dimen -
with manufacturing. implemented in steps. Under this design sional sensing is incorporated in tool-
Under the NBS research program, a approach, a shop or factory can under- setting stations which use linear variable
topic o f interest to an industrial sponsor take automation in steps which are eco- differential transformers (LVDT S) that
is investigated by a research associate nomically tolerable and still reasonably have been "hyperlinearized " by software
employed by the sponsoring firm and expect the various pieces to fit together correction techniques developed by
assigned to work at NBS. Frequently. a into an integrated production facility as N BS scientists.
program uses a machine tool which belongs more cornponenu am added to the system. Design of the control structure is
to the sponsor, but which is temporarily The AMRF itself i5 being constructed b a d on distributed computing power
incorporated into the AMRF. Such coop- in a similar steplike fashion. &fore which takesadvantage of r m n t
erative research activities serve to assure control system development advances to in microeltctronics. Because of thisdesign
that the AMRFcombiM placticaieconomic higher levels of demonstration. a working decision, a high level of computational
solutions to near-term problems with the unit consisting of a single machining capacity can be economically incorpo -
capacity to advance the state of the art. center, its robot tender, fixturing. mater- rated into sensory systems and other ele-
Developing the AMRF '
ial handling interfaces, and various ments low in the control hierarchy. For
sensory systems is fully developed and example. a dedicated safety system with
Developmental planning, procure - tested as a freestanding work station its own controller is being incorporated
ment,and teotingofthiemanyelementsof driven by manual or simulated hlgher- into each work station in order to provide
t h e A M ~ ~ % toextend wellinto
~ e ~ ~ t ~ k v c l commands. - redundant safety checking for the protec -
1986. ~ ~ ~hasalready begun
h ~ i
~ Coardinated ~ control yof several work tion of both humans and machina
hr elcmmts of the laciiity az ,
s ~ t d ~ l ~~ g ~ the moment, with a
j Using Avaihbk Equipment. in con-
incorporated. The AMRF, &$ cornm~nfamily af parts will be main- trast with the approach adopted in the
presently defined, consists British ASP Planand the Japanese FMC
machining centers (a vertical Project, the NBS is using standard,
machining center, a horizontal modern, commercially available machine
machining center, a large tunring center, tools in the A MRF. The ASP and FMC
and a smaller turning center), B c Interconnections. A major a
approaches c l for the design of new
and deburring station, an automatic objmive of the A MRF is to study the machine tools for automation. The NBS,
inspection station, and a material hand- interface problems which arise when based on reviews of published studies of
ling complex. The two machiningccntm. the international state of the art in
one of the turning centers, and rwo machine tool design, has chosen instead
robots are present now on the shop floor, into an integrated facility. In the process to rely upon the engineering experience
the automatic inspection station w bc l
i of solving such interface problems. NBS of a mature machine tool industry and to
installed by late 1983. scientists can propose changes in equip- avoid radical design changes for the
ment to improve interconnection and AMRF machining centers and other
control. Work with softwan ascuracy facility hardware.
~ n h ~ n has defined a twhole range
~ ~ c ~ The AMRF is meant to be a national
lnstrumt Shop. Anaddittons1 I , O f*
OO Ot of new interface requirements for machine project which draws upon and contri -
(930 mz) off the main shop floor is tool controllers. In a slmilar way, butes to work on automation of small
utilized for a toolcrib area, computer development of off-line programming batch manufacturing being conducted by
space, electronic and mechanical support and ml-timecontrol for robots has given Industry, universities, and other labora -
laboratories, and a small conference new insight into the need for more tories. While theeffortsofthe NBS relate
training facility. sophisticated robot controller interfaces. to problems ofstandards, the bureaualso
MduLr Architecture. An early deci- Various sensor types are being em- encourages close cooperation wlth those
slon in planning for the AMRF determined ployed in the design of the AMRF, mclud- who are addressing other components o f
that the control architecture must be ing force. proximity, vision, temperature, the field.
Manufacturing Engineering/October 1983 69