Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

Lasky, T.A., Hsia, T.C., Tummala, R.L., Odrey, N.G.

Robotics
The Electrical Engineering Handbook
Ed. Richard C. Dorf
Boca Raton: CRC Press LLC, 2000
2000 by CRC Press LLC
101
RolofIcs
101.1 Robot Confguiation
Caitesian Confguiation Cylindiical Confguiation Spheiical
Confguiation Aiticulated Confguiation SCARA
Confguiation Gantiy Confguiation Additional Infoimation
101.2 Dynamics and Contiol
Independent Joint Contiol of the Robot Dynamic Models
Computed Toique Methods Adaptive Contiol Resolved
Motion Contiol Compliant Motion Flexible Manipulatois
101.3 Applications
Justifcation Implementation Stiategies Applications in
Manufactuiing Emeiging Issues
101.1 Rubut Cunhguratiun
Ty A. Io|y ond Ten C. Ho
Confguraon is a fundamental classifcation foi industiial iobots. Confguiation iefeis to the geometiy of the
iobot manipulatoi, i.e., the mannei in which the links of the manipulatoi aie connected at each joint. The
Robotic Industiies Association (RIA) defnes a iobot as a manu|aor JesgneJ o moe maera|, ars, oo|s,
or seta|:eJ Jetes, |roug| ara||e rogrammeJ moons [or |e er[ormante o[ a arey o[ as|s. With this
defnition, attention heie is focused on industiial manipulatoi aims, typically mounted on a fxed pedestal
base. Mobile iobots and haid automation e.g., Computei Numeiical Contiol (CNC) machines] aie excluded.
The emphasis heie is on seiial-chain manipulatoi aims, which consist of a seiial chain of linkages, wheie each
link is connected to exactly two othei links, with the exception of the fist and last, which aie connected to
only one othei link. Additionally, the fist thiee links, called the majoi linkages, aie focused on, with only a
biief mention of the last thiee links, oi wiist joints, also called the minoi linkages.
Robot confguiation is an impoitant consideiation in the selection of a manipulatoi. Confguiation iefeis
to the way the manipulatoi links aie connected at each joint. Each link will be connected to the subsequent
link by eithei a lineai (sliding oi piismatic) joint, which can be abbieviated with a P, oi a ievolute (oi iotaiy)
joint, abbieviated with an R. Using this notation, a iobot with thiee ievolute joints would be abbieviated as
RRR, while one with a iotaiy joint followed by two lineai (piismatic) joints would be denoted RPP. Each
confguiation type is well suited to ceitain types of tasks and ill suited to otheis. Some confguiations aie moie
veisatile than otheis. In addition to the geometiical consideiations, iobot confguiation affects the stiuctuial
stiffness of the iobot, which may be an impoitant consideiation. Also, confguiation impacts the complexity
of the foiwaid and inveise kinematics, which aie the mappings between the iobot actuatoi (joint) space, and
the Caitesian position and oiientation of the iobot end-effectoi, oi tool.
Theie aie six majoi iobot confguiations commonly used in industiy. Details foi each confguiation aie
piesented in subsequent subsections. The simplest confguiation is the Caitesian iobot, which consists of thiee
oithogonal, lineai joints (PPP), so that the iobot moves in the x, y, and : diiections in the joint space. The
Ty A. LasIy
Inverry of Co|forno, Dov
TIen C. HsIa
Inverry of Co|forno, Dov
R. LaI TummaIa
Mc|gon Srore Inverry
IchoIas C. drey
Ie|g| Inverry
2000 by CRC Press LLC
cylindiical confguiation consists of one ievolute and two lineai
joints (RPP), so that the iobot joints coiiespond to a cylindiical
cooidinate system. The spheiical confguiation consists of two iev-
olute joints and one lineai joint (RRP), so that the iobot moves in
a spheiical, oi polai, cooidinate system. The aiticulated (aim-and-
elbow) confguiation consists of thiee ievolute joints (RRR), giving
the iobot a somewhat human-like iange of motion. The SCARA
(Selectively Compliant Assembly Robot Aim) confguiation con-
sists of two ievolute joints and one lineai joint (RRP), aiianged in
a diffeient fashion than the spheiical confguiation. It may also be
equipped with a ievolute joint on the fnal sliding link. The gantiy confguiation is essentially a Caitesian
confguiation, with the iobot mounted on an oveihead tiack system. One can also mount othei iobot confg-
uiations on an oveihead gantiy system to give the iobot an extended woikspace, as well as fiee up valuable
factoiy ooi space. The peicentage usage of the fist fve confguiation types is listed in Table 101.1. This table
does not include gantiy iobots, which aie assumed to be included in the Caitesian categoiy. Additionally, this
infoimation is fiom 1988, and does not accuiately iepiesent cuiient usage.
In geneial, iobots with a iotaiy base have a speed advantage. Howevei, they have moie vaiiation in iesolution
and dynamics compaied to Caitesian iobots. This can lead to infeiioi peifoimance if a fxed contiollei is used
ovei the iobot`s entiie woikspace.
Cartesian Cunhguratiun
The Caitesian confguiation consists of thiee oithogonal, lineai
axes, abbieviated as PPP, as shown in Fig. 101.1. Thus, the joint
space of the iobot coiiesponds diiectly with the standaid iight-
handed Caitesian xy:-cooidinate system, yielding the simplest
possible kinematic equations. The woik envelope of the Caite-
sian iobot is shown in Fig. 101.2. The woik envelope encloses all
the points that can be ieached by the iobot aim oi the mounting
point foi the end-effectoi oi tool. The aiea ieachable by an end
effectoi oi tool is not consideied pait of the woik envelope. All
inteiaction with othei machines, paits, oi piocesses must take
place within this volume Ciitchlow, 1985]. Heie, the woikspace
of a iobot is assumed to be equivalent to the woik envelope.
Theie aie seveial advantages to this confguiation. As noted
above, the iobot is kinematically simple, since motion on each
Caitesian axis coiiesponds to motion of a single actuatoi. This
eases the piogiamming of lineai motions. In paiticulai, it is easy
to do a stiaight veitical motion, the most common motion in
assembly tasks. The Caitesian geometiy also yields a constant
aim iesolution thioughout the woikspace; i.e., foi any confgu-
iation, the iesolution foi each axis coiiesponds diiectly to the
iesolution foi that joint. The simple geometiy of the Caitesian
iobot leads to coiiespondingly simple manipulatoi dynamics.
The disadvantages of this confguiation include inability to ieach objects on the ooi oi points invisible fiom
the base of the iobot, and slow speed of opeiation in the hoiizontal plane compaied to iobots with a iotaiy
base. Additionally, the Caitesian confguiation iequiies a laige opeiating volume foi a ielatively small woikspace.
Caitesian iobots aie used foi seveial applications. As noted above, they aie well suited foi assembly opeia-
tions, as they easily peifoim veitical stiaight-line inseitions. Because of the ease of stiaight-line motions, they
aie also well suited to machine loading and unloading. They aie also used in clean ioom tasks.
FIGURE 101.1 The Caitesian confguiation.
(Sourte. T. Owen, ssem||y w| Ro|os, Engle-
wood Cliffs, N.J.: Pientice-Hall, 1985. With pei-
mission.)
TABLE 101.1 Robot Aim Geometiy Usage
Aim Geometiy Peicent of Use
Caitesian 18
Cylindiical 15
Spheiical 10
Aiticulated 42
SCARA 15
Sourte. V. D. Hunt, Ro|ots Sourte|oo|, New
Yoik: Elseviei, 1988. With peimission.
2000 by CRC Press LLC
Cy!indrica! Cunhguratiun
The cylindiical confguiation consists of one veitical ievolute joint and
two oithogonal lineai joints (RPP), as shown in Fig. 101.3. The iesulting
woik envelope of the iobot is a cylindiical annulus, as shown in
Fig. 101.4. This confguiation coiiesponds with the cylindiical cooidi-
nate system.
As with the Caitesian iobot, the cylindiical iobot is well suited foi
stiaight-line veitical and hoiizontal motions, so it is useful foi assembly
and machine loading opeiations. It is capable of highei speeds in the
hoiizontal plane due to the iotaiy base. Howevei, geneial hoiizontal
stiaight-line motion is moie complex and coiiespondingly moie diff-
cult to cooidinate. Additionally, the end-point iesolution of the cylin-
diical iobot is not constant but depends on the extension of the
hoiizontal linkage. A cylindiical iobot cannot ieach aiound obstacles.
Additionally, if a monomast constiuction is used on the hoiizontal
linkage, then theie can be cleaiance pioblems behind the iobot.
Spherica! Cunhguratiun
The spheiical (oi polai) confguiation consists of two ievolute joints
and one lineai joint (RRP), as shown in Fig. 101.5. This iesults in a set
of joint cooidinates that matches with the spheiical cooidinate system.
A typical woik envelope foi a spheiical iobot is shown in Fig. 101.6.
FIGURE 101.2 Caitesian iobot woik envelope.
FIGURE 101.4 Cylindiical iobot woik envelope.
max
min min
max
max
min
FIGURE 101.3 The cylindiical con-
fguiation. (Sourte. T. Owen, ssem||y
w| Ro|os, Englewood Cliffs, N.J.:
Pientice-Hall, 1985. With peimission.)
max
max
min
min
max
min
0
2000 by CRC Press LLC
Spheiical iobots aie typically heavy-duty iobots. They have the advan-
tages of high speed due to the iotaiy base, and a laige woik volume, but
aie moie kinematically complex than eithei Caitesian oi cylindiical iobots.
Geneially, they aie used foi heavy-duty tasks in, foi example, automobile
manufactuiing. They do not have the dexteiity to ieach aiound obstacles
in the woikspace. Spheiical iobots also do not have fxed iesolution
thioughout the woikspace.
Articu!ated Cunhguratiun
The aiticulated (oi anthiopomoiphic, jointed, aim-and-elbow) confgu-
iation consists of thiee ievolute joints (RRR), as shown in Fig. 101.7. The
iesulting joint cooidinates do not diiectly match any standaid cooidinate
system. A slice of a typical woik envelope foi an aiticulated iobot is shown
in Fig. 101.8.
The aiticulated iobot is cuiiently the most commonly used in ieseaich.
It has seveial advantages ovei othei confguiations. It is closest to dupli-
cating the motions of a human assemblei, so theie should be less need to
iedesign an existing woikstation to utilize an aiticulated iobot. It has a veiy laige, dexteious woik envelope;
i.e., it can ieach most points in its woik envelope fiom a vaiiety of oiientations. Thus, it can moie easily ieach
aiound oi ovei obstacles in the woikspace oi into paits oi machines. Because all the joints aie ievolute, high
FIGURE 101.5 The spheiical confguiation. (Sourte. T. Owen, ssem||y w| Ro|os, Englewood Cliffs, N.J.: Pientice-Hall,
1985. With peimission.)
FIGURE 101.6 Spheiical iobot woik envelope.
max
max
min
min
max
min
0
FIGURE 101.7 The aiticulated
confguiation. (Sourte. T. Owen,
ssem||y w| Ro|os, Englewood
Cliffs, N.J.: Pientice-Hall, 1985. With
peimission.)
2000 by CRC Press LLC
speeds aie possible. The aiticulated aim is good foi tasks involving multiple inseitions, complex motions, and
vaiied tool oiientations. The veisatility of this confguiation makes it applicable to a vaiiety of tasks, so the
usei has fewei limitations on the use of the iobot. Howevei, the same featuies that give this iobot its advantages
lead to ceitain disadvantages. The geometiy is complex, and the iesulting kinematic equations aie quite intiicate.
Stiaight-line motion is diffcult to cooidinate. Contiol is geneially moie diffcult than foi othei geometiies,
with associated inciease in cost. Heie again, aim iesolution is not fxed thioughout the woikspace. Additionally,
the dynamics of an aiticulated aim vaiy widely thioughout the woikspace, so that peifoimance will vaiy ovei
the woikspace foi a fxed contiollei. In spite of these disadvantages, the aiticulated aim has been applied to a
wide vaiiety of ieseaich and industiial tasks, including spiay painting, clean ioom tasks, machine loading, and
paits-fnishing tasks.
SCARA Cunhguratiun
The SCARA (Selectively Compliant Assembly Robot Aim) confguiation consists of two ievolute joints and a
lineai joint (RRP), as shown in Fig. 101.9. This confguiation is signifcantly diffeient fiom the spheiical
confguiation, since the axes foi all joints aie always veitical. In addition to the fist thiee degrees of freedom
(DOF), the SCARA iobot will often include an additional iotation about the last veitical link to aid in oiientation
of paits. The woik envelope of the SCARA iobot is illustiated in Fig. 101.10. The SCARA confguiation is the
newest of the confguiations discussed heie, and was developed by Piofessoi Hiioshi Makino of Yamanashi
Univeisity, Japan.
FIGURE 101.8 Aiticulated iobot woik envelope.
FIGURE 101.9 The SCARA confguiation. (Sourte. T. Owen, ssem||y w| Ro|os, Englewood Cliffs, N.J.: Pientice-Hall,
1985. With peimission.)
max
min max
max
min =
0.0
min
0
2000 by CRC Press LLC
This confguiation has many advantages and is quite populai in industiy. The confguiation was designed
specifcally foi assembly tasks Tiuman, 1990], so has distinct advantages when applied in this aiea. Because
of the veitical oiientation of the joints, giavity does not affect the dynamics of the fist two joints. In fact, foi
these joints, the actuatois can be shut off and the aim will not fall, even without the application of biakes. As
the name SCARA implies, this allows compliance in the hoiizontal diiections to be selectively vaiied; theiefoie,
the iobot can comply to hoiizontal foices. Hoiizontal compliance is impoitant foi veitical assembly opeiations.
Because of the veitical lineai joint, stiaight-line veitical motions aie simple. Also, SCARA iobots typically have
high positional iepeatability. The ievolute joints allow high-speed motion. On the negative side, the iesolution
of the aim is not constant thioughout the woikspace, and the kinematic equations aie ielatively complex. In
addition, the veitical motion of the SCARA confguiation is typically quite limited. While the SCARA iobot
can ieach aiound objects, it cannot ieach ovei them in the same mannei as an aiticulated aim.
Gantry Cunhguratiun
The gantiy confguiation is geometiically equiva-
lent to the Caitesian confguiation, but is sus-
pended fiom an oveihead ciane and typically can
be moved ovei a laige woikspace. It consists of
thiee lineai joints (PPP), and is illustiated in
Fig. 101.11. In teims of woik envelope, it will have
a iectangulai volume that sweeps out most of the
innei aiea of the gantiy system, with a height lim-
ited by the length of the veitical mast, and the
headioom above the gantiy system. One consid-
eiation in the selection of a gantiy iobot is the
type of veitical linkage employed in the : axis. A
monomast design is moie iigid, yielding tightei
toleiances foi iepeatability and accuiacy, but
iequiies signifcant headioom above the gantiy to
have a laige iange of : axis motion. On the othei
hand, a telescoping linkage will iequiie signif-
cantly less headioom but is less iigid, with coiie-
sponding ieduction in iepeatability and accuiacy.
Othei iobot confguiations can be mounted on
gantiy systems, thus gaining many of the advan-
tages of this geometiy.
FIGURE 101.10 SCARA iobot woik envelope.
Horizontal
Reach
Work
Envelope
Swing
Horizontal
Stroke
Horizontal
Reach
Plan View
Elevation
Work Envelope
Vertical
Stroke
Vertical
Reach
FIGURE 101.11 The gantiy confguiation. (Sourte. T. Owen,
ssem||y w| Ro|os, Englewood Cliffs, N.J.: Pientice-Hall, 1985.
With peimission.)
2000 by CRC Press LLC
Gantiy iobots have many advantageous piopeities. They aie geometiically simple, like the Caitesian iobot,
with the coiiesponding kinematic and dynamic simplicity. Foi the same ieasons, the gantiy iobot has a constant
aim iesolution thioughout the woikspace. The gantiy iobot has bettei dynamics than the pedestal-mounted
Caitesian iobot, as its links aie not cantileveied. One majoi advantage ovei ievolute-base iobots is that its
dynamics vaiy much less ovei the woikspace. This leads to less vibiation and moie even peifoimance than
typical pedestal-mounted iobots in full extension. Gantiy iobots aie much stiffei than othei iobot confguia-
tions, although they aie still much less stiff than Numeiical Contiol (NC) machines. The gantiy iobot can
stiaddle a woikstation, oi seveial woikstations foi a laige system, so that one gantiy iobot can peifoim the
woik of seveial pedestal-mounted iobots. As with the Caitesian iobot, the gantiy iobot`s simple geometiy is
similai to that of an NC machine, so technicians will be moie familiai with the system and iequiie less tiaining
time. Also, theie is no need foi special path oi tiajectoiy computations. A gantiy iobot can be piogiammed
diiectly fiom a Computei-Aided Design (CAD) system with the appiopiiate inteiface, and stiaight-line motions
aie paiticulaily simple to piogiam. Laige gantiy iobots have a veiy high payload capacity. Small, table-top
systems can achieve lineai speeds of up to 40 in./s (1.025 m/s), with a payload capacity of 5.0 lb (2.26 kg),
making them suitable foi assembly opeiations. Howevei, most gantiy iobot systems aie not as piecise as othei
confguiations, such as the SCARA confguiation. Additionally, it is sometimes moie diffcult to apply a gantiy
iobot to an existing woikstation, as the woikpieces must be biought into the gantiy`s woik envelope, which
may be haidei to do than foi a pedestal-mounted manipulatoi.
Gantiy iobots can be applied in many aieas. They aie used in the nucleai powei industiy to load and unload
ieactoi fuel iods. Gantiy iobots aie also applied to mateiials-handling tasks, such as paits tiansfei, machine
loading, palletizing, mateiials tianspoit, and some assembly applications. In addition, gantiy iobots aie used
foi piocess applications such as welding, painting, diilling, iouting, cutting, milling, inspection, and nonde-
stiuctive testing.
The gantiy iobot confguiation is the fastest-giowing segment of the iobotics industiy. While gantiy iobots
accounted foi less than 5% of the units shipped in 1985, they aie piojected to account foi about 30% of the
iobots by the end of the 1990s. One ieason foi this is summed up in Long 1990], which contains much moie
infoimation on gantiy iobots in geneial:
Cuiiently gantiy iobot cells aie being set up which allow manufactuieis to place a sheet of mateiial in
the gantiy`s woik envelope and begin automatic cutting, tiimming, diilling, milling, assembly and fnishing
opeiations which completely manufactuie a pait oi subassembly using quick-change tools and pio-
giammed subioutines.
Additiuna! Inlurmatiun
The above six confguiations aie the main types cuiiently used in industiy. Howevei, theie aie othei confgu-
iations used in eithei ieseaich oi specialized applications. Some of these confguiations have found limited
application in industiy and may become moie pievalent in the futuie.
All the above confguiations aie seiial-chain manipulatois. An alteinative to this common appioach is the
paiallel confguiation, known as the Stewait platfoim Waldion, 1990]. This manipulatoi consists of two
platfoims connected by thiee piismatic linkages. This aiiangement yields the full six DOF motion (thiee
position, thiee oiientation) that can be achieved with a six-axis seiial confguiation but has a compaiably veiy
high stiffness. It is used as a motion simulatoi foi pilot tiaining and viitual ieality applications. The negative
aspects of this confguiation aie its ielatively iestiicted motion capability and geometiic complexity.
The above confguiations aie iestiicted to a single manipulatoi aim. Theie aie tasks that aie eithei diffcult
oi impossible to peifoim with a single aim. With this iealization, theie has been signifcant inteiest in the use
of multiple aims to peifoim cooidinated tasks Bonitz and Hsia, 1996]. Possible applications include caiiying
loads that exceed the capacity of a single iobot, and assembling objects without special fxtuiing. Multiple aims
aie paiticulaily useful in zeio-giavity enviionments. While theie aie signifcant advantages to the use of multiple
iobots, the complexity, in teims of kinematics, dynamics, and contiol, is quite high. Howevei, the use of multiple
iobots is opening new aieas of application foi iobots.
2000 by CRC Press LLC
Typical industiial iobots have six oi fewei DOF. With six DOF, the iobot can, within its woik envelope,
ieach aibitiaiy positions and oiientations. At the edge of the woik envelope, a six-DOF iobot can attain only
one oiientation. To inciease the geometiic dexteiity of the manipulatoi, it is useful to considei iobots with
moie than six DOF, i.e., iedundant iobots. These iobots aie highly dexteious and can use the extia DOF in
many ways: avoidance obstacle, joint toique minimization, kinematic singularity (points wheie the manipulatoi
cannot move in ceitain diiections) avoidance, biacing stiategies wheie pait of the aim is biaced against a
stiuctuie, which iaises the lowest stiuctuial iesonant fiequency of the aim, etc. While the redundant manip-
ulator confguiation has many desiiable piopeities, the geometiic complexity has limited theii application in
industiy.
Foi any of the six standaid iobot confguiations, the oiientation capability of the majoi linkages is seveiely
limited. Thus, it is ciitical to piovide additional joints, known as the minoi linkages, to piovide the capability
of vaiied oiientations foi a given position. Most iobots include a thiee-DOF ievolute joint wiist that is connected
to the last link of the majoi linkages. The thiee ievolute axes will be oithogonal and will usually inteisect in a
common point, known as the wiist centei point. Then, the kinematic equations of the manipulatoi can be
paititioned into locating the Caitesian position of the wiist centei point and then deteimining the oiientation
of a Caitesian fiame fxed to the wiist axes.
Cunc!usiuns
Each of the six standaid confguiations has specifc advantages and disadvantages. When choosing a manipulatoi
foi a task, the piopeities of the manipulatoi geometiy aie one of the most impoitant consideiations. If the
manipulatoi will be used foi a wide vaiiety of tasks, one may need to tiade off peifoimance foi any given task
foi the exibility that will allow the manipulatoi to woik foi the vaiious tasks. In such a case, a moie exible
geometiy should be consideied. The futuie of iobotics will be inteiesting. With the steady inciease in compu-
tational capabilities, the moie complex geometiies, including iedundant and multiple iobots, aie beginning to
see incieased applications in industiy.
Dehning Terms
Degrees of freedom: The numbei of degiees of fieedom (DOF) of a manipulatoi is the numbei of independent
position vaiiables that must be specifed in oidei to locate all paits of the manipulatoi. Foi a typical
industiial manipulatoi, the numbei of joints equals the numbei of DOF.
Kinematics: The kinematics of the manipulatoi iefeis to the geometiic piopeities of the manipulatoi. ForwarJ
|nemats is the computation of the Caitesian position and oiientation of the iobot end-effectoi given
the set of joint cooidinates. Inerse |nemats is the computation of the joint cooidinates given the
Caitesian position and oiientation of the end-effectoi. The inveise kinematic computation may not be
possible in closed foim, may have no solution, oi may have multiple solutions.
Redundant manipulator: A iedundant manipulatoi contains moie than six DOF.
Singularity: A sngu|ary is a location in the woikspace of the manipulatoi at which the iobot loses one oi
moie DOF in Caitesian space, i.e., theie is some diiection (oi diiections) in Caitesian space along which
it is impossible to move the iobot end-effectoi no mattei which iobot joints aie moved.
Re!ated Tupic
101.2 Dynamics and Contiol
Relerences
R.G. Bonitz and T.C. Hsia, Inteinal foice-based impedance contiol foi coopeiating manipulatois," IEEE
Transatons on Ro|ots anJ uomaon, Feb. 1996.
J.J. Ciaig, InroJuton o Ro|ots. Met|ants anJ Conro|, Reading, Mass.: Addison-Wesley, 1986.
A. J. Ciitchlow, InroJuton o Ro|ots, New Yoik: Macmillan, 1985.
2000 by CRC Press LLC
E. Long, Gantiy iobots," in Contse Inernaona| Entyt|oeJa o[ Ro|ots, R. C. Doif, Ed., New Yoik: Wiley-
Inteiscience, 1990.
R. Tiuman, Component assembly onto piinted ciicuit boaids," in Contse Inernaona| Entyt|oeJa o[ Ro|ots,
R. C. Doif, Ed., New Yoik: Wiley-Inteiscience, 1990.
K. J. Waldion, Aim design," in Contse Inernaona| Entyt|oeJa o[ Ro|ots, R. C. Doif, Ed., New Yoik: Wiley-
Inteiscience, 1990.
Further Inlurmatiun
The jouinal IEEE Transatons on Ro|ots anJ uomaon is a valuable souice foi a wide vaiiety of iobotics
ieseaich topics, occasionally including new iobot confguiations. Additionally, IEEE`s Conro| Sysems Maga:ne
occasionally publishes an issue devoted to iobotic systems. The home page foi the IEEE Robotics and Auto-
mation Society can be found at http://www.acim.usi.edu/RAS/".
Anothei jouinal that often has iobotics-ielated aiticles is the ASME Journa| o[ Dynamt Sysems, Measuremen
anJ Conro|.
An additional souice of iobotics infoimation is T|e ProteeJngs o[ |e IEEE Inernaona| Con[erente on
Ro|ots anJ uomaon. This confeience is held annually.
Useful souices on the Woild Wide Web include the Ro|ots Inerne Resourtes page, located at http://piglet.cs.
umass.edu:4321/iobotics.html", and Ro|ots Resourtes, located at http://www.eg.bucknell.edu/~iobot-
ics/iiic.html". Consult youi system administiatoi foi infoimation on this web access.
101.2 Dynamics and Cuntru!
F. Io| Tummo|o
The piimaiy puipose of the iobot contiol system is to issue commands to joint actuatois to faithfully execute
a planned tiajectoiy in the tool space. This may involve position contiol when the manipulatoi is following a
tiajectoiy thiough fiee space oi a combination of position and foice contiol if the manipulatoi is to ieact
continuously to contact foices at the tool oi end-effectoi.
Contiol systems can opeiate eithei in open loop oi closed loop. In open-loop systems, the output has no
effect on the input. On the othei hand, closed-loop systems continuously sense the output and make appiopiiate
adjustments to the input in oidei to keep the output at the desiied level.
The majoiity of the cuiient industiial iobots use the independent joint control method and close the loop
aiound the joints of the iobot. The desiied joint positions coiiesponding to a tool tiajectoiy aie eithei taught
by using a eat| |ox oi geneiated by solving an inveise kinematics pioblem. The independent joint contiol
method, howevei, is effective only at low speeds. As the speeds inciease, the coupling effects between the joints
inciease and waiiant the inclusion of these effects in the contiollei development. Advanced contiollei devel-
opment and implementation based on full dynamics is one of the active aieas of cuiient ieseaich. New advances
in sensoi technology, fastei computeis, advanced iobots such as diiect diive aims and industiial competition
piovide new oppoitunities and motivation foi acceleiating the development and implementation of advanced
contiolleis foi iobots in the neai futuie.
Independent Juint Cuntru! ul the Rubut
The independent joint contiol method assumes that a single joint of a iobot is moving while all the othei joints
aie fxed. A typical joint position contiol system is shown in Fig. 101.12, wheie the actuatoi used is a dc
seivomotoi Luh, 1983]. In geneial, any one oi a combination of electiic motois and hydiaulic oi pneumatic
pistons can be used to move the joint thiough the desiied positions. These motois may be connected diiectly
to the joint oi indiiectly thiough geais, chains, cables, oi lead sciews. The desiied joint positions that aie inputs
to the position loops aie obtained fiom the tiajectoiy plannei. The actual position of the joint is obtained by
using a position sensoi, such as a potentiometei oi an optical encodei. An amplifei is used foi incieasing the
system gain, denoted by K
a
. The velocity feedback K

is used to ieinfoice the effect of back emf foi contiolling


2000 by CRC Press LLC
the damping of the system. This can be done eithei using a tachometei oi computing the diffeience in angulai
displacements of the actuatoi shaft ovei a fxed time inteival.
The design of the contiol system involves fxing the values of K
a
and K

to achieve the desiied iesponse.


Considei the closed-loop tiansfei function of the system shown in Fig. 101.12 (assuming nT
l
0),
(101.1)
wheie K
a
gain of the amplifei, K
T
toique constant of the motoi, K
|
back emf constant, K

position
sensoi constant (volts/iad), R iesistance of the motoi winding (ohms), and n geai iatio.
L
link position
(iad) and
m
angulai displacement at the actuatoi side (iad).
The effective ineitia, J
eff
, and damping, B
eff
, aie defned as
J
eff
J
m
- n
2
J
L
(101.2)
and
B
eff
B
m
- n
2
B
L
(101.3)
wheie J
m
total ineitia on the motoi side, B
m
damping coeffcient at the motoi side, J
L
ineitia of the iobot
link, and B
L
damping coeffcient at the load side.
This is a second-oidei system and stable foi all values of K
a
and K

. The values of K
a
and K

aie selected to
achieve a desiied tiansient iesponse by fxing the damping iatio and the natuial fiequency of the system and
aie desciibed below.
The chaiacteiistic equation foi the above system is
(101.4)
This can be conveniently wiitten as
(101.5)
wheie the natuial fiequency u
n
and the damping iatio of the system aie given as
FIGURE 101.12 Closed-loop contiol of a iobot joint. (Sourte. Adapted fiom J. Y. S. Luh, Conventional contiollei design
foi industiial iobots: A tutoiial," IEEE Trans. Sysems, Man, Cy|ernets, vol. SMC-13, no. 3, June 1983. C 1983 IEEE.)

1
2
( )
( ) ( )
s
s
nK K K
s RJ s RB K K K K K nK K K
J
a T
T | a T T a

+ + + +
eff eff
s s
RB K K K K K
RJ
nK K K
RJ
T | a T T a 2
0 +
+ +
+
eff
eff eff

s s
n n
2 2
2 0 + + u u
2000 by CRC Press LLC
(101.6)
(101.7)
These systems aie designed to opeiate with ciitical damping ( 1) because an undeidamped system ( < 1)
has fast iesponse but iesults in an oveishoot, wheieas an oveidamped system ( > 1) is too slow. Howevei, this
is not always possible, because the damping iatio given by Eq. (101.7) depends on B
eff
and J
eff
which vaiy duiing
the actual opeiation of the manipulatoi. B
eff
changes with age oi iepeated use of the manipulatoi. J
eff
vaiies with
the payload. Foi example, the vaiiation of J
eff
foi the Stanfoid manipulatoi undei vaiious loading conditions
is shown in Fig. 101.13. J
eff
also vaiies with the confguiation of the manipulatoi duiing the actual opeiation.
So a compiomise solution will be to design the contiollei such that > 1 thioughout the intended opeiation.
The undamped natuial fiequency u
n
is selected to be no moie than half the iesonance fiequency of the
iobot to avoid any stiuctuial damage to the iobot Paul, 1981]. These iesonances aie possible due to the
exibilities associated with the links of the iobot and the shafts within the diive system, to name a few. These
aie called unmoJe|eJ iesonances because they aie not explicitly included in the model. In oui case, if K
eff
and
J
eff
aie the effective stiffness and the ineitias of the joint, iespectively, then the iesonance fiequency u
r
is given by
(101.8)
Since K
eff
is diffcult to estimate but constant foi a given joint, we can expeiimentally deteimine the iesonance
fiequencies foi a known ineitia and use this infoimation foi fxing the gain. Suppose u is the iesonance
fiequency foi a given value of effective ineitia J, then
FIGURE 101.13 Vaiiations of link ineitias foi JPL-Stanfoid manipulatoi. (Sourte. A.K. Bejczy, Jet Piopulsion Lab, Pasadena,
Calif., Ameiican Automatic Contiol Confeience Tutoiial Woikshop, Washington, D.C., June 18, 1982.)
u

n
a T
nK K K
RJ
>
eff
0

+ + RB K K K K K
nK K RJ K
T | a T
a T
eff
eff
2
u
r
K
J

eff
eff
2000 by CRC Press LLC
(101.9)
To minimize the effects of unmodeled iesonances, we use
(101.10)
The selection of K
a
and K

depends on selecting and u


n
. Using Eqs. (101.6) and (101.10), we can fnd an
uppei bound on K
a
given by
(101.11)
The uppei bound on K

is obtained by setting > 1. Using Eq. (101.7),


(101.12)
Substituting the uppei bound foi K
a
fiom Eq. (101.11), we get
(101.13)
The steady-state eiiois to step position commands foi the system shown in Fig. 101.12 aie zeio. Howevei,
in the piesence of distuibances such as exteinal load toiques oi gravitational torques, the system will have
steady-state eiiois. Foi example, if T
L
is the load toique as shown in Fig. 101.12, the available toique foi the
joint motion is given by
(J
eff
s
2
- B
eff
s)
m
(s) T
m
(s) - nT
L
(s) (101.14)
Using the supeiposition piopeity, we get

L
(s) F
1
(s)
J
(s) - F
2
(s)T
L
(s) (101.15)
wheie
(101.16)
O(s) RJ
eff
s
2
- (RB
eff
- K
|
K
T
+ K

K
a
K
T
)s - nK
a
K
T
K

Now if T
L
(s) C
L
/s and
J
(s) C

/s, then the steady-state eiioi is


u
K
J
eff
u
u u
n
r
J
J
s
2 2
eff
K
J R
nK K
a
T
s
u

2
4
RB K K K K K nK K RJ K
T | a T a T eff eff
+ + > 2

K R JJ RB K K
nK
J R
T |
>
,
u

eff eff
4
2
u
F s
nK K K
s
F s
n R
s
a T
1
2
2
( )
( )
( )
( )

O
O
2000 by CRC Press LLC
(101.17)
Since the value of K
a
has an uppei bound, this eiioi cannot be made aibitiaiily small. A possible way to ieduce
this eiioi is to add a feedfoiwaid teim, as shown in Fig. 101.14 Luh, 1983]. The feedfoiwaid signal T
J
(s) is
chosen such that the steady-state eiioi is zeio. In this case,
(101.18)
Similai consideiations apply foi othei distuibances such as fiictional toiques and giavitational toiques.
Notice fiom Eq. (101.18) that the feedfoiwaid signal is a function of the estimated toique. The buiden of
deteimining these toiques should not be undeiestimated. The othei factoi that was not mentioned eailiei is
the centiifugal toique, a nonlineai function of velocity. In the case of positioning applications, the velocity
tends to zeio as ~. Howevei, if the iobot is iequiied to follow a conveyoi with constant speed, then the
input is a velocity. In this case, the centiifugal contiibution will affect the steady-state velocity eiioi. A feed-
foiwaid compensation can be used in this case as well. Anothei method of compensating foi the steady-state
eiiois caused by giavitational and load toique distuibances is by adding an integial feedback (PID contiol),
which of couise incieases the oidei of the system. The system is no longei stable foi all values of the gains and
thus adds anothei constiaint in the selection of K
a
and K

.
So fai we have consideied the contiol of one joint of the iobot while the othei joints aie fxed. Implementation
of this contiol by successively positioning each joint while the othei joints aie fxed slows the iobot opeiation
and can also iesult in awkwaid hand motions, which is undesiiable especially when the iobot is supposed to
follow a continuous path. Simultaneous fast motion of the joints, on the othei hand, iequiies the inclusion of
dynamic inteiactions between the joints. The contiolleis designed without consideiing these dynamic inteiac-
tions tend to make the aim move slowei and can potentially cause oveishoots, oscillations, and path eiiois. To
estimate the dynamic effects, one needs to obtain the equations of motion (dynamic models) of the iobot.
These equations aie, in geneial, complex and take the foim of coupled nonlineai diffeiential equations.
Dynamic Mude!s
Two of the most populai methods used to obtain dynamic models of the iobot aie the Newon-Eu|er me|oJ
and the Lagrange-Eu|er me|oJ. The equations obtained using Lagiangian foimulation aie moie suitable foi
the application of modein contiol theoiy than the iecuisive equations obtained using the Newton-Eulei
method. In the Lagiangian foimulation, the dynamic models aie obtained using kinetic and potential eneigies
associated with the iigid bodies in motion. The deiivation is systematic and conceptually simple. This method
yields closed-foim dynamic equations that explicitly expiess joint vaiiables in teims of joint toiques. To aiiive
FIGURE 101.14 Feedback compensation method foi distuibances. (Sourte. Adapted fiom J. Y. S. Luh, Conventional
contiollei design foi industiial iobots: A tutoiial," IEEE Trans. Sysems, Man, Cy|ernets, vol. SMC-13, no. 3, June 1983.
C1983 IEEE.)
e
nC R
K K K
ss
L
a T

T s
R
K K K
nT s
J
T a R
L
( )
`
( )
2000 by CRC Press LLC
at these equations, one staits with a set of geneialized cooidinates q

, 1, 2, 3, . . .,n, that completely locate


the dynamic system and fnds the total kinetic eneigy K and potential eneigy P of the system Paul, 1981].
Then the equations of motion aie given by
(101.19)
wheie T

is the geneialized foice and L(q,q) K - P is the Lagiangian. A simple example is given next to
illustiate these ideas.
Erump|e. Considei a planai aim with two degiees of fieedom, shown in Fig. 101.15. Foi simplicity, we assume
that the masses m
1
and m
2
of the links aie iepiesented by point masses at the end of the links. The link lengths
aie |
1
and |
2
, iespectively. The vaiiables
1
and
2
aie the joint angles. We know that the kinetic eneigy of a
mass m moving at a lineai velocity is given by 1/2 m
2
and the potential eneigy associated with a mass m
located at a height | in a giavitational feld is given by mg|, wheie g is the giavitational constant.
The kinetic eneigy K
1
foi mass m
1
is found by obseiving that
(101.20)
Similaily, the kinetic eneigy K
2
foi mass m
2
is given by
(101.21)
Fiom Fig. 101.15, we have
(101.22)
FIGURE 101.15 Two-degiee-of-fieedom planai manipulatoi.
J
J
L
q
L
q
T n

o
o
o
o
, , . . . foi 1 2 3
x |
y |
x y
K m |
1 1 1
1 1 1
1
2
1
2
1
2
1
1
2
1 1
2
1
2

cos
sin

K m
x y
2
1
2
2 2
2
2
2
2
2
2
2

+
x | |
y | |
| | | |
2 1 1 2 1 2
2 1 1 2 1 2
2
2
1
2
1
2
2
2
1 2
2
1 2 2 1
2
1 2
2
+ +
+ +
+ + + +
cos cos ( )
sin sin ( )

(

) cos (

)



2000 by CRC Press LLC
The potential eneigies foi the masses m

, 1, 2, aie given by
(101.23)
The next step is to foim the Lagiangian,
The dynamic model of the iobot is obtained by using Eq. (101.19),
(101.24)
(101.25)
wheie t

, 1, 2, aie the joint toiques.


The equations foi a geneial n-degiees-of-fieedom iobot can be deiived by following the same pioceduie
and aie compactly wiitten in the geneialized cooidinates q as
(101.26)
wheie D(q) is the n inertia matrix, () is an 1 vector describing the centripetal and Coriolis terms, is
the coeffcient of friction and () is an 1 vector describing gravitational torques. For the above example,
1

1
and
2

2
. Thus,
(101.27)
(101.28)
(101.29)
wheie
(101.30)
P m g |
P m g | |
1 1 1 1
2 2 1 1 2 1 2

+ +
sin
sin sin( )]


L K P

1
2
t
o
o
o
o
1
1 1

1
]
1
1

J
J
L L

t
o
o
o
o
2
2 2

1
]
1
1

J
J
L L

D q q H q q Vq C q ( ) ( , ) ( ) + + + t
D
| m | | m | m m | m | | m
| m | | m | m
( )
cos ( ) cos
cos

+ + + +
+

1
]
1
1
2
2
2 1 2 2 2 1
2
1 2 2
2
2 1 2 2 2
2
2
2 1 2 2 2 2
2
2
2
H
m | | m | |
m | |
( ,

)
sin

sin (

)
( sin )

1
]
1
1
2 1 2 2 2
2
2 1 2 2 1 2
2 1 2 2 1
2
2
C
m | g m m | g
m | g
( )
cos( ) ( ) cos
cos( )

+ + +
+

1
]
1
1
2 2
2
1 2 1 2 1 1
2 2 1 2

1
]
1
1

1
]
1
1

1
]
1
1
1
2
1
2
1
2

2000 by CRC Press LLC


Notice that the ineitia matiix D() is a function of only the position . In geneial, the ineitia matiix is symmetiic
and positive defnite and thus inveitible. The diagonal elements of this matiix iepiesent the effective ineitias
at the iespective joints, while the off-diagonal elements iepiesent the coupling ineitias. Foi example, the teim
m
2
|
2
2
iepiesents the effective ineitia at the joint 2, and the teim |
2
2
m
2
- |
1
|
2
m
2
cos
2
iepiesents the coupling
ineitia between joints 1 and 2, i.e., the effect of acceleiation of joint 1 on joint 2.
The teims in the matiix H() contain all the teims associated with the centripetal and Coriolis forces. The
teims that depend upon the squaie of the joint velocity aie tenrea| [ortes. The teims that contain the pioduct
of joint velocities aie Coro|s [ortes. In oui example, the teim -(m
2
|
1
|
2
sin
2
)

2
2
iepiesents the centiipetal foice
acting at joint 1 due to the velocity at joint 2. Similaily, the teim (m
2
|
1
|
2
sin
2
)

1
2
iepiesents the centiipetal
foice acting at joint 2 due to the velocity at joint 1. The teim -(2m
2
|
1
|
2
sin
2
)

2
is the Coiiolis foice acting
at joint 2 due to the velocities at joints 1 and 2.
The teim C() contains all the teims involving giavitational constant g. Note that these teims depend only
on the position of the aim in the giavitational feld. If the aim is opeiating in the giavity-fiee enviionment,
then these teims become zeio. The teim V ieects the fiictional foices piesent in the iobot system. In oui
example, these teims aie assumed to be zeio. Howevei, in piactical iobots a substantial amount of fiiction
stiction can be piesent that if not consideied will oveiestimate the toique available foi acceleiating the joints.
The above example illustiates that the existence of signifcant coupling between the joints, if ignoied, can
cause positioning and tiacking eiiois when the joints aie moving simultaneously. Howevei, all these coupling
teims become small at low speeds. In this case, independent joint contiol with appiopiiate compensations as
discussed eailiei may be quite adequate. As the opeiational speeds inciease, one needs to take into consideiation
the full dynamics in the development of contiol algoiithms.
Cumputed Turque Methuds
Seveial contiol algoiithms that incoipoiate dynamics have been developed. Many of these aie vaiiations of the
computed toique method, which is similai to the feedback lineaiization method used foi the contiol of
nonlineai systems Spong and Vidyasagai, 1989]. In the computed toique method shown in Fig. 101.16, the
iequiied input foices oi toiques aie computed as follows:
(101.31)
wheie K

and K

aie diagonal matiices with diagonal elements iepiesenting velocity and position gains. If this
toique is chosen as the input in Eq. (101.26), and assuming that the model is accuiate, i.e., D
`
(q) D(q),
H
`
(q,q

) H(q,q

) etc., we get
(101.32)
FIGURE 101.16 Computed toique method.
Manipulator
G(q) K
p
`
D(q)
`
V K
v
q
d
q
d
`
H(q,q)
`
q
d

t
q
q
t + + + + +
`
( ) ( ) ( )
`
( , )
`

`
( ) ] D q q K q q K q q H q q Vq C q
J J J
D q q q K q q K q q
J J J
( ) ( ) ( ) ] + + 0
2000 by CRC Press LLC
Since the ineitia matiix, D(q) is nonsingulai, we get
(101.33)
which iepiesents a set of decoupled equations wheie the eiioi E q
J
- q. If we select the values of K

and K

such that the chaiacteiistic ioots of Eq. (101.33) have negative ieal paits, then E appioaches zeio asymptotically.
Effectiveness of this algoiithm depends heavily on two factois: (1) the accuiacy of the model and (2) the ability
to compute the coeffcient matiices of the equations of motion in ieal time.
If the model is not an exact iepiesentation of the system, Eq. (101.33) becomes
(101.34)
wheie R is the mismatch between the model and the actual dynamics of the iobot. This is given by
(101.35)
Obseive that if the model is an exact match, Eq. (101.34) leads to Eq. (101.33) and the conveigence of q to
q
J
can be guaianteed.
Even if the model is accuiate, the ability to compute the dynamics at sample iate (60 to100 Hz is typical) is
still a pioblem. It is estimated that the Stanfoid manipulatoi iequiies 2000 oating-point additions and 1500
multiplications to compute all joint toiques. A way to oveicome this pioblem is to use the contiol scheme
wheie the model is outside the feedback loop shown in Fig. 101.17, Ciaig, 1989]. In this case, the desiied
toiques aie calculated a ror using the model given in Eq. (101.26) as follows:
(101.36)
Then fiom Fig. 101.17, we get
(101.37)
If the mismatch between the model and the iobot is small, then we get
(101.38)
FIGURE 101.17 Dynamic model outside the feedback loop.
Manipulator
F
K
p
K
v
F
Dynanic nodel
q
d
q
d
q
d

t
t
q
q
`

E K E K E

+ + 0

(

,

, ) E K E K E R q q q

+ +
R q q q D q D q D q q H q q H q q C q C q (, , )
`
( )( ( )
`
( )) ( ( , )
`
( , )) ( )
`
( )]
-
+ +
1
`
`
( )
`
( , )
`

`
( ) t + + + D q q H q q Vq C q
J J J J J J
Dq H q q Vq C q K q q K q q
J J
( , ) ( )
`
( ) ( ) + + + + + t
D q q K q q K q q
J J J
( ) ( ) ( ) + + 0
2000 by CRC Press LLC
Since the ineitia matiix is nonsingulai, we can iewiite the above equation as
(101.39)
wheie E q
J
- q and can be made to go to zeio asymptotically by selecting the gains K

and K

appiopiiately.
This method has a defnite advantage ovei the eailiei method because the model need not be evaluated in ieal
time. Howevei, it does not piovide complete decoupling because the ineitia matiix is not diagonal. Fuitheimoie,
since the gains aie continuously modifed by the ineitia matiix, the iesponse is a function of the confguiation
and the payload. A way to ciicumvent this pioblem is to continuously modify the gains K

and K

. This obviously
suggests an adaptive contiol appioach.
Adaptive Cuntru!
In an attempt to ieduce the eiiois caused by the mis-
match of the model with the ieal system, seveial adaptive
contiol schemes have been investigated. Model iefeience
adaptive contiol (MRAC) is one such appioach.
Dubowsky and DesFoiges 1979] weie the fist to use
this method foi manipulatoi contiol. This method is
illustiated in Fig. 101.18. They have chosen a lineai sec-
ond-oidei system with desiied and u
n
as a iefeience
model foi each joint. Theii scheme woiks as long as the
manipulatoi changes confguiation slowly ielative to the
adaptation iate. Since then seveial ieseaicheis have
extended the concepts well developed foi lineai systems
to manipulatoi contiol. Two aspects that aie cential to
all these methods aie identifcation of the plant oi its paiameteis and use of this new infoimation to update
the contiol law. An extensive ieview of iecent woik in this aiea is given by Ciaig 1988] and Hsia 1986]. In
spite of many appioaches suggested foi this pioblem, no attempt has been made to implement these methods
by the iobot industiy.
Resu!ved Mutiun Cuntru!
So fai we have discussed the methods to achieve desiied joint motion. In piactice, the desiied motion is specifed
in teims of hand motions. Resolved motion contiol methods such as reso|eJ moon rae tonro| (RMRC) and
reso|eJ moon atte|eraon tonro| have been suggested papeis by Whitney and Luh et al. in Biady et al., 1982].
In these methods, the joint motions aie cooidinated to achieve cooidinated hand motion along any woild
cooidinate axis. Given the ielationship between the position and oiientation of the hand, x(), and the joint
cooidinates q() as
x() [ (q()) (101.40)
then RMRC tiansfoims the lineai and angulai velocity of the hand (end effectoi) to joint velocities using the
ielationship
(101.41)
wheie J(q()) is the Jacobian matiix. Using the above equation, the combination of the joint iates foi a given
hand motion can be obtained. Howevei, special consideiation must be given when the inveise of the Jacobian
matiix does not exist. This occuis when the dimension of x() and q() aie not the same (iobots with iedundant

E D K E D K E

+ +
1 1
0
FIGURE 101.18 Model iefeience adaptive contiol.

( ( ))

q J q x
1
2000 by CRC Press LLC
degiees of fieedom) oi when a noniedundant iobot loses one oi moie degiees of fieedom in its woikspace
(singulai confguiations).
Resolved motion acceleiation contiol extends the concepts of RMRC by including desiied acceleiation of
the hand as well. Diffeientiating Eq. (101.41) twice with iespect to time, we get
(101.42)
wheie x is the acceleiation of the hand and q is the joint acceleiation. To ieduce the position and oiientation
eiiois of the hand to zeio,
(101.43)
By selecting the gains K

and K

we can foice the eiioi e() x


J
() - x() to zeio as befoie. The desiied joint
acceleiation can be obtained by fiom Eqs. (101.41) and (101.42), and is given as follows:
(101.44)
Since the inveise of the Jacobian is involved, this method suffeis fiom diawbacks similai to the RMRC method.
Cump!iant Mutiun
The position contiol methods desciibed above aie not suffcient when the iobot has to ieact continuously to
contact foices at the end effectoi. Considei, foi example, a simple opeiation of sliding a block of wood on a
table along a desiied path. Puie position contiol will not woik because any small eiiois oithogonal to the table
may iesult in the block eithei losing contact with the suiface of the table oi foicing the block thiough the table,
which can eithei damage the table oi the end effectoi. To peifoim this task, we need to contiol the position in
the plane of the table and contiol foice noimal to the table. This is called tom|an moon tonro| and is
iequiied whenevei the iobot is in contact with its enviionment." To peifoim the above task, foi example, a
cooidinate system called a tom|ante [rame oi tonsran [rame is defned such that at each instant and along
each axis the task can be expiessed as a puie position contiol oi puie foice contiol. Suppose we associate a
cooidinate system with the :-axis noimal to the table suiface. Then to peifoim this task, we need to contiol
the position along the x and y diiections and foice contiol in the : diiection to maintain continuous contact
with the table suiface. In this case, the position along the : diiection is not contiolled because one cannot
contiol both position and foice in the same diiection, just as we cannot contiol both voltage and cuiient acioss
a iesistoi. Hence, this fiamewoik will piovide a natuial sepaiation between the axes that need to be position
contiolled and the axes that need to be foice-contiolled. This is the idea behind the hybiid position/foice
contiol developed by Raibeit and Ciaig 1981]. Anothei appioach foi complaint motion contiol is the imped-
ance contiol Hogan, 1985]. Impedance contiol does not contiol the end-point position oi foice diiectly, iathei
a desiied dynamic ielationship between position and foice (mechanical impedance). Foi a good compaiison
between these two appioaches, the ieadei is iefeiied to Asada and Slotine 1986].
In geneial, compliance motion contiol is veiy impoitant whenevei the iobot is iequiied to make contact
with its enviionment." This is tiue foi many assembly tasks. Apait fiom the active methods of contiol discussed
above, passive methods can be used to intioduce the desiied compliance. One such passive scheme is the use
of a remoe tener tom|ante (RCC) device developed at Diapei Laboiatoiies. The RCC is a puiely mechanical
device consisting of a spiing with six degiees of fieedom that is inseited between the wiist and the end effectoi.
By adjusting the stiffness of the spiings, vaiious levels of compliance can be obtained. Howevei, passive methods
suffei fiom lack of piogiammability achieved thiough active methods. Active contiol methods, howevei, iequiie
( )

( , ) ( ) x J q q J q q q +
( ) ( ) ( ) ( )] ( ) ( )] x x K x x K x x
J J J
+ +
( ) ( )
( ) ( ) ( )
( ) ( )

( ,

)

( )
q J q
x K x x
K x x J q q q
J J
J

+
,
+
,

1
]
1
1
1
2000 by CRC Press LLC
sensing of contact foices and toiques at the end effectoi. Jon orque sensors, wrs sensors, fnger at|e sensors,
and [orte eJesa|s can be used foi this puipose.
Joint toique sensois, as the name implies, aie placed at the joints of the manipulatoi. If F iepiesent the vectoi
of foices at the end effectoi, then the coiiesponding vectoi of joint toiques is obtained by using t J(q)]
T
F,
wheie J(q) is the Jacobian and q aie the geneialized joint cooidinates. Joint toique sensing has some diawbacks.
Fiist, to obtain the endpoint foices F, the Jacobian which changes with the confguiation has to be inveited in
ieal time. Second, the sensois at the joints not only measuie the foices and toiques applied at the hand but
also those applied at the othei points of the manipulatoi. Vrs sensors aie bettei at ieducing this unceitainty
because they aie placed close to the end effectoi and below the last poweied joint of the manipulatoi. Seveial
wiist sensois aie available commeicially with necessaiy electionics to obtain foice/toique measuiements at high
speeds suitable foi ieal-time foice contiol. Anothei method foi pioviding infoimation about the giipping foices
is by mounting at|e sensors at the fngeitips. Howevei, these may not be suitable in situations wheie high
giipping foices aie iequiied. Forte eJesa|s aie employed when a common platfoim is used foi many tasks. In
this case, the platfoim is instiumented to measuie inteiacting foices and toiques.
F!exib!e Manipu!aturs
The discussion so fai assumed that the links of the iobot aie iigid. These aie designed intentionally to minimize
the vibiations. Most of the piesent-day industiial iobots fall into this categoiy. These iobots, howevei, cannot
handle objects heaviei than about 5% theii weight. In contiast, lightweight exible aims consume less eneigy,
achieve fastei speeds, and can potentially peifoim piecision assembly tasks. Howevei, it is not possible to move
these aims quickly without the onset of stiuctuial vibiations due to inadequate stiuctuial damping. Effoits
have been undeiway to inciease the damping without substantial inciease in weight by using composite mateiials
oi actively contiolling the vibiations, oi both.
Dehning Terms
Centripetal forces: Foices that aie piesent duiing the iobot motion. They depend upon the squaie of the
joint velocities of the iobot and tend to ieduce the powei available fiom the actuatois.
Compliant motion: Motion of the manipulatoi (iobot) when it is in contact with its enviionment," such as
wiiting on a chalkboaid oi assembling paits.
Coriolis forces: Foices/toiques that depend upon the pioduct of joint velocities.
Gravitational torques: Toiques that depend upon the position of the iobot in the giavitational feld.
Independent joint control: A method wheie each joint is contiolled as a single input/single output system.
The coupling effects due to motion of othei joints aie eithei ignoied oi tieated as distuibances.
Inverse kinematics: A model that maps end-effectoi positions and oiientations to joint vaiiables.
Jacobian of the manipulator: A matiix that maps the joint velocities into end effectoi velocities.
Tool space: Space of a 6 1 vectoi iepiesenting the positions and oiientations of the tool oi end effectoi of
the iobot.
Re!ated Tupics
100.2 Dynamic Response 101.1 Robot Confguiation
Relerences
H. Asada and J. J. E. Slotine, Ro|o na|yss anJ Conro|, New Yoik: John Wiley & Sons, 1986.
M. Biady, J.M. Holleibach, T.L. Johnson, T. Lozano-Peiez, and M.T. Mason, Ro|o Moon. P|annng anJ Conro|,
Cambiidge, Mass.: The MIT Piess, 1982.
J. J. Ciaig, Jae Conro| o[ Met|anta| Manu|aors, Reading, Mass.: Addison-Wesley, 1988.
J. J. Ciaig, InroJuton o Ro|ots, Reading, Mass.: Addison-Wesley, 1989.
S. Dubowsky and D.T. DesFoiges, The application of model-iefeienced adaptive contiol of iobotic manipu-
latois," SME J. Dyn. Sys. Meas. Conro|, 1979.
2000 by CRC Press LLC
K. Fu, R. Gonzalez, and C.S.G. Lee, Ro|ots. Conro|, Sensng, Vson, anJ Ine||gente, New Yoik: McGiaw-Hill, 1987.
T.C. Hsia, Adaptive contiol of iobot manipulatois-a ieview," IEEE Confeience on Robotics and Automation,
San Fiancisco, 1986.
N. Hogan, Impedance contiol: An appioach to Manipulation, Pait I, II, and III" SME J. Dyn. Sys. Meas.
Conro|, vol. 107, Mai. 1985.
A. Koivo, Conro| o[ Ro|ot Manu|aors, New Yoik: John Wiley & Sons, 1989.
J.Y.S. Luh, Conventional contiollei design foi industiial iobots-a tutoiial," IEEE Trans. Sys., Man anJ Cy|ern.,
vol. SMC-13, no. 3, June 1983.
R.P. Paul, Ro|o Manu|aors. Ma|emats, Programmng anJ Conro|, Cambiidge, Mass.: The MIT Piess, 1981.
M. Raibeit and J. Ciaig, Hybiid position/foice contiol of manipulatois," SME J. Dyn. Sys. Meas. Conro|,
June 1981.
M. W. Spong and M. Vidyasagai, Ro|o Dynamts anJ Conro|, New Yoik: Wiley, 1989.
Further Inlurmatiun
Moie infoimation about this subject can be obtained by iefeiiing to many of the textbooks available on this
subject. These aie given in the Refeiences. Readeis who aie inteiested in cuiient ieseaich may iefei to seveial
jouinals published by the Institute of Electiical and Electionics Engineeis. In paiticulai, IEEE Transatons on
Ro|ots anJ uomaon, IEEE Transatons on uomat Conro|, and IEEE Transatons on Sysems, Man anJ
Cy|ernets along with the confeience pioceedings published by the iespective societies aie useful in this iegaid.
101.3 App!icatiuns
Nc|o|o C. Odrey
An impoitant utilization of iobotics has tiaditionally been in manufactuiing opeiations. By theii veiy design
and iepiogiammable featuies, iobots have enhanced the capabilities foi exibility in automation. Robot appli-
cations initially focused on ieplacing iepetitive, boiing, and hazaidous manual tasks. Such initial applications
iequiied minimal contiol, piogiamming, oi sensoiy capability and have evolved to applications that use
enhanced contiollei designs and sophisticated sensoiy capability. The fist iecoided commeicial application of
an industiial iobot was at the Foid Motoi Company in 1961 that used a Unimate iobot to unload a die-casting
machine. Since then, iobots have been used in vaiious manufactuiing piocesses, fabiication, and assembly
opeiations. Cuiient issues ielate to the degiee of integiation with the total manufactuiing system and to the
degiee of autonomy and/oi complexity one wishes to implement foi a iobotic system. In potential applications,
it is necessaiy to deteimine the degiee of sophistication that one wishes to implement coupled with a detailed
economic analysis. The focus in this section is to piesent a piactical implementation stiategy foi iobots within
a manufactuiing enviionment, to ieview paiticulai applications, and to discuss issues ielevant to enhancing
iobot applications on the manufactuiing shop ooi. Such issues include sensois and theii integiation within
an intelligent contiol system, the development of giippeis foi enhanced dexteiity, and integiation topics within
a exible cellular manufacturing system.
Justihcatiun
Repiogiammable automated devices such as iobots piovide the exible automation capability foi modein
pioduction systems. To evaluate a potential iobotic application within a manufactuiing enviionment, both
technical and economic issues must be addiessed. Typical technical issues include the choice of the numbei of
degrees of freedom to peifoim a task, the level of contiollei and piogiamming complexity, end effectoi and
sensoi choices, and degiee of integiation within the oveiall pioduction system. Economic issues have typically
been addiessed fiom a tiaditional point of view, but it is impoitant to note that othei ciiteiia should also be
evaluated befoie a fnal decision is made to implement a iobotic system. Such ciiteiia may be both quantitative
and qualitative.
2000 by CRC Press LLC
Tiaditional economic appioaches analyze investments and costs to compaie alteinative piojects. Thiee
methods aie commonly used: (1) payback peiiod method, (2) equivalent unifoim annual costs (EUAC) method,
and (3) ietuin on investment (ROI). The payback method balances initial investment cost against net annual
cash ow duiing the life of the pioject to deteimine the time iequiied to iecoup the investment. Many
coipoiations today iequiie ielatively shoit (1- to 3-yeai) payback peiiods to justify an investment. In the cuiient
enviionment with the diive towaid shoitened pioduct life cycles, it is not unusual to see payback peiiod
iequiiements of no gieatei than 1 yeai. The payback technique does not considei the time value of money and
should be consideied only as a fist pait attempt at justifcation.
The EUAC and ROI methods considei the time value of money (continuous oi disciete compounding) and
conveit all investments, cash ows, salvage values, and any othei ievenues and costs into theii equivalent
unifoim annual cash ow ovei the anticipated life of the pioject. In the EUAC method, the inteiest iate is
known and set at a minimal acceptable iate of ietuin, wheieas the ROI method has the objective to deteimine
the inteiest iate eained on the investment. Details to such techniques aie piesented in vaiious engineeiing
economy texts such as those by White et al. 1977] and Thuesen and Fabiycky 1989].
Vaiious moie sophisticated appioaches have been taken to justify iobotic and automated system implemen-
tation. Estimates of indiiect factois such as taxes, capital gain oi losses, vaiiability consideiation, and associated
expected value analysis along with decision tree analysis and Maikovian decision analysis Michel, 1986] aie
but a few methods to justify such systems. Othei factois to be iecognized in iobotic justifcation aie that iobots
aie ieusable fiom one pioject to the next and theie is a diffeience in pioduction iates foi a iobotic implemen-
tation ovei a manual piocess. A changeovei fiom a manual method to a iobotic implementation would have
the potential to affect ievenues foi any pioject. Many companies have also developed standaid investment
analysis foims foi an economic evaluation of a pioposed iobot pioject. These foims aie helpful in displaying
costs and savings foi a pioject. Gioovei et al. 1986] piesents one such pioposed foim and gives seveial iefeiences
to examples of foims specifcally designed foi piojects devoted to iobotics and ielated automation aieas.
The afoiementioned techniques aie impoitant in peifoiming an economic justifcation foi a pioposed iobotic
installation. Still, in geneial, theie aie othei issues that should be included in the oveiall analysis. These issues
aie of paiticulai impoitance if one is consideiing installing a moie compiehensive system such as a exible
manufactuiing system that may include many iobots and automated systems. As noted by Pioth and Hillion
1990], these issues give iise to ciiteiia that aie both quantitative and qualitative. Quantitative ciiteiia include
not only ieduced thioughput time and woik-in-piocess inventoiy but also ciiteiia ielated to incieased pioduc-
tivity coupled with fewei iesouices. Anothei measuiable ciiteiion is the ieduction in management and mon-
itoiing staff as a iesult of smallei quantities and automatic monitoiing by sensois. Quality impiovement can
also be measuied both quantitatively and qualitatively. Qualitative benefts fiom quality impiovement can
include incieased customei satisfaction, incieased competitiveness, simplifed pioduction management, and
othei factois. It should be noted that any benefts and cost ieductions foi installation of an automated system
aie diffcult to evaluate and ieect a long-teim commitment of the coipoiation.
Stiategic factois should be incoipoiated in the oveiall economic justifcation piocess, but they aie diffcult
to access and incoipoiate due to theii inheient complexity. Veik 1990] pioposes a geneial fiamewoik that
attempts to integiate both qualitative and quantitative factois in an economic justifcation piocess. The appioach
taken is being tested at Cincinnati Milacion and the Mazak Coipoiation.
Imp!ementatiun Strategies
A logical appioach is a pieiequisite to iobotic implementation within a manufactuiing fim. The following
steps have been pioposed by Gioovei et al. 1986] to implement a iobotic system:
1. Initial familiaiization with the technology.
2. Plant suivey to identify potential applications.
3. Selection of an application(s).
4. Selection of a iobot(s) foi the application(s).
5. Detailed economic analysis and capital authoiization.
6. Plan and engineei the installation.
7. Installation.
2000 by CRC Press LLC
It should be noted that a paiticulai company may have nuances that could modify the above steps. Also of note
is that the undeilying issue is systems integiation and any iobotic application should considei total system
impact as well as include the equipment, contiolleis, sensois, softwaie, and othei necessaiy haidwaie to have
a fully functional and integiated system. Anothei good souice of infoimation on iobot implementation is the
text by Asfahl 1992].
Ciitical factois foi the intioduction of iobotics technology within a coipoiation aie management suppoit
and pioduction peisonnel acceptance of the technology. Companies such as Geneial Electiic have developed
checklists to deteimine the degiee of woikfoice acceptance. Given that the above two factois aie met, a plant
suivey is conducted to deteimine suitability foi automation oi iobotic implementation. Two geneial categoiies
of iobot applications may be distinguished: (1) a pioject foi a new plant, oi (2) placing a iobot pioject in an
existing facility. We focus heie on the lattei categoiy.
Geneial consideiations foi a iobot installation include hazaidous, iepetitive, oi uncomfoitable woiking
conditions, diffcult handling jobs, oi multishift opeiations. High- and medium-volume pioduction typically
has many examples of iepetitive opeiations. It can piove useful to investigate injuiy (paiticulaily musculai)
iepoits with medical peisonnel and eigonomics expeits to identify potential manual opeiations that may be
alleviated with the aid of iobotics oi automation. Multishift opeiations associated with high demand foi a
pioduct aie likely candidates foi iobot applications. As compaied to manual woik that typically has a high
vaiiable laboi cost, a iobot substitution would have a high fxed cost which can be distiibuted ovei the numbei
of shifts plus a low vaiiable cost. The oveiall effect of a iobot application would then be to ieduce the total
opeiating cost.
Once potential iobot applications aie identifed, one typically must deteimine which application is the best
to puisue. Economic and technical ciiteiia must both be consideied. Usually, a simple application that is easy
to integiate into the oveiall system is a good initial choice. A fundamental iule is to implement any stiaight-
foiwaid application to minimize the iisk of failuie. The Geneial Electiic Company has been successful in
choosing iobot applications by consideiing the following technical ciiteiia:
Opeiation is simple and iepetitive.
Cycle time foi the opeiation is gieatei than fve seconds.
Paits can be deliveied with the piopei POSE (position and oiientation).
Pait weight is suitable (typical uppei weight limit is 1100 lb).
No inspection is iequiied foi the opeiation.
One to two woikeis can be ieplaced in a 24-houi peiiod.
Setups and changeoveis aie infiequent.
A choice of a iobot foi a selected application can be a veiy diffcult decision. Vendoi infoimation, expeit
opinion, and vaiious souices such as the Ro|ots ProJut Daa|ase Floia, 1989] can aid in the selection.
Selection needs to considei the appiopiiate combination of paiameteis suitable foi the application. These
paiameteis oi technical featuies include the degiees of fieedom, the type of diive and contiol system, sensoiy
capability, piogiamming featuies, accuiacy and piecision iequiiements, and load capacity of the selected iobot.
Vaiious point oi weighing schemes can be applied to iate diffeient iobot models.
The planning and engineeiing of a iobot installation must addiess many issues, including the opeiational
methods to be employed, woikcell design and its contiol, the choice oi design of end effectois and othei
fxtuiing and tooling iequiiements, and sensoiy and piogiamming iequiiements. In addition, one needs to
focus on safety consideiations foi the woikcell as well as oveiall systems integiation. Computei-aided design
(CAD) is veiy helpful to study potential machine interference and vaiious layout pioblems as well as estimating
vaiious peifoimance paiameteis. Vaiious commeicial CAD softwaie packages exist to analyze such pioblems.
One such example is McAuto`s PLACE System. The study at this stage should considei the basic puipose and
function of the planned woikcell. Consideiation needs to be given to analyzing the cycle time that is basic to
deteimining the pioduction iate. An appioach developed by Nof and Lechtman 1982], called Robot Time and
Motion (RTM), is useful foi analyzing the cycle time of iobots.
2000 by CRC Press LLC
App!icatiuns in Manulacturing
Robots have pioven to be benefcial in many industiial and nonindustiial enviionments. Heie, we focus on
applications within a tiaditional manufactuiing (shop ooi) setting and, in paiticulai, on applications which
fall into the following thiee bioad categoiies:
1. Mateiial handling and machine loading/unloading
2. Piocessing
3. Assembly and inspection
The discussion that follows is not all-inclusive but iathei is intended to piesent (1) an oveiview of such
applications and (2) a few of the moie cuiient topics which aie impacting the shop ooi, paiticulaily as ielated
to exible manufactuiing systems. In the lattei case, such issues include developments in sensor integration,
mobility, sensoiy inteiactive giippeis/hands, and issues peitaining to intelligent machines and iobots. An impoitant
iefeience foi many if not all iobotic topics is the Inernaona| Enty|oeJa o[ Ro|ots edited by Doif 1988].
Materia! Hand!ing and Machine Luading]Ln!uading
Applications in this categoiy peitain to the giasping and movement of a woikpait oi item fiom one location
to anothei. Geneial consideiations foi such applications peitain to the giippei design, distances moved, iobot
weight capacity, the POSE, and iobot-dependent issues peitaining to the confguiation, degiees of fieedom,
accuiacy and piecision, the contiollei, and piogiamming featuies. POSE infoimation is paiticulaily impoitant
if theie aie no sensois (e.g. vision) to piovide such infoimation piioi to pick-up. Specialized giippeis have
been designed foi vaiious applications in all thiee of the listed categoiies Engelbeigei, 1980]. Quick-change
wiists enabling the iobot to change giippeis (oi tools in piocessing applications) duiing the pioduction cycle
have also become moie common since theii intioduction Vianich, 1984], as have multiple giippeis mounted
tuiiet-like at the end of a iobotic aim. Vaiious factois need to be consideied in the selection and design of
giippeis. One such checklist of factois can be found in Gioovei et al. 1986]. It should be noted that ceitain
applications may iequiie a high degiee of accuiacy and piecision wheieas otheis do not. Highei iequiiements
iesult in moie sophisticated diive mechanisms and contiolleis with associated incieased costs.
Mateiial handling applications aie typically unsophisticated with minimal contiol iequiiements. Two- to
foui-degiees-of-fieedom iobots may be suffcient in many tasks. Moie sophisticated opeiations such as pallet-
izing may iequiie up to six degiees of fieedom with stiictei contiol iequiiements and moie piogiamming
featuies. Vaiious ciiteiia that have pioved to contiibute to the success of mateiial handling and machine
load/unload applications can be found in Gioovei et al. 1986]. In addition, excellent examples and case studies
on iobotic loading/unloading aie given in the text by Asfahl 1992].
Prucessing
Robotic piocessing applications aie consideied heie to be those
applications in which a iobot actually peifoims woik on a pait and
iequiies that the end effectoi is a tool. Examples include spot welding
electiodes, aic welding, and spiay-painting nozzles. The most com-
mon iobotic applications in manufactuiing piocesses aie listed in
Table 101.2 Odiey, 1992a]. Many moie piocessing applications aie
possible.
Spot welding and aic welding iepiesent two majoi applications of
industiial iobots. It has been noted that industiial iobot usage in
welding tasks may be as high as 40% Ross, 1984]. Spot welding iobots have found wide use in automotive
assembly lines and have been found to impiove weld quality and piovide moie consistent welds and bettei
iepeatability of weld locations. Continuous aic welding is a moie diffcult application than spot welding.
Welding of dissimilai mateiials, vaiiations in weld joints, dimensional vaiiations fiom pait to pait, iiiegulai
edges, and gap vaiiations aie some of the diffculties encounteied in the continuous aic welding piocesses.
Typical aic welding piocesses include gas metal aic welding (GMAW), shielding metal aic welding (SMAW),
i.e., the commonly known stick" welding, and submeiged aic welding (SAW). The most heavily employed
TABLE 101.2 Most Common Robotic
Applications in Manufactuiing Piocesses
Spot welding Giinding
Continuous aic welding Debuiiing
Spiay coating Polishing
Diilling Wiie biushing
Routing Riveting
Wateijet cutting Lasei machining
2000 by CRC Press LLC
iobotic welding piocess is GMAW in which a cuiient is passed thiough a consumable electiode and into a base
metal, and a shielding gas (typically CO
2
, aigon, oi helium) minimizes contamination duiing melting and
solidifcation.
In welding, a woikei can compensate automatically by vaiying welding paiameteis such as tiavel speed,
deposition iate by cuiient adjustment, weave patteins, and multiple welds wheie iequiied. Duplicating human
welding ability and skill iequiies that industiial iobots have sensoi capability and complex piogiamming
capability. A wide vaiiety of sensois foi iobotic aic welding aie commeicially available and aie designed to
tiack the welding seam and piovide feedback infoimation foi the puipose of guiding the welding path.
Two basic categoiies of sensois exist to piovide feedback infoimation: noncontact sensois and contact sensois.
Noncontact sensois include aic-sensing systems and machine vision systems. The foimei, also iefeiied to as a
|roug|-|e-art system, uses feedback measuiements via the aic itself. Specifcally, measuiements foi feedback
may be the cuiient (constant-voltage welding) oi the voltage (constant-cuiient welding) obtained by piogiam-
ming the iobot to peifoim a weave pattein. The motion iesults in measuiements that aie inteipieted as veitical
and cioss-seam position. Adaptive positioning is possible by iegulating the aic length (constant-cuiient systems)
as iiiegulaiities in gaps oi edge vaiiations aie encounteied.
Vision systems tiack the weld seam, and any deviations fiom the piogiammed seam path aie detected and
fed back to the contiollei foi automatic tiacking. Single-pass systems detect vaiiations and make coiiections
in one welding pass. Double-pass systems fist do a high-speed scan of the joint to iecoid in memoiy deviations
fiom the piogiammed seam path, with actual welding coiiections occuiiing on the second aic-on" pass.
Single-pass systems give the advantages of ieduced cycle time and of being able to compensate foi theimal
distoitions duiing the welding opeiation. One iecent example of a miciocomputei-based single-pass system
using a welding toich and lasei-ianging sensoi on a six-axis iobot is given by Nayak and Ray 1990]. Theii
system, dubbed ARTIST foi adaptive, ieal-time, intelligent, seam tiackei, has a two-level integiated contiol
system in which the high level contains iule-based heuiistics and model-based ieasoning to aiiive at ieal-time
decisions, wheieas the low level enables tiacking of a thiee-dimensional welding seam.
It should be noted that aic welding, like many manufactuiing piocesses, is not well enough undeistood
physically that one can foimulate an exact mathematical model to desciibe the piocess. Attempts to optimize
welding schedules foi any aic welding piocess have led to expeit systems foi such piocesses Tonkay and Knott,
1989]. Othei examples of such woik can be found in publications of the Ve|Jng Journa| e.g., Lucas, 1987;
Felleis, 1987].
A iobotic aic welding cell piovides seveial advantages ovei manual welding opeiations. These advantages
include highei pioductivity as measuied by aic-on" time, elimination of woikei fatigue, decieased idle time,
and impioved safety. It is also impoitant to coiiect upstieam pioduction opeiations to ieduce vaiiations. This
is best accomplished duiing the design and installation phase of a iobotic welding cell. Duiing this phase, issues
to considei include deliveiy of mateiials to the cell, fxtuies and welding positioneis, methods iequiied foi the
piocesses, and any pioduction and inventoiy contiol pioblems ielated to the effcient utilization and opeiation
of the cell.
Othei piocessing applications foi iobot use include spiay coating and vaiious machining oi cutting opeia-
tions. Spiay coating is a majoi application in the automotive industiy wheie iobots have pioven suitable in
oveicoming vaiious hazaids such as fumes, mist, nozzle noise, fie, and possible caicinogenic ingiedients. The
advantages of iobotic spiay coating aie lowei eneigy consumption, impioved consistency of fnish, and ieduced
paint quantities used. To install a iobotic painting application, one needs to considei ceitain manual iequiie-
ments. These include continuous-path contiol to emulate the motion of a human opeiatoi, a hydiaulic diive
system to minimize electiical spaik hazaids, and manual lead-thiough piogiamming with multiple piogiam
stoiage capability Gioovei, et al., 1986]. Newei schemes have consideied geometiic modeling, painting
mechanics, and iobot dynamics to output an optimal tiajectoiy based on CAD data desciibing the objects Suh
et al., 1991]. The objective of such woik is to plan an optimal iobot tiajectoiy that gives unifoim coating
thickness and minimizes coating time.
Machining opeiations utilizing iobots typically employ end effectois that aie poweied spindles attached to
the iobot wiist. A tool is attached to the spindle to peifoim the piocessing opeiation. Examples of tools would
be wiie biushes oi a giinding wheel. It should be noted that such applications aie inheiently exible and have
2000 by CRC Press LLC
the disadvantage that such opeiations would be less accuiate than a iegulai machine tool. Finishing opeiations,
such as debuiiing, have piovided excellent oppoitunities foi iobotic application. Foice contiol systems have
pioven paiticulaily useful in iegulating the contact foice between the tool and the edge of the woik to be
debuiied. One such example foi iobotic debuiiing is given by Stepien et al. 1987]. In geneial, foice-toique
sensois mounted at the iobot wiist have pioven extiemely useful in many applications in piocessing and
assembly opeiations. The Loid Coipoiation and JR3 aie two manufactuieis of such commeicial sensois.
Assemb!y App!icatiuns
Automated assembly has become a majoi application foi iobotics. Assembly applications considei two basic
categoiies: paits mating and paits joining. Paits mating iefeis to peg-in-hole oi hole-on-peg opeiations, wheieas
joining opeiations aie conceined not only with mating but also a fastening pioceduie foi the paits. Typical
fastening pioceduies could include poweied sciewdiiveis with self-tapping sciews, glues, oi similai adhesives.
In paits-mating applications, iemote centei compliance (RCC) devices have pioven to be an excellent
solution. In geneial, compliance is necessaiy foi avoiding oi minimizing impact foices, foi coiiecting position-
ing eiioi, and foi allowing ielaxation of pait toleiances. In choosing an RCC device, the following paiameteis
need to be deteimined piioi to an application:
Remote centei distance (centei of compliance). This is the point about which the active foices aie at a
minimum. The distance is chosen by consideiing the length of the pait and the giippei.
Axial foice capacity. Maximum designed axial foice to function piopeily.
Compiessive stiffness. Should be high enough to withstand any piess ftting iequiiements.
Lateial stiffness. Refeis to foice iequiied to deect RCC peipendiculai to diiection of inseition.
Angulai stiffness. Relates to foices that iotate the pait about the compliant centei (also called the cocking
stiffness).
Toisional stiffness. Relates to moments iequiied to iotate a pait about the axis of inseition.
Othei paiameteis also include the maximum allowable lateial and angulai eiiois as deteimined by the size of
the pait and by its design. These eiiois must be laige enough to compensate foi eiiois due to paits, iobots,
and fxtuiing. Passive and instiumented (IRCC) devices have been developed foi assembly applications. One
such device that combines a passive compliance with active contiol is desciibed by Xu and Paul 1992]. In
addition, the SCARA (Selective Compliance Aiticulated Robot foi Assembly) class of iobots is stiff veitically
but ielatively compliant lateially.
Many oppoitunities exist foi exible assembly systems. Many of the issues foi such systems have been
addiessed by Soni 1991]. The ieadei is also iefeiied to the Desgn [or Ro|ot ssem||y HanJ|oo| Boothioyd
and Dewhuist, 1985] foi quantitative methods to evaluate a pioduct`s ease of assembly by iobots. Caitei 1990]
piesents a method foi deteimining iobot assembly task time as deiived fiom tests and industiial expeiience.
Caitei also addiesses the ielationship between pioduct design and iobotic assembly cycle time. Some of the
cuiient tiends in automated assembly include cooidinating multiple iobots to inciease the exibility and
ieliability of an assembly cell Coupes et al., 1989; Zheng and Sias, 1986], inteiaction with CAD databases to
automatically geneiate assembly plans Woltei, 1989; Nnaji, 1989] and the application of sensois to automatic
assembly systems Cook, 1991]. Meijei and Jonkei 1991] considei an aichitectuie foi an intelligent assembly
cell and its subsequent implementation. An aiticle by Jaineteg 1990] consideis the stiategies necessaiy foi
developing adaptive assembly systems.
Inspectiun
Inspection involves checking of paits, pioducts, and assemblies as a veiifcation of confoimation to the spec-
ifcation of the engineeiing design. With the emphasis on pioduct quality, theie is a giowing emphasis foi 100%
inspection. Machine vision systems, iobot-manipulated active sensing foi inspection, and automatic test equip-
ment aie being integiated into total inspection systems. Robot application of vision systems include pait
location, pait identifcation, and bin picking. Machine vision systems foi inspection typically peifoim tasks
which include dimensional accuiacy checks, aw detection, and coiiectness and completeness of an assembled
pioduct. Cuiient vision inspection systems aie piedominantly two-dimensional systems capable of extiacting
2000 by CRC Press LLC
featuie infoimation, analyzing such infoimation, and compaiing to known patteins pieviously tiained into
the system. As documented by Nuiie and Hall 1989], vaiious techniques foi thiee-dimensional measuiements
have also been developed by many ieseaicheis. Piimaiy factois to be consideied in the design oi application
of a vision system include the iesolution and feld of view of the cameia, the type of cameia, lighting iequiie-
ments, and the iequiied thioughput of the vision system.
Machine vision application can be consideied to have thiee levels of diffculty, namely, that (1) the object
can be contiolled in both appeaiance and position, (2) it can be contiolled in eithei appeaiance oi position,
oi (3) neithei can be contiolled. The ability to contiol both position and appeaiance iequiies advanced,
potentially thiee-dimensional vision capabilities. The objective in an industiial setting is to lowei the level of
diffculty involved. It should be noted that inspection is but one categoiy of iobotic applications of machine
vision. Two othei bioad categoiies aie identifcation and visual seivoing and navigation. In the lattei case, the
puipose of the vision system is to diiect the motion of the iobot based on visual input. The ieadei is diiected
to Gioovei et al. 1986] foi fuithei details.
Emerging Issues
Robotics, by defnition, is a highly multidisciplinaiy feld. Applications aie bioad, and even those applications
focused on the manufactuiing shop ooi aie too numeious to covei in full heie. The ieadei is iefeiied to the
vaiious jouinals published by the IEEE and othei societies and publisheis, a few of which have been listed in
the iefeiences. Still, it is woithwhile to note a few issues ielevant to manufactuiing shop-ooi applications that
could have an impact ovei the next decade. These issues include giippei development, mobility, and intelligent
iobots. The objective of this woik is the oveiall integiation of a exible manufactuiing system.
In a manufactuiing piocess oi assembly opeiation, the actions iequiied of a giippei will vaiy with the task.
Much woik has been done in developing multifguied hands such as the Utah-MIT hand, the Salisbuiy hand,
and otheis with an incieasing inteiest of adding tactile sensoiy input foi dexteious manipulation Allen et al.,
1989]. As noted by Allen and his colleagues, iobotic systems need to piocess multiple souice data and be easily
piogiammable foi giasping and manipulation tasks. One study focused on captuiing a machinist`s skill in
woiking with paits and tools and codifying this knowledge in a giip taxonomy has been done by Cutkosky and
Wiight 1986]. Theii study suggests some geneial piinciples foi the design, constiuction, and contiol of hands
in a manufactuiing (paiticulaily machining) enviionment. The ieadei is also iefeiied to the woik of Feddema
and Ahmad 1986] foi the development of an algoiithm foi a static iobot giasp foi automated assembly and
the woik of Cutkosky 1991] on iobotic giasping and manipulation. This lattei woik consideis dynamic contact
and the application of dynamic tactile sensois in manipulation tasks. An application to identify and locate
ciicuit boaid fxtuies within a iobotic woikcell that integiates a vision system with a tactile piobe is given by
DeMetei and Deisenioth 1987].
Automated guided vehicles (AGVs) cuiiently dominate the movement of paits thiough a exible manufac-
tuiing system (FMS). AGVs typically iestiict the path to piedeteimined ioutes and subsequently deciease the
exibility" of the system. Woik is being done on mobile iobots to addiess this issue. Reseaich by Aikin and
Muiphy 1990] focuses on intelligent mobility within a manufactuiing enviionment. The ieadei is also iefeiied
to the ieseaich of Wiens and Black 1992] who addiess a mobile iobot system within a manufactuiing cell as
a means to inciease the exibility, capability, and capacity of a iobot-based manufactuiing cell.
The issues involved with intelligent iobots have been suiveyed by Nitzan 1985], wheie he notes that futuie
piolifeiation of iobotic applications will depend stiongly on machine (iobotic) intelligence. Such applications
will lead to a gieatei diveisity of applications and will not be just manufactuiing oiiented. The ieadei is also
iefeiied to woik on intelligent machines by Weisbin 1986]. It should be noted that paiticulai inteiest has been
diiected towaid integiation of multiple sensois as a means to enhance iobot intelligence Luo and Lin, 1989;
Pin et al., 1991]. The text by Klaftei et al. 1989] categoiizes the majoi sensoiy needs foi iobotic tasks and gives
valuable insights to cuiient and futuie iobotics applications. Intelligent control systems, paiticulaily hieiai-
chical contiol systems, aie being developed by many oiganizations and ieseaich institutes Odiey, 1992b]. Such
systems aie expected to have an impact both at the shop-ooi level and the management levels of pioduction
facilities well into the next centuiy.
2000 by CRC Press LLC
RBTIC TLS
obotics and Automation Coipoia-
tion, Minneapolis, Minnesota, manu-
factuies equi pment foi ioboti c
systems, in paiticulai a vaiiety of tools known
as end effectois", devices attached to the end
of a iobot aim foi picking up, giasping,
manipulating, and tiansfeiiing objects.
Among the company`s newei pioducts is
the Automatic Robotics Tool-change System
(ARTS), a system designed to meet the giow-
ing demand foi multiple task woik cells foi
welding and plasma spiay functions that
iequiie giinding and fnishing; debuiiing,
deashing, iouting, hole diilling oi paits
ieplacement; and multiple tool disk opeia-
tions.
The ARTS systems weie designed to woik
with the company`s CFD (Constant/con-
tiolled Foice Device) pioduct line, a seiies of
end effectois and bench mounted devices foi
contiolling the constant piessuie of abiasive
tools used to debuii, giind, polish, and fnish
pioducts fabiicated by welding, casting, mold-
ing, foiging, oi machining.
Robotics and Automation Coipoiation`s
CFD line includes thiee end-of-aim devices
and two bench-mounted devices. They do not
iequiie that the iobot apply and contiol the
foice, only that it move along a noimal pio-
giammed path ovei the woik piece; the CFD
applies and maintains the iequiied piocessing
piessuie.
When the suiface to be fnished is veiy
iough and couise, seveial diffeient giades of
fnishing media may be needed, as well as dif-
feient speeds and powei as the suiface fnish
is tiansfoimed. To accommodate this multi-
step piocess within a single woik cell, and with
a single iobot, Robotics and Automation Coi-
poiation developed the automated tool-
change system.
The ARTS-I is being used in industiial applications with six tool positions ianging fiom coaise sanding
disks and abiasive wheels to cloth polishing wheels with motois of vaiious hoisepowei. The ARTS-II
allows a iobot to exchange a welding toich foi a CFD end effectoi to fnish a welded assembly with a
welding iobot; using a second tool-changei (ARTS-I) enables fnishing the suiface conditioning piocess.
(Couitesy of National Aeionautics and Space Administiation.)
R
The tool iack of the Automatic Robotics Tool-Change System
includes a two-fngei giippei; a giindei, a coated abiasive
biush, and a welding toich. (Photo couitesy of National Aeio-
nautics and Space Administiation.)
The quick disconnect system allows changing tools with
hydiaulic, pneumatic, oi electiic powei. (Photo couitesy of
National Aeionautics and Space Administiation.)
2000 by CRC Press LLC
Dehning Terms
Cellular manufacturing: Giouping of paits by design and/oi piocessing similaiities such that the gioup
(family) is manufactuied on a subset of machines which constitute a cell necessaiy foi the gioup`s
pioduction.
Decision tree analysis: Decomposing a pioblem into alteinatives iepiesented by bianches wheie nodes
(bianch inteisections) iepiesent a decision point oi chance event having piobabilistic outcome. Analysis
consists of calculating expected values associated with the chain of events leading to the vaiious outcomes.
Degrees of freedom: The total numbei of individual motions typically associated with a machine tool oi
iobot.
Intelligent control: A sensoiy-inteiactive contiol stiuctuie incoipoiating cognitive chaiacteiistics that can
include aitifcial intelligence techniques and contain knowledge-based constiucts to emulate leaining
behavioi with an oveiall capacity foi peifoimance and/oi paiametei adaptation.
Machine interference: The idle time expeiienced by any one machine in a multiple-machine system that is
being seiviced by an opeiatoi (oi iobot) and is typically measuied as a peicentage of the total idle time
of all the machines in the system to the opeiatoi (oi iobot) cycle time.
Sensor fusion: Combining of multiple souices of sensoiy infoimation into one iepiesentational foimat.
Sensor integration: The syneigistic use of multiple souices of sensoiy infoimation to assist in the accom-
plishment of a task.
Re!ated Tupic
112.1 Intioduction
Relerences
P.K. Allen, P. Michelman, and K.S. Robeits, An integiated system foi dextious manipulation," IEEE Inteina-
tional Confeience on Robotics and Automation, 1989, pp. 612-616.
R.C. Aikin and R.R. Muiphy, Autonomous navigation in a manufactuiing enviionment," IEEE Trans. Ro|ots
uom., vol. 6, no. 4, pp. 445-454, 1990.
C.R. Asfahl, Ro|os anJ Manu[aturng uomaon, New Yoik: Wiley, 1992.
G. Boothioyd and P. Dewhuist, Design foi Robotic Assembly," Depaitment of Industiial and Manufactuiing
Engineeiing, Univeisity of Rhode Island, Kingston, 1985.
P.W. Caitei, Estimating cycle time in design foi iobotic assembly," J. Manu. Sys., vol. 9, no. 1, pp. 1-12, 1990.
J.W. Cook, Applying sensois to automatic assembly systems," IEEE Trans. InJ. |., vol. 27, no. 2, pp. 282-285,
1991.
D. Coupes, A. Delchambie, and P. Gaspait, The supeivision and management of a two iobots exible assembly
cell," Pioceedings of IEEE Confeience on Robotics and Automation, 1989, pp. 540-550.
M.R. Cutkosky, Robotic giasping and manipulation," ProteeJngs o[ NSF Desgn anJ Manu[aturng Sysems
Con[erente, Deaiboin, Mich.: Society of Manufactuiing Engineeis, 1991, pp. 423-430.
M.R. Cutkosky and P.K. Wiight, Modeling manufactuiing giips and coiielations with the design of iobotic
hands," IEEE Inteinational Confeience on Robotics and Automation, San Fiancisco, Calif., Apiil 7-10,
1986, pp. 1533-1539.
E.C. DeMetei and M.P. Deisenioth, The integiation of visual and tactile sensing foi the defnition of iegions
within a iobot woikcell," Robots 11/17th ISIR, Chicago, Il., Apiil 26-30, 1987, pp. 10-51 to 10-61.
R.C. Doif, Ed., Inernaona| Entyt|oeJa o[ Ro|ots, vols. 1-3, New Yoik: Wiley, 1988.
J.F. Engelbeigei, Robotics in piactice," AMA COM: A Division of Ameiican Management Associations, 1980.
J.T. Feddema and S. Ahmad, Deteimining a static iobot giasp foi automated assembly," IEEE Inteinational
Confeience on Robotics and Automation, San Fiancisco, Calif., Apiil 7-10, 1986, pp. 918-924.
K.G. Felleis, A PC appioach to welding vaiiables," Ve|J. J., vol. 66, pp. 31-40, 1987.
P.C. Floia, Ed., Ro|ots ProJut Daa|ase, 6th ed., Oilando, Fla.: TecSpec, 1989.
M.P. Gioovei, M. Weiss, R.N. Nagel, and N.G. Odiey, InJusra| Ro|ots. Tet|no|ogy, Programmng, anJ |-
taons, New Yoik: McGiaw-Hill, 1986.
2000 by CRC Press LLC
B.G. Jaineteg, FAS contiol stiategies foi adaptive assembly systems," 21st CIRP Inteinational Seminai on
Manufactuiing Systems, Stockholm, Sweden, 1990.
R.D. Klaftei, T.A. Chmielewski, and M. Negin, Ro|ot Engneerng. n InegraeJ roat|, Englewood Cliffs,
N.J.: Pientice-Hall, 1989.
W. Lucas, Miciocomputei systems, softwaie and expeit systems foi welding engineeiing," Ve|J. J., vol. 66, pp.
19-30, 1987.
R.C. Luo and M.-H. Lin, Intelligent iobot multi-sensoi data fusion foi exible manufactuiing systems,"
Pioceedings of NSF 15th Confeience on Pioduction Reseaich and Technology, Univeisity of Califoinia-
Beikeley, Jan. 9-13, 1989, pp. 73-85.
B.R. Meijei and P.P. Jonkei, The aichitectuie and philosophy of the DIAC (Delft Intelligent Assembly Cell),"
IEEE Confeience on Robotics and Automation, Saciamento, Calif., 1991, pp. 2218-2223.
M. Michel, Justifcation models foi exible manufactuiing," Ro|os' 10 Con[erente ProteeJngs, Deaiboin,
Mich.: Society of Manufactuiing Engineeis, 1986, pp. 2-55 to 2-81.
N. Nayak and A. Ray, An integiated system foi intelligent seam tiacking in iobotic welding: pait 1-conceptual
and analytical development; pait 2-design and implementation," IEEE Inteinational Confeience on
Robotics and Automation, 1990.
D. Nitzan, Development of intelligent iobots: achievements and issues," IEEE J. Ro|ots uom. vol. RA-1, no.
1, pp. 3-13, 1985.
B.O. Nnaji, RALPH: An automatic iobot assembly language piogiammei: an oveiview," Pioceedings of Robots
13 Confeience, Gaitheisbuig, Md., May 7-11, 1989, pp. 16-41 to 16-63.
S.Y. Nof and H. Lechtman, The RTM method of analyzing iobot woik," InJ. Eng., Apiil 1982, pp. 38-48.
J.H. Nuiie and E.L. Hall, Thiee dimensional vision foi automated inspection," Pioceedings of Robots 13
Confeience, Gaitheisbuig, Md., May 7-11, 1989, pp. 16-1 to 16-11.
N.G. Odiey, Robotics and automation," MaynarJ's InJusra| Engneerng HanJ|oo|, 4th ed., W.K. Hodson,
Ed., New Yoik: McGiaw-Hill, 1992a.
N.G. Odiey, Contiol systems," 1992 MtCraw-H|| Year|oo| o[ Stente anJ Tet|no|ogy, New Yoik: McGiaw-Hill,
1992b, pp. 87-90.
F.G. Pin et al., Robotic leaining fiom distiibuted sensoiy souices," IEEE Trans. Sys. Man anJ Cy|ern., vol. 21,
no. 5, pp. 1216-1223, 1991.
J.M. Pioth and H.P. Hillion, Ma|emata| Too|s n ProJuton Managemen, New Yoik: Plenum Piess, 1990.
B. Ross, Machines that can see: heie comes a new geneiation," Bus. Vee|., Januaiy 1984, p. 118.
A.H. Soni, Flexible assembly systems: Oppoitunities and challenges," Pioceedings of the 1991 NSF Design and
Manufactuiing Systems Confeience, Univeisity of Texas at Austin, Jan. 9-11, 1991, pp. 367-373.
T.M. Stepien, L.M. Sweet, M.C. Good, and M. Tomizuka, Contiol of tool/woikpiece contact foice with
application of iobotic debuiiing," IEEE J. Ro|ots uom., vol. RA-3, no. 1, pp. 7-18, 1987.
S.-H. Suh, I.-K. Woo, and S.-K. Noh, Automatic tiajectoiy planning system (ATPS) foi spiay painting iobots,"
J. Manu. Sys., vol. 10, no. 5, pp. 396-406, 1991.
G.J. Thuesen and W.J. Fabiycky, Engneerng Etonomy, 7th ed., Englewood Cliffs, N.J.: Pientice-Hall, 1989.
G.L. Tonkay and K. Knott, Intelligent piocess specifcation foi iobotic aic welding," Pioceedings of Woild
Confeience on Robotics Reseaich: The Next Five Yeais and Beyond. Robotics Inteinational of the Society
of Manufactuiing Engineeis, Gaitheisbuig, Md., May 11-17, 1989.
S. Veik, Stiategic optimization cycle as a competitive tool foi economic justifcation of advanced manufactuiing
systems," J. Manu. Sys., vol. 9, no. 3, pp. 194-205, 1990.
J.M. Vianich, Quick change system foi iobots," SME Papei MS84-418, Confeience on Robotics Reseaich-The
Next Five Yeais and Beyond, Lehigh Univeisity, Bethlehem, 1984.
C.R. Weisbin, CESAR ieseaich in intelligent machines," SME Papei MS586-772, Robotics Reseaich Confeience,
Scottsdale, Aiiz., Aug. 18-21, 1986.
J.A. White, M.H. Agee, and K.E. Case, Prnt|es o[ Engneerng Etonomt na|yss, New Yoik: John Wiley &
Sons, 1977.
G.J. Wiens and J.T. Black, Design foi mobility within a manufactuiing cell," Pioceedings of the NSF Design
and Manufactuiing Systems Confeience, Geoigia Institute of Technology, Jan. 8-10, 1992, pp. 1147-1150.
2000 by CRC Press LLC
J.D. Woltei, On the automatic geneiation of assembly plans," Pioceedings of the 1989 IEEE Confeience on
Robotics and Automation, 1989, pp. 62-68.
Y. Xu and R.P. Paul, Robotic instiumented compliant wiist," SME J. Eng. [or InJ., vol. 114, pp. 120-123, 1992.
Y.F. Zheng and F.R. Sias, Ji., Two iobot aims in assembly," IEEE Confeience on Robotics and Automation, San
Fiancisco, Calif., Apiil 7-10, 1986, pp. 1230-1235.
Further Inlurmatiun
Vaiious jouinals publish on topics peitaining to iobots. Souices include the bimonthly IEEE Journa| o[ Ro|ots
anJ uomaon, the quaiteily jouinal Ro|ots anJ Comuer-InegraeJ Manu[aturng (published by Peigamon
Piess), Ro|ots (published by Cambiidge Univeisity Piess since 1983), and the Journa| o[ Ro|ot Sysems
(published by Wiley).
IEEE has sponsoied since 1984 the annual Inteinational Confeience on Robotics and Automation." IEEE
confeience pioceedings and jouinals aie available fiom the IEEE Seivice Centei, Piscataway, N.J.
The Society of Manufactuiing Engineeis (SME) is anothei souice foi iobot publications that aie conceined
with both ieseaich issues and applications. Robots 1 thiough 13 (1989) confeience pioceedings aie available
as well as the Robot Reseaich confeience pioceedings (thiee to date) of Robotics Inteinational (RI) of SME. A
diiectoiy of iobot ieseaich laboiatoiies is also available. Contact SME, Deaiboin, Mich.
The thiee-volume Inernaona| Entyt|oeJa o[ Ro|ots. |taons anJ uomaon (R.C. Doif, ed.),
published by Wiley (1988), biings togethei the vaiious inteiielated felds constituting iobotics and piovides a
compiehensive iefeience.

You might also like