Professional Documents
Culture Documents
kopacek1988
kopacek1988
Abstract: This survey gives a short introduction to the today robot structure and problems
of motion control and some remarks for future aspects of lightweight , high performance
robots, that will include parts with non negligible flexibilities . How the modelling of this
time variant elastic deformations, a substantial part of any control strategy, can principally
be done is shown based on the recent literature.
The quasistatic method uses substitutive springs to describe the elastic deformat ions at spe-
ciallocations. For the dynamic behaviour of vibrating flexible links FE-methods, a vibration
mode or shape functions approach or other combined methods may be applied. The char-
acteristics of the FE-methods are their advantage of a realistic modelling of complex shaped
links but with the drow back of high evaluation time. The vibration mode methods need
a special knowledge to select the right set of functions but lead to short evaluation time.
Combined methods are also based on a fracturizing of the deformable elements into smaller
units, whose elastic properties can be described more easily.
The modelling of the elastic properties and an adequate control will be an essential basis for
a future time and energy efficient robot performance.
Keywords: Flexible Robots, Modelling, Systems of Elastic Bodies .
21
22 P. Kopacek , K. Deso"er and P. Lugner
calculation of points on a desired path - e. g. proximity sensors provide information about pres-
straight line between two points in the working ence, distance, speed of approximation . . , They
space work without contact .
determination of the necessary command values, visual sensors are of great importance for tasks in
resulting from the tasks mentioned before, for the connection with assembly operations (part identi-
joint controllers fication , orientation ... ). They work mainly with
position control of the axes in accordance to the TV cameras.
calculated command values. auditive sensors are being de\'eloped and serve for
speech recognition.
While the calculation of the world coordinates from the
robot coordinates is always easy, the reverse case might be
difficult for some types of industrial robots . Whether a 2 General considerations on flexi-
closed solution for calculating the robot coordinates from bility
world coordinates exists depends on the kinematic structure
and the construction parameters of the robot. E. g. eccen- Kinematics, kinetics and control of industrial robots have
tricities in the joints might be responsible that the control been studied extensively under the assumption that the links
computer has to solve the inverse transformation in a recur- can be modelled as "rigid bodies", an assumption which is
sive way. warranted in the great majority of manipulators in use today,
The term "path calculation" has usually different mean- so called "heavy constructions" - in order to avoid untoler-
ings. The a priori information could consist in able posi tioning inaccuracies that may be caused by elastic
deflections and vibrations. The advantage of such a con-
a total determination of the path or struction is that angular encoders at the act uator shafts or
a determination of distinct points of a desired at the joints can be used to gain information of the actual
path together with information on the kind of the position of the end-effector in world coordinates in a purely
connecting slope. geometric manner. Therefore the control device can use this
information directly. But that is not enough if the deflections
Both, the path or distinct points might be given in robot of the substructures of a manipulator cannot be neglected
coordinates or in world coordinates. anymore .
The robot may be programmed by the "teach-in meth- For assembling tasks, for example, robots should be faster
ode" or by point or path calculation (e. g. by optimization and therefore light weight constructions are necessary. Never-
criteria) . theless, they should have better positioning accuracy. Thus,
The problem of path generation leads to the problem of lightweight constructions and flexibility of the components
calculating points along a desired path. This path might be will be consequences of compromises between
determined under different assumptions
higher acceleration and speed ,
minimum time for the motion
higher ratio payload mass/ moved total mass,
realizing the shortest distance (straight line be-
smaller actuators, lower energy consumption and
tween two points)
safer operation due to reduce masses ,
minimum energy consumption for the motion
positioning accuracy,
along the path.
range of working space ,
Calculating only the position, the robot would stop at complexity of control system,
each of the calculated or predescribed points. Therefore, for
- total costs, etc.
a continous motion of the robot, it is necessary to determine
in addition a speed and probably an acceleration in each of There are three possi bili ties for the reduction of moved
these points. mass:
Generally the control computer is structured as a host
computer for the tasks mentioned before connected to mi- 1. The links are characterized by
crocomputers which control the position of each joint. The
(a) their geometric size
outputs of the host computer are command values resulting
from the path generation in the two described ways . (b) their material , (Yo ung's modulus, mass density,
Basis of each position control design for robots are the permi t ted maximal stress)
kinematic and kinetic equations of the robot arm. For the (c) their permitted maximal deflexion .
main types of robots commercially available today these
equations are derived in the sense of rigid body systems These items determine the to tal mass . Considerations
e.g. in Desoyer, Kopacek and Troch (1985). in optimizing are presented e. g. by Book and Majette
External sensors give the control computer of the robot (1983), Kiedrzynski (1986), Weck and Stave (1986),
additional information about the surrounding. External sen- and especially in considerations of new materials by
sors available today may be classified into four main groups: Urbanek (1988).
With the high gear ratios used today the fast rotat-
ing rotors attached to moving vibrating links cause kinetostatic method dynamic methods
periodic gyroscopic moments which may produce ad- /
ditional disturbing vibrations transversal to these vi-
brations of the links_ Those are further reasons for
keeping the moment of momentum of the rotors low
(Desoyer, Kopacek, Troch, (1985); Desoyer, Lugner,
Springer, (1985)) .
• finite element methods The general form of the equations of motion for such a
other or combined methods. robot model are formally equal to those stated in equation
(2) .
The following characterization of these methods is based The numerical calculation provides directly the signals
on the well-known basic considerations and the recent litera- for the control system for positions used to determine the
ture . Due to the great number of publications and their not substitutive springs whereas for other locations an interpo-
always clear association with one of the methods only a lim- lation, may be including the static deflections , has to be
ited number of references, without classification, are stated done .
in the following chapter.
24 P. Kopacek . K. DesO\·er and P. Lugner
I
Flexible Robot
I robot with a number of L t
i .Ie=l
j. elastic OOF, where the qkj
! are assembled in the vector '1. and the OOF for the rigid body
motion assembled in vector E. Its general mathematical form
Single Structure for an autonomous system also suitable for the controller
e. g. Link i Link i +1 design is
I I (2)
ni Finite Elements
~
of Link i
For this system of generally nonlinear equations the matrix
jk Elastic OOF for r- H(~) represents a generalized mass matrix and B(!!) the
Finite Element k generalized control forces as a function of the control vector
I
Equations of Motion
I l!. The main parts of the vector L are inertial and gyroscopic
terms and the elastic and damping properties of the system.
for the Element k
I The vector;;. comprises all (generalized) coordinates of the
robot.
Geometric no I Transformation
of figure 2, takes into account that the deflections of the link
itself with respect to its rigid body reference configuration
Linearization
I to Link Frame are no longer small. Then also the derivatives of the trans-
formation matrices from the local elements to the link frame
I yc.s
I lead to nonlinearities. The dashed line in figure 2 indicates
+
To separate what can be done by off-line mathematical
- I - -
I modelling and preparation and the following online effort
Transformation
to Global System
I
I a demarcation line is set in fig . 2. It seems appropriate to
include the transformation from the link frames to the ref-
erence frame of the world coordinates for the global robot
I
Assembling of I
I motion into the numerical evaluation for each time step.
As most investigations show, even a small number of fi-
Link Equations
n;
L L jk Elastic OOF
r- ,...J I nite elements per link and a restriction to bending in only
one plane make a real time computing for a robot with two
i 11:::- I or more flexible links impossible today.
I I 3.2.2 Vibration mode approach
Solution Solution
The elastic deformations of a flexible link i are described by
a set of ni shape functions !Iij(r:;) with their time dependent
Figure 2: Scheme of FE-modelling. amplitudes qij( t) representating the elastic OOF of the link
Modelling of Flexible Robots 25
- e. g. Book , Maizza-Neto and Whitney (1975) , Matsuoka in the highest order time derivatives. The nonlinear branch,
and Citron (1985) , Cetinkunt, Siciliano and Book (1986) , indicated by the dashed line in the right branch, comprises
Nicosia, Tomei and Tornambe (1986), Johanni (1986) . So the same steps, but will naturally lead to a more complex
the elastic deformation !!!i(r;, t) at a position r; of the link i set of system equations . Again the resulting equations can
can be formulated by a series expansion or modal expansion be written in the form of equation (2) .
As indicated by the demarcation , most of the preparation
Yl;(r;,t) = I:llj;(r;)qj;(t) (3) work for the evaluation can be done omine. The selection of
j =l proper shape functions will always be a problem connected to
the special application, but it will be increasingly supported
For these functions l!.j;(r;) a set of vibration modes or eigen-
by computer aided algorithms. So e. g. a precalculation with
functions may be used, that are directly derived from the
FE-methods for a link may lead to approximat ions for the
partial differential equations and the individual boundary
vibration modes. Because most of the offline work could be
conditions describing the behaviour of the considered flex-
done in analytical handy form , the real evaluation will be
ible part . So this approach can be applied for relatively
very time efficient.
simple boundary conditions and geometry of the parts only
- e. g. Sakawa and Matsuno (1986), Balas (1978) - because A comparison of these two dynamic methods with re-
for a fast moving part connected to arbitrary moving neigh- spect to quality of information, cal c ula~ion and evaluation
bours the boundary conditions will be very complex. efforts will remain a problem, because of the number of FE
Therefore generally a Ritz approach, an appropriate set or number and kind of the shape functions chosen, e. g. Ho-
of shape functions not directly related to the basic differ- henbichler (1987), Plockinger (1987).
ential equation and boundary conditions will be employed The FE-methods have better possibilities to model links
- e.g. Truckenbrodt (1978), (1981), (1982). So e.g. often of geometrically complex shapes, but at the cost of a greater
shape functions corresponding to the eigenfunctions of a number of FE-elements and thereby evaluation time. On
clamped-free beam are used to describe the deflections of the other hand , eigenfunctions or shape functions for com-
the joint at the end of the beam, thus providing the position plex shaped links need a great preparation effort (e. g. with
and the attitude for the clamped end of the next beam in a employing FE-methods as mentioned before) and experience
chain structured robot arm. to utilize the smaller number of DOF.
In figure 3, showing a similar overview like fig . 2, for a Using the FE-methods no problems occur connecting the
robot with flexible links, the third level indicates the decom- links, whereas using shape functions which violate boundary
position of the elastic deflexion of a link i into n; shape func- conditions introduce a systematic incoherence into the sys-
tions with an equal number of elastic DOF. As indicated in tem. But increasing the accuracy means more shape func-
tions , taking into account the boundary conditions more pre-
the left branch, a (geometrical) linearization with respect to
cisely, and that will result in a higher evaluation time.
the influences of preceding elastic deformations will provide
In contrast to the FE-methods a description of the flex-
a set of differential equations which has no strong coupling
ibilities by the vibration mode methods , together with an
individual adaptation, are more likely to enable a real time
I
I Flexible Robot
I 1
evaluation in the soon future .
I yes
I I edged that a proper combination of different features of the
known methods may lead to an improved precision of the
Assembling of
Link Equations ~.-J I I:::
o
calculated results together with high time efficiency of the
evaluation.
2:: Tt; Elastic DOF
;
- --
I -
I I 4 Conclusions
Solution Solution c
I I CJ
c
~.,!g
c :>
.2
~
The construction and the design of future high performance
robots have to consider all components from configurations
0:':
:; and shapes of links, joints, actuators , the system modelling
Figure 3: Scheme of the vibration mode modelling. with the background of available computer capacity to the
26 P. Kopace k. K. Deso\'er and P. Lugner
possibility of advanced control algorithms to achieve the re- Cetinkunt, S., Siciliano, B. and Book, W.J. (1986) . Sym-
quired precision and working speed. bolic Modelling and Dynamic Analysis of Flexible Ma-
The elastic deformations and their prediction with math- nipulators, Proc. IEEE Symp. Mcc 1986, pp . 6.
ematical models will be essential as input quantities for fu- Chang, K. H., Funakubo, C. , Dohi , T ., Komeda , T. and
ture control systems and will thereby determine to a large Sakuma, 1. (1986). .4.n Approach to the Vibration
extend the performance range of the robot. Analysis of a Rob ot Arm Based on a Linearized Dy·
While the FE-methods offer relatively high precision, that namic Equation. Proc. "16 th ISIR", Brussels pp.
means results close to reality, today their evaluations cannot 355-364.
be done in real time and therefore the methods cannot be
Chassiakos, A. G. and Bekey, G. A. (1986). On the Modell-
applied in real time contro!' The movement of simple arm
ing and Control of a Flexible Manipulator Arm by
structures or arms with only one f\eJcible component are con-
Point Actuators. Proc. 25 th Conf. on Decision and
trolled today by using a mechanical system modelling similar
Control, Athens, Greece, Dec. 1986, TP3-4:30.
to the vibration mode methods. We think, that an expansion
of these methods together with improved control concepts Desoyer, K., Kopacek, P. and Troch, 1. (1985). Industriero-
will set the lead to first steps to an industrial application of boter und Handhabungsgerate, Oldenbourg Verlag
light weight, high speed robots. Munchen, 1985.
Desoyer, K., Lugner, P. and Springer, H. (1985) . Dynamic
Effects of Active Elements in Manipulators and their
5 References Influence upon the Controlling Drives,
Alberts, Th. E ., Dickerson, St. and Book W. J . (1985). Mod - IUTAM/IFToMM Symp. "Dynamics of Multi-
body Systems", Udine 1985, Ed.: G. Bianchi and
elling and Control of FleJcible J" Ianipulators, Proc. Ro-
W. Schiehlen, Springer- Verlag Berlin - Heidelberg,
bots 9 Conf., Detroit, Michigan, June 1985, pp. 1/1-
1986, pp. 43-45.
59.
Alberts, Th.E., Dickerson, St. and Book W.J. (1986). On Desoyer, K., Kopacek, P., Lugner P. and Troch 1. (1986) .
the Transfer Function Modelling of FleJcible Structures Flexible Robots - A Survey. Preprints IFAC Symp.
with Distributed Damping. Prepr. ASME Winter "Theory of Robots", Vienna 1986, late paper
Annual Meeting, pp. 22 . Volume, pp. 41·52 and (1988) IFAC Proc. Series,
Nr. 3, pp. 23-34.
Austin, F . and Pan, H. H. (1970) . Planar Dynamics of Free
Rotating Flexible Beams with Tip Masses, AIAA- Drent, K. H. (1986) . Dynamic Compensation of Elastic De·
Journal, Vo!. 8, No. 4, pp. 726-733. formations of Industrial Robot .4.rms, Proc. "16 th
ISIR", Brussels, pp. 2;1-281.
Bagci, C. and Kalaycioglu, S. (1979). Elastodynamics of
Planar Mechanisms Using Planar Actual Finite Line Dubowsky, S. and Gardner T . N. (1977). Design and Anal-
Elements, Lumped Mass Systems, Matrix-exponential ysis of MultjJink Flexible Mechanism With Multiple
Method, and the Method of "Critical Geometry Kineto Clearance Connections. Journal of Engineering for
Elasto Statics", Journal of Mechanical Design. Industry, Vo!. 99, pp. 88-96.
Balas, M. J. (1978). Feedback Control of Flexible Systems. Geradin, M., Robert, G., and Bernardin, C. (1984). Dy-
IEEE Trans. on Automatic Control, Vo!. AC-23, namic Modelling of Manipulators with Flexible
Nr. 4, pp. 6i3-679 . Members . Advanced Software in Robotics
(Eds. A. Danthine and M. Geradin), Elsevier Science
Barraco, A., Cuny, B. and Ishiomin, G. (1985) . Dynamics of
Publishers, pp. 27-39.
Flexible Systems, IUTAM / IFToMM-Symposium,
"Dynamics of Multibody Systems", Udine, Ed .: Goldenberg, A. A. and Rakhsha, F . Simulation Model of a
G. Bianchi and W. Schiehlen, Springer- Verlag Berlin Structurally Flexible Robot, Int. Conf. on Ind. Pro-
1986, pp. 1-16. cess Modelling and Control, Hangzhou, China.
Book, W. J., Maizza-Neto, W. and Whitney, D. E. (19;5). Haugh, E. J., Wu, S. C., Kim S. S. (1985). Dynamics of Flex·
Feedback Control of Two Beam. Two Joint Systems ible Machines: A Variational Approach,
with Distributed Flexibility. Journal of Dyn. Sys- IUTAM / IFToMM-Symposium "Dynamics of
tems, Meas. and Control, pp. 424-431. Multibody Systems", Udine, Ed.: G . Bianchi
Book, W . J. and Majette, M. (1983) . Controller Design for and W. Schiehlen, Springer· Verlag Berlin 1986, pp. 55-
68 .
Flexible Distributed Parameter Mechanical Arms via
Combined State Space and Frequency Domain Tech- Hemami , A. (1985). Studies on a Light Weight and Flexible
niques. Journal of Dyn. Systems, Meas. and Robot Manipulator. Robotics, Vo!. 1, pp. 27-36.
Control, Vo!. 105, pp. 245-254. Hirschberg, W. (1985). Theoretische Untersuchungen des
Bremer, H. (1977). Bewegungsgleichungen hybrider Systeme dynamischen Verhaltens geregelter, elastischer Indus-
beliebiger Anordnung. ZAMM 57, T5;-T59. trieroboter, (Ph. D. Thesis), Technical University
Bremer, H. (1978). Zur Dynamik hybrider Mehrkorpersys- Graz, Austria, 1985, 151 pp .
teme. Diss. TU Munchen. Hohenbichler, G. (1987). Simulation flexibler Roboter: Mo·
Bremer, H. (1980). Dampfungsdurchdringungen bei hybri- dellbildung mit Ansatzfunktionen, Diploma Thesis,
den Mehrkorpersystemen. ZAMM 60, T51-T53. Techn. Univ. Vienna, Oct. 198;.
Bremer, H. (1981) . Bewegungsgleichungen hybrider Hu A., Skelton , R. E. and Yang, T. Y. (1985). On the Se-
Mehrkorpersysteme: .4.nalytische und synthetische Ver- lection of Finite Element Models for Control Design,
fahren. ZAMM 61, T27-T29. Proc. 1985 Amer. Control Conf., pp. 1648-1656.
Huebner, H. K. (19;5). The Finite Element Methods for En·
Bremer, H. (1983). Kinetik starr-elastischer Mehrkorpersy-
steme. Fortschritt Berichte der VDI-Zeitschrift gineers. J . Wiley 19;5.
Reihe 11, Nr. 53. '
\Iodelling of Flexihle Robot s
Huston , R. L. and Passarello, C . E. (1984). Finite Element Midha, A. , Erdman, A. G . and Frorib . D. A. (1978). Fi-
.'Hethods - An Introduction. Mech. Eng., Vol. 25. nite Element Approach to :\[athematical Jlodelling of
Huston, R. L. (1980). Flexibility Effects in ,Y lultibody Sys- High-Speed Elastic Linkages. Mechanism and
tem Dynamics. l\lechanics Res_ Communications, Machine Theory, Vol. 13 , pp. 603-6 18.
Vol. 7, pp. 261-268 . Natke, H. G . (1978). Fergleich von Algorithmen fiir die An-
Johanni R. (1986). On the Automatic Generation of the passung des Rechenmodells einer schlVingunsfiihigen
Equations of Motion for Robots with Elastically De- elasto-mechanischen Struktur an \'ersuchswerte.
formable Arms, Prepr_ IFAC Symp. on "Theory ZAMM, Vol. 59, pp. 257- 268.
of Robots" , Vienna, Dec. 1986 , pp. 195-199 and (1988) Nicosia , S., Tomei, P. and Tornambe, A. (1986). Dynamic
"Theory of Robots", IFAC Proc_ Series Nr . 3, pp. Modelling of Flexible Robot Manipulators, Proc.IEEE
143-147. Conf. Rob. & Autom . , Vol. 1, pp. 365-372.
Jonker, J . B. (1985). Dynamics of .4cti\·e Mechanisms with Pl6ckinger, P. (1987). Simulation flexibler Roboter: Mod-
Flexible Links, IUTAM / IFToMM Symposium ellbildung mit Methoden der nniten Elemente, Diplo-
"Dynamics of Multibody Systems", lJdine, Ed.: ma Thesis, Techn . Univ. Vienna, Oct. 1987.
G. Bianchi and W. Schiehlen , Springer- Verlag Berlin Rivin, E.!. and Zeid, A. (1986). Structural Compliance and
1986, pp . 103-118. the De\,iations from Programmed Path in Jointed Ma-
Jonker, J. B. (1986) . Dynamic Simulation of Robotic Ma- nipulators. Proc . of the "16 th ISIR", Brussels, pp .
nipulators with Flexible Links, Proe. of the "16 th 45-53.
ISIR", Brussels, pp. 365-376. Rubin, S. (1975). Impro\'ed Component -Mode Representa-
J udd, R. P. and Falkenburg, D . R . (1985). Dynamics of Non- tion for Structural Dynamic .4nalysis, AIAA-Jour-
rigid _4rticulated Robot Linkages. Trans. on Auto- nal, Vol. 13, No. 8, pp . 995-1006.
matic Control, Vol. AC-30 , No. 5, pp . 499-502. Sakawa, Y. and Matsuno , F. (1986). Modelling and Control
Kanoh, H., Tzafestas, S. , Lee , H. G. and Kalat J. of a Flexible ,Io,fanipulator with a Parallel Drive Mech-
(1986). Modelling and Control of Flexible Robot Arms. anism, Int. J. Control 44, pp . 299-313.
Proc. 25 th Conf. on Decision and Control, Athens , Sakawa, Y., Matsuno, F. and Fukushima, S. (1985). Mod-
Greece, Dec. 1986, FP3-3:30. elling and Control of a Flexible Robot Arm. Journal
Kelly, F.A. and Huston, R.L. (1981). ModellingofFlexibil- of Robotic Systems, Vol. 2, pp. 453-472.
ity Effects in Robot Arms. Proc. "JACC", Char- Schiehlen, W . (1984) . Dynamik von Robot ern , MAN-
lottesville, Vol. 1, pp. WP-2C / 1-5. Technologie Bericht.
Kelly, F. A. and Huston, R . L. (1985). Statics and Dynam- Schiehlen, W . and Rauh , J. (1986) . Modelling of Flexible
ics of a Flexible Manipulator , Proc. 1985 Comput- Multibeam Systems in Rigid-Elastic Supereiements ,
ers in Eng. Conf., pp . 87-94. Revista Brasiliera.
Kiedrzynski, A. (1985). Modelling of the Elastic Links of Schmitz, E. (1986). Dynamics and Control of a Planar Ma-
Closed-Loop-Robots. Proc. Conf. on "Advanced nipulator with Elastic Links. Proc. 25 th Conf. on
Robots", Tokyo , pp. 283-288 . Decision and Control, Athens, Greece, Dec.
Kiedrzynski, A. (1986). Mass-Stiffness Analysis of Robot 1986 , TP3-3 :30 .
Links. Proc. "16 th ISIR", Brussels, pp. 151-1.58. Singh, R . P ., Van der Voort , R . J. and Likins , P.W. (1985).
Likins, P. W . and Wirsching , P. H. (1968). Use of Synthetic Dynamics of Flexible Bodies in Tree Topology - A
Modes in Hybrid Coordinate Dynamic Analysis. Computer-Orient ed Approach. J ourn. Guidance and
AIAA-Journal, Vol. 6. pp. 1867-1873. Control, Vol. 8, pp. 584-590.
Lilov, L. (1985). Bewegungsgleichungen elastischer _II,Iehr- Spong, M . W. (19 86). Modelling and Control of Elastic Joint
korpersysteme. ZAM,\ !. \"01. 65 , pp . T74-T76. Robots. Prepr. 1986 ASl\IE \Vinter Annual Meet-
Lilov, L. and Wittenburg, J. (1985) . Dynamics of Chains of ing, Anaheim , California, pp. 9.
Rigid Bodies and Elastic Rods with Re\'olute and Pris- Springer, H. , Lugner, P. and Desoyer , K. ( 1985). Equations
matic Joints . IUTAM / IFToMM Symp. "Dynam- of Motion for Manipulators lncluding Dynamic Effects
ics of Multibody Systems", LTdine. Ed.: of Active Elements , Preprints of the
G. Bianchi and W. Schiehlen , Springer- Verlag Berlin "Symp. on Robot Control", Barcelona, Late Paper
1986, pp.141-152. Vol., pp. 32-37.
Lilov, L. (1985). Dynamics of Elastic Multibody Systems In- Sunada, W. and Dubowsky, S. (1981). The Application of
\'GIving Closed Loops. IUTAM / IFToMM Symp. Finite Element Methods to the Dynamic Analysis of
"Dynamics of l\lultibody Systems", Udine. Ed. : Flexible Spatial and Co-Planar Linkage Systems.
G. Bianchi and W . Schiehlen, Springer- Verlag Berlin Journ. Mech. Design, Vol. 103, pp. 643-651.
1986, pp. 129-140 . Sunada. W . and Dubowsky, S. (1983). On the Dynamic
Maizza-Neto, O. (1980). Modal Approach for Modelling Analysis and Beha\'iour of Industrial Robotic Manip-
Flexible Manipulators: Experimental Results. Proc. of ulators \I'ith Elastic ,\Iembers. Trans. ASME, J. of
the "8 th IFAC Symp. Automatic Control in Mechanisms, Trans. and Automation in Design,
Space", pp. 405 - 411. Vol. 105 , pp. 42 - 51.
Matsuoka, K . and Citron. S. J . (1985). Symbolic Processing Truckenbrodt, A. ( 1978) . Regel ung eines flex..i blen Manipu-
and Dynamic Simulation of Equations of .\lotion for latorarms. ZAI\1:\I 58, T!84-T! 85.
Flexible Manipulators . Proc. ISA / S5, (Ad\". [nstr. Truckenbrodt, A. (19 80 ). Be wegungs\'erhalten und Regelung
40 pt. 2), pp. 1601 - 1610. hybrider Mehrkorpersysteme mit .4.n\\'endung auf In-
dustrieroboter. VDI-Bericht, :"ir. 33. Reihe 8.
RC - B-
28 P. Kopacek. K. DesO\'er and P. Lugner