IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 40, NO.

3, MARCH 1995 419

A New Model for Control of Systems with Friction

[Link] de Wit, Associate, IEEE, H. Olsson, Student Member, IEEE, K. J. h t r o m , Fellow, IEEE, and P. Lischinsky

Abstract-In this paper we propose a new dynamic model for very much resembles that of a connection with a stiff spring
friction. The model captures most of the friction behavior that has with damper and is sometimes referred to as the Dahl effect.
been observed experimentally. This includes the Stribeck effect, Later studies (see, e.g., [l], [2]) have shown that a friction
hysteresis, spring-like characteristics for stiction, and varying
break-away force. Properties of the model that are relevant to model involving dynamics is necessary to describe the friction
control design are investigated by analysis and simulation. New phenomena accurately.
control strategies, including a friction observer, are explored, and A dynamic model describing the spring-like behavior during
stability results are presented. stiction was proposed by Dahl [3]. The Dahl model is essen-
tially Coulomb friction with a lag in the change of friction
force when the direction of motion is changed. The model has
I. INTRODUCTION many nice features, and it is also well understood theoretically.

F RICTION is an important aspect of many control sys-

tems both for high quality servo mechanisms and simple
pneumatic and hydraulic systems. Friction can lead to tracking
Questions such as existence and uniqueness of solutions and
hysteresis effects were studied in an interesting paper by
Bliman [4]. The Dahl model does not, however, include the
errors, limit cycles, and undesired stick-slip motion. Control Stribeck effect. -An attempt to incorporate this into the Dahl
strategies that attempt to compensate for the effects of friction, model was done in [5] where the authors introduced a second-
without resorting to high gain control loops, inherently require order Dahl model using linear space invariant descriptions. The
a suitable friction model to predict and to compensate for Stribeck effect in this model is only transient, however, after a
the friction. These types of schemes are therefore named velocity reversal and is not present in the steady-state friction
model-based friction compensation techniques. A good friction characteristics. The Dahl model has been used for adaptive
model is also necessary to analyze stability, predict limit friction compensation [6], [7], with improved performance as
cycles, find controller gains, perform simulations, etc. Most the result. There are also other models for dynamic friction.
of the existing model-based friction compensation schemes Armstrong-HClouvryproposed a seven parameter model in [ 11.
use classical friction models, such as Coulomb and viscous This model does not combine the different friction phenomena
friction. In applications with high precision positioning and but is in fact one model for stiction and another for sliding
with low velocity tracking, the results are not always satis- friction. Another dynamic model suggested by Rice and Ruina
factory. A better description of the friction phenomena for [8] has been used in connection with control by Dupont [9].
low velocities and especially when crossing zero velocity This model is not defined at zero velocity. In this paper we
is necessary. Friction is a natural phenomenon that is quite will propose a new dynamic friction model that combines the
hard to model, and it is not yet completely understood. The stiction behavior, i.e., the Dahl effect, with arbitrary steady-
classical friction models used are described by static maps state friction characteristics which can include the Stribeck
between velocity and friction force. Typical examples are effect. We also show that this model is useful for various
different combinations of Coulomb friction, viscous friction, control tasks.
and Stribeck effect [l]. The latter is recognized to produce a
destabilizing effect at very low velocities. The classical models
explain neither hysteretic behavior when studying friction 11. A NEW FRICTION MODEL
for nonstationary velocities nor variations in the break-away The qualitative mechanisms of friction are fairly well un-
force with the experimental condition nor small displacements derstood (see, e.g., [l]). Surfaces are very irregular at the
that occur at the contact interface during stiction. The latter microscopic level and two surfaces therefore make contact at
a number of asperities. We visualize this as two rigid bodies
that make contact through elastic bristles. When a tangential
Manuscript received August 27, 1993; revised February 15, 1994. Recom- force is applied, the bristles will deflect like springs which
mended by Associate Editor, T. A. Posbergh. This work was supported in part
by the Swedish Research Council for Engineering Sciences (TFR) Contract gives rise to the friction force; see Fig. 1.
91-721, the French National Scientific Research Council (CNRS), and the EU If the force is sufficiently large some of the bristles deflect
Human Capital and Mobility Network on Nonlinear and Adaptive Control so much that they will slip. The phenomenon is highly
ERBCHRXCT 93-0380.
C. Canudas de Wit and P. Lischinsky are with Laboratoire d'Automatique random due to the irregular forms of the surfaces. Haessig
de Grenoble, URA CNRS 228, ENSIEG-INPG, B.P. 46, 38402, Grenoble, and Friedland [101 proposed a bristle model where the random
France. behavior was captured and a simpler reset-integrator model
H. Olsson and K. J. Astrom are with the Department of Automatic Control,
Lund Institute of Technology, Box 118, S-221 00 Lund, Sweden. which describes the aggregated behavior of the bristles. The
IEEE Log Number 9408272. model we propose is also based on the average behavior of
0018-9286/95$04.00 0 1995 IEEE

~ ?

T 1 1 1 1
1 ' '

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 , ,,,

420 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 40, NO. 3, MARCH 1995

Note, however, that when velocity is not constant, the dy-

namics of the model will be very important and give rise
to different types of phenomena. This will be discussed in
Section IV.

Relation to the Dahl Model

The model reduces to the Dahl model if g(v) = FC/(TO,
and (TI = ( ~ =
2 0. Equations (1) and ( 3 ) then give

Fig. 1. The friction interface between two surfaces is thought of as a contact

dF -
_ - u0-
dz = (1 - $sgn(v)). (5)
between bristles. For simplicity the bristles on the lower part are shown as dt dt
being rigid.
Dahl actually suggested the more general model
the bristles. The average deflection of the bristles is denoted
by z and is modeled by
see [l 11. Most references to Dahl's work, however, do use
the simpler model (5). Dahl's model accounts for Coulomb
where v is the relative velocity between the two surfaces. friction but it does not describe the Stribeck effect.
The first term gives a deflection that is proportional to the
integral of the relative velocity. The second term asserts that An Extension of the Dahl Model
the deflection z approaches the value An attempt to extend Dahl's model to include the Stribeck
effect was made by Bliman and Sorine [5]. They replaced the
time variable t by a space variable s through the transformation
in steady state, i.e., when is constant. The function g is Pt

positive and depends on many factors such as material prop-

erties, lubrication, temperature. It need not be symmetrical.
Direction dependent behavior can therefore be captured. For Equation ( 5 ) then becomes
typical bearing friction, g (v ) will decrease monotonically from dF =
- -go- F + aosgn(v)
g(0) when v increases. This corresponds to the Stribeck effect. ds Fc
The friction force generated from the bending of the bristles
is described as which is a linear first-order system if sgn(v) is regarded as an
input. Bliman and Sorine then replaced (6) by the second-order
dz
F = (TOZ 01- +dt
model
d2F dF
where (TO is the stiffness and 01 a damping coefficient. A ds2 + 2(w- d s
- + w 2 F = w2Fcsgn(v)
term proportional to the relative velocity could be added to
the friction force to account for viscous friction so that to imitate the Stribeck effect with an overshoot in the response
dz to sign changes in the velocity. This model, however, will only
F = (TOZ 01- +dt
+
022). (3) give a spatially transient Stribeck effect after a change of the
direction of motion. The Stribeck effect is not present in the
The model given by (1) and (3) is characterized by the
steady-state relation between velocity and friction force.
function g and the parameters (TO, 01 and 0 2 . The function
+
aog(v) ( T ~ Wcan be determined by measuring the steady-
state friction force when the velocity is held constant. A 111. MODELPROPERTIES
parameterization of g that has been proposed to describe the The properties of the model given by (1) and (3) will now
Stribeck effect is be explored. To capture the intuitive properties of the bristle
aog(v) = FC + ( F s - Fc)e-("/"a)2 (4) model in Fig. 1, the deflection z should be finite. This is indeed
the case because we have the following property.
where FC is the Coulomb friction level, FS is the level of the Property I : Assume that 0 < g(v) 5 a. If lz(0)I 5 a then
stiction force, and U, is the Stribeck velocity; see [I]. With Iz(t)l 5 a v t 2 0.
this description the model is characterized by six parameters Proof: Let V = z2/2, then the time derivative of V
(TO, 01, ( ~ 2 Fc,
, Fs, and U,. It follows from (2)-(4) that for evaluated along the solution of (1) is
steady-state motion the relation between velocity and friction
force is given by d V = z(v - - 21211
- )
dt g(v)
= oog(v)sgn(v) + (TZU
FS5(v)
= Fcsgn(v) + ( F S - Fc)e-("/"s)2sgn(v) +(TZU.

- 1 7 1 1 -

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 CANUDAS DE WIT et al.: CONTROL OF SYSTEMS WITH FRICTION 42 1

TABLE I 1 Friction force [NI A

PARAMETER
VALUES USEDIN ALL SIMULATIONS

Parameter Value unit

The derivative is negative when IzI > g(u). Since g ( u ) Fig. 2. Presliding displacement as described by the model. The simulation
is strictly positive and bounded by a, we see that the set was started with zero initial conditions.

R = { z : IzI 5 a } is an invariant set for the solutions of

(l), i.e., all the solutions of z ( t ) starting in R remain there. very large, and therefore it is essential to have (TI # 0 to have
a sufficiently damped motion. The viscous friction coefficient,
Dissipativity u2, is normally not sufficiently large to provide good damping.

Intuitively we may expect that friction will dissipate energy.

Since our model given by (1) and (3) is dynamic, there may be MODELBEHAVIOR
IV. DYNAMICAL
phases where frictian stores energy and others where it gives As a preliminary assessment of the model we will inves-
energy back. It can be proven that the map cp : U H z is tigate its behavior in some typical cases. They correspond
dissipative for our model. For more details on the concepts to standard experiments that have been performed. In all the
and definitions concerning dissipative systems, see [ 121. simulations the function g has been parameterized according
Property 2: The map cp : U H z , as defined by (l), is to (4) and the parameter values in Table I have been used.
dissipative with respect to the function V ( t )= 4 z 2 ( t ) , i.e., The parameter values have to some extent been based on

I’ z(r)u(~)d.r 2 V ( t )- v ( 0 ) .
experimental results [l]. The stiffness 00 was chosen to give
a presliding displacement of the same magnitude as reported
in various experiments. The value of the damping coefficient
Proofi It follows from (1) that <
01 was chosen to give a damping of = 0.5 for the linearized
equation (9) with a unit mass. The Coulomb friction level FC
corresponds to a friction coefficient p M 0.1 for a unit mass,
and Fs gives a 50% higher friction for very low velocities.
dz The viscous friction u2 and the Stribeck velocity U , are also
2 z-.
dt of the same order of magnitude as given in [I].
Hence The different behaviors shown in the following subsections
cannot be attributed to single parameters but rather to the
L t z ( 7 ) u ( r )d r 2 L t z ( T ) F d r 2 V ( t )- V(0). behavior of the nonlinear differential equation (1) and the
shape of the function g. The presliding displacement and
the varying break-away force are due to the dynamics. This
Linearization in Stiction Regime
behavior is also present in the Dah1 model. The Stribeck shape
To get some insight into the behavior of the model in the of g together with the dynamics give rise to the type of
stiction regime we will consider a mass m in contact with a hysteresis observed in the subsection on frictional lag.
fixed horizontal surface. Let x be the coordinate of the mass,
i.e., U = d x / d t . The equation of motion becomes A. Presliding Displacement

m-
d 2 x = -F = -o0z - g l -d z - d x Courtney-Pratt and Eisner have shown that friction behaves
dt2 dt dt (7) like a spring if the applied force is less than the break-away
where z is given by (1). Linearizing (1) around z = 0 and force. If a force is applied to two surfaces in contact there will
U = 0 we get
be a displacement. A simulation was performed to investigate
if our model captures this phenomenon. An external force was
applied to a unit mass subjected to friction. The applied force
was slowly ramped up to 1.425 N which is 95% of Fs. The
Inserting (8) into (7) gives force was then kept constant for a while and later ramped down
to the value -1.425 N, where it was kept constant and then
d2x dx
m-
dt2
(CTI + +
~ 2 ) -
dt
+
uox = 0. (9) ramped up to 1.425 N again. The results of the simulation
are shown in Fig. 2 where the friction force is shown as a
This shows that the system behaves like a damped second- function of displacement. The behavior shown in Fig. 2 agrees
order system. Notice that the bristle stiffness, CO, is usually qualitatively with the experimental results in [ 131.

7-
~ ~

I’ 1 ‘I 7

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 422 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 40,NO. 3, MARCH 1995

1Friction force [NI b

1.4.

Y ,
1.2.
Fig. 5. Experimental setup for stick-slip motion.

1Position [m]
1.

1 Velocity [ds]
0 I. io-) 2. 3.G
Fig. 3 . Hysteresis in friction force with varying velocity. The velocity
variation with the highest frequency shows the widest hysteresis loop.
k4
0
13J-Friction force [NI
io m
Time [SI
c

Y Velocity [ds] F

03
Time [SI
0 10 m
Fig. 6 . Simulation of stick-slip motion.

1.2-
did depend on the rate of increase of the force but not on the
0
dwell-time; see also [161. Simulations were performed using
0

1.1 - 0
our model to determine the break-away force for different rates
0
of force application. Since the model is dynamic, a varying
1-
0
0 0 0 0
break-away force can be expected. A force applied to a unit
mass was ramped up at different rates, and the friction force
Force rate [Nk] when the mass started to slide was determined. Note that since
the model behavior in stiction is essentially that of a spring,
there will be microscopic motion, i.e., velocity different from
zero, as soon as a force is applied. The break-away force was
therefore determined at the time where a sharp increase in the
B. Frictional Lag velocity could be observed. Fig. 4 shows the force at break-
Hess and Soom [14] studied the dynamic behavior of fric- away as a function of the rate of increase of the applied force.
tion when velocity is varied during unidirectional motion. They The results agree qualitatively with the experimental results
showed that there is hysteresis in the relation between friction in [151 and [16].
and velocity. The friction force is lower for decreasing veloci-
ties than for increasing velocities. The hysteresis loop becomes [Link]-Slip Motion
wider at higher rates of the velocity changes. Hess and Soom Stick-slip motion is a typical behavior for systems with
explained their experimental results by a pure time delay in friction. It is caused by the fact that friction is larger at rest than
the relation between velocity and friction force. Fig. 3 shows
during motion. A typical experiment that may give stick-slip
a simulation of the Hess-Soom experiment using our friction motion is shown in Fig. 5. A unit mass is attached to a spring
model. The input to the friction model was the velocity which with stiffness k = 2 N/m. The end of the spring is pulled
was changed sinusoidally around an equilibrium. The resulting with constant velocity, i.e., d y l d t = 0.1 m/s. Fig. 6 shows
friction force is given as a function of velocity in Fig. 3. Our results of a simulation of the system based on the friction
model clearly exhibits hysteresis. The width of the hysteresis model in Section 11. The mass is originally at rest and the
loop also increases with frequency. Our model thus captures force from the spring increases linearly. The friction force
the hysteretic behavior of real friction described in [14]. counteracts the spring force, and there is a small displacement.
When the applied force reaches the break-away force, in this
C . Varying Break-Away Force case approximately aog(O), the mass starts to slide and the
The break-away force can be investigated through exper- friction decreases rapidly due to the Stribeck effect. The spring
iments with stick-slip motion. In [15] it is pointed out that contracts, and the spring force decreases. The mass slows
in such experiments the dwell-time when sticking and the down and the friction force increases because of the Stribeck
rate of increase of the applied force are always related and effect and the motion stops. The phenomenon then repeats
hence the effects of these factors cannot be separated. The itself. In Fig. 6 we show the positions of the mass and the
experiment was therefore redesigned so that the time in stiction spring, the friction force and the velocity. Notice the highly
and the rate of increase of the applied force could be varied irregular behavior of the friction force around the region where
independently. The results showed that the break-away force the mass stops.

- 1 I I 1 -1
I -

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 ~

CANUDAS DE WIT et al.: CONTROL OF SYSTEMS WITH FRICTION 423

Fig. 7. Block diagram for the servo problem with PID controller
I I
Fig. 9. Block diagram for the position control problem using a friction
observer.

it. Since the model is dynamic and has an unmeasurable state,

some kind of observer is necessary. This will be discussed
next.
Position Control with a Friction Observer: Consider the
problem of position tracking for the process (10). Assume that
I "e [SI the parameters 00, 01, and 02, and the function g in the friction
0 io 40 &, 80
.-
100
model are known. The state z is, however, not measurable and
hence has to be observed to estimate the friction force. For this
Fig. 8. Simulation of the PID position control problem in Fig. 7.
we use a nonlinear friction observer given by

V. FEEDBACK
CONTROL - 14
d i =U--i-ke, k>O
dt dv)
To further illustrate the properties of our friction model we
will investigate its application to some typical servo problems.
First we will use it to show that it predicts limit cycle
oscillations in servos with PID control. We will then use it and the following control law
to design observer based friction compensators.
u = -H(s)e + F + m-d2xd
dt2
A. Limit Cycles Caused by Friction where e = x - X d is the position error and X d is the desired
It has been observed experimentally that friction may give reference which is assumed to be twice [Link] term
rise to limit cycles in servo drives where the controller has ke in the observer is a correction term from the position error.
integral action; for references see [2]. This phenomenon is The closed-loop system is represented by the block diagram
often referred to as hunting. in Fig. 9. With the observer based friction compensation, we
Consider the linear motion of a mass m at position x. The achieve position tracking as shown in the following theorem.
equation of motion is Theorem 1 : Consider system (10) together with the friction
model (1) and (3), friction observer (12) and (13), and control
d2x
m-=u-~ law (14). If H ( s ) is chosen such that
dt2
where d x / d t = U is the velocity, F the friction force given by
(3) and U the control force which is given by the PID controller
is strictly positive real (SPR) then the observer error, F - F,
and the position error, e, will asymptotically go to zero.
Proof: The control law yields the following equations
A block diagram of the system is shown in Fig. 7. In Fig. 8 1 - OlS+(TO
we show the results of a simulation of the system. The friction e= (-F) = (-5) = -G(s)i
parameters are given by Table I, m = 1 and the controller
+
ms2 H ( s ) +
ms2 H ( s )
parameters are K , = 6, K p = 3, and Ki = 4. The reference
position is chosen as X d = 1. The model clearly predicts limit
cycles as have been observed experimentally in systems of where F = F - F and z" = z - 1. Now introduce
this type.
52
V=<TP<+-
B. Friction Compensation k
as a Lyapunov function and
Only linear feedback from the position was used in the PID
control law (1 1). Knowledge about friction was not used. It "
- = A< + B(-5)
is of course more appealing to make a model-based control dt
that uses the model to predict the friction to compensate for e = C<

-r I 1'I 1

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 424 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 40, NO. 3, MARCH 1995

i
- -W) -
e
is strictly positive real, then the observer error, F - F , and
the velocity error will asymptotically go to zero.
Proofi The theorem is proven in the same way as Theo-
rem 1 after observing that the control law yields the following
error equations
1 -
(-F) =
OlS+(TO
(-5) = - G ( s ) ~
Fig. 10. The block diagram in Fig. 9 redrawn with e and Z as outputs of a
linear and a nonlinear block, respectively.
w - v d =
+
ms H(s) ms H(s) +
which is a state-space representation of G(s). Since G(s) is
SPR [17] it follows from the Kalman-Yakubovitch Lemma This is again an interconnection of a dissipative system with 2
[17] that there exist matrices P = PT > 0 and Q = QT > 0 as its output and a linear SPR system with w - V d as its output.
such that To assume that the friction model and its parameters are
known exactly is of course a strong assumption. Investigation
+
A ~ PP A = -Q of the sensitivity of the results to these assumptions is an
P B = CT. interesting problem that is outside the scope of this paper. The
accuracy required in the velocity measurement is a similar
Now
problem.

VI. CONCLUSIONS
A new dynamic model for friction has been presented. The
model is simple yet captures most friction phenomena that
are of interest for feedback control. The low velocity friction
characteristics are particularly important for high performance
I -IT&<. pointing and tracking. The model can describe arbitrary steady-
state friction characteristics. It supports hysteretic behavior due
The radial unboundedness of V together with the semi- to frictional lag, spring-like behavior in stiction and gives a
definiteness of dV/dt implies that the states are bounded. We varying break-away force depending on the rate of change of
<
can now apply LaSalle’s theorem to see that 4 0 and 2 + 0 the applied force. All these phenomena are unified into a first-
which means that both e and F tends to zero and the theorem order nonlinear differential equation. The model can readily
is proven. be used in simulations of systems with friction.
The theorem can also be understood from the following Some relevant properties of the model have been investi-
observations. By introducing the observer we get a dissipative gated. The model was used to simulate position control of a
map from e to 2 and by adding the friction estimate to servo with a PID controller. The simulations predict hunting
the control signal, the position error will be the output of as has been observed in applications of position control with
a linear system operating on 5. This means that we have integral action. The model has also been used to construct
an interconnection of a dissipative system and a linear SPR a friction observer and to perform friction compensation for
system as seen in Fig. 10. Such a system is known to be position and velocity tracking. When the parameters are known
asymptotically stable. the observer error and the control error will asymptotically go
Velocity Control: The same type of observer-based control to zero. Sensitivity studies, parameter estimation and adapta-
can be used for velocity control. For this control problem the tion are natural extensions of this work.
controller is changed to
.* dud REFERENCES
~ = - H ( s ) e + F + m -dt
[ 11 B. Armstrong-Htlouvry, Control of Machines with Friction. Boston,
MA: Kluwer, 1991.
[2] B. Armstrong-Htlouvry, P. Dupont, and C. Canudas de Wit, “A survey
of models, analysis tools and compensation methods for the control of
machines with friction,” Automatica, vol. 30, no. 7, pp. 1083-1138,
1994.
[3] P. Dahl, “A solid friction model,” Aerospace Corp., El Segundo, CA,
where w - V d is the velocity error and V d the desired velocity Tech. Rep. TOR-0158(3 107-1 8)- 1, 1968.
[4] P.-A. Bliman, “Mathematical study of the Dahl’s friction model,”
which is assumed to be differentiable. Velocity tracking is European J . Mechanics. AISolids, vol. 11, no. 6, pp. 835-848, 1992.
achieved as shown in the following theorem. [5] P.-A. Bliman and M. Sorine, “Friction modelling by hysteresisoperators.
Application to Dahl, Sticktion, and Stribeck effects,” in Proc. Con6
Theorem 2: Consider system (10) together with friction Models of Hysteresis, Trento, Italy, 1991.
model (1) and (3) and observer based control law (15). If [6] C. Walrath, “Adaptive bearing friction compensation based on re-
H ( s ) is chosen such that cent knowledge of dynamic friction,” Automatica, vol. 20, no. 6, pp.
717-727, 1984.
G(s) =
O1S +00 [7] N. Ehrich Leonard and P. Krishnaprasad, “Adaptive friction compen-
ms +H(s) sation for bi-directional low-velocity position tracking,” in Proc. 31st
Conf Decis Contr, 1992, pp. 267-273.

-r n -1- I-

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
CANUDAS DE WIT ef al.: CONTROL OF SYSTEMS WITH FRICTION 425

[8] J. R. Rice and A.L. Ruina, “Stability of steady frictional slipping,” J . Henrik Olsson (S’91) received the [Link]. degree
Applied Mechanics, vol. SO, no. 2, 1983. in electrical engineering from Lund Institute of
191 P. Dupont, “Avoiding stick-slip through PD control,” lEEE Trans. Technology, Lund, Sweden in 1989.
Automat. Contr., vol. 39, pp. 1094-1097, 1994. He spent the academic year 1989-1990 at the
[IO] D. Haessig and B. Friedland, “On the modeling and simulation of Department of Electrical Engineering at University
friction,” in Proc. I990 Amer. Conrr. Conf., San Diego, CA, 1990, pp. of Califomia, Santa Barbara. Since 1990 he has
1256-1261, been with the Department of Automatic Control at
[I 11 P. Dahl, “Solid friction damping of spacecraft oscillations,” in PI-OC. Lund Institute of Technology where he is currently
AIAA Guidance and Conrr. C o n f , Boston, MA, Paper 75-1 104, 1975. completing the Ph.D. degree. His main research
[ 121 J. Willems, “Dissipative dynamical systems part I: General theory,” interest is in control of nonlinear servosystems.
Arch. Rational Mech. Anal., vol. 45, pp. 321-51, 1972.
[I31 J. Courtney-Pratt and E. Eisner, “The effect of a tangential force on the
contact of metallic bodies,” in Proc. Royal Society, vol. A238, 1957,
pp. 529-550.
[ 141 D. P. Hess and A. Soom, “Friction at a lubricated line contact operating
at ocillating sliding velocities,” J . Tribology, vol. 112, pp. 147-152, Karl Johan Astrom (M’71-SM’77-F’79) receievd
1990. the Ph D. degree in automatic control and mdthemat-
[ 151 V. I. Johannes, M. A. Green, and C. A. Brockley, “The role of the rate ics from the Royal Institue of Technology (KTH),
of application of the tangential force in determining the static friction Stockholm, in 1960
coefficient,” Wear, vol. 24, no. 381-385, 1973. He has been Professor of Automatic Control at
[I61 R. S. H. Richardson and H. Nolle, “Surface friction under time- Lund Institute of Technology/Lund University since
dependent loads,” Wear, vol. 37, no. 1, pp. 87-101, 1976. 1965 His research interests include broad aspect\ of
1171 H. K. Khalil, Nonlinear Systems. New York: Macmillan, 1992. automatic control, stochastic control, system iden-
tification, adaptive control, computer control, and
computer-aided control engineering.
Dr. Astrom has published five books and many
papers. He is a member of the Royal Swedish Academy ot Sciences, and
the Royal Swedish Academy of Engineering Sciences (IVA) He has received
many awards among them the Quazza medal from IFAC in 1987 and the
IEEE Medal of Honor in 1993

Carlos Canudas d e Wit (A-93) received the B Sc

degree in electronics and communications from the
Technologic of Monterrey, Mexico in 1980. He re- Pablo Lischinsky was bom in Montevideo,
ceived the [Link] and the Ph D. degrees in automatic Uruguay, on September 24, 1960. He received
control from the Polytechnic of Grenoble, France, in the B.S. degree and the M.S. degree in control
1984 and 1987, respectively engineering from the Escuela de Ingenieria de
From 1981 to 1982 he worked as a Research En- Sistemas, Universidad de Los Andes, Mtrida,
gineer at the Department of Electrical Engineering Venezuela, in 1985 and 1990, respectively. He
at the CINVESTAV-IPN in Mexico City He was received the M.S. degree in automatic control in
a Visiting Researcher in 1985 at Lund Institute of 1993 from the Insitut Nationale Polytechnique
Technology, Sweden Since 1987 he has been an (INPG-ENSIEG), Laboratoire d’Automatique,
Associate Professor in the Department of Automatic Control, Polytechnic of Grenoble, France. Since 1990, he has been with the
Grenoble, where he teache3 and conducts research in the area of adaptive Department of Automatic Control at the Universidad
and robot control. He wrote Adaptrve Control of Par-trally Known Systemr de Los Andes. Currently he is on leave, working on his Ph.D. dissertation
Theor v and Applicatrons (Elsevier, 1988). He is an editor of Advanced Robot at the Insitut Nationale Polytechnique (INPG-ENSIEG). His current research
Control (Springer-Verlag) and also an associate editor of IEEE TRANSACTIONSinterests are in adaptivc2 control, identification, and computer control of
ON AUTOMATIC CONTROL mechanical systems.

Authorized licensed use limited to: TU Delft Library. Downloaded on June 02,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.