CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol.

XIII - Passivity Based Control - Antonio Loría and Henk

Nijmeijer

PASSIVITY BASED CONTROL

Antonio Loría
CNRS, LSS-Supélec, Plateau de Moulon, 91192, Gif sur Yvette, France

Henk Nijmeijer
Departments of Mech. Engg., Eindhoven Univ. of Technology, The Netherlands

Keywords: Passivity, automatic control, mechanical systems, robot manipulators,

Euler-Lagrange systems, nonlinear systems, input-output stability.

Contents

1. Introduction
2. Passivity: mathematically speaking

S
TE S
2.1. In a General Input-Output Framework

R
AP LS
3. Stability of passive systems
3.1. L2 -Stability
3.2. From L2 -Stability to Lyapunov Stability
C EO
4. PBC of Euler-Lagrange systems
4.1 Passivity of EL Systems
4.2. PBC of EL Systems
4.2.1. An Introductory Example
E –

4.2.2 Lyapunov Stability of the ES+DI Controllers

H

4.3. EL Controllers
PL O

4.4. Tracking a Time-varying Reference

5. Epilogue
M SC

Glossary
Bibliography
Biographical Sketches
SA NE

Summary

Passivity is a fundamental property of many physical systems which may be roughly

U

defined in terms of energy dissipation and transformation. It is an inherent Input-Output

property in the sense that it quantifies and qualifies the energy balance of a system when
stimulated by external inputs to generate some output. Passivity is therefore related to
the property of stability in an input-output sense, that is, we say that the system is stable
if bounded “input energy” supplied to the system, yields bounded output energy. This is
in contrast to Lyapunov stability which concerns the internal stability of a system, that
is, how “far” the state of a system is from a desired value. In other words, how
differently a system behaves with respect to a desired performance.

Passivity based control is a methodology which consists in controlling a system with the
aim at making the closed loop system, passive. The field constitutes an active research
direction and therefore in this chapter we give only a basic overlook of the most
important concepts involved. A section is also devoted to a wide class of physical

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

passive systems: the Euler-Lagrange (EL) systems and their passivity-based control.

The reader should rather consider this presentation as very concise image of the material
cited in the Bibliography. Therefore, we invite the reader who wishes to obtain a deeper
knowledge in the subject, to see those references.

1. Introduction

To better understand the passivity concept and passivity-based control (PBC), we need
to leave behind the notion of state of a system and think of the latter as a device which
interacts with its environment by transforming inputs into outputs. From an energetic
viewpoint we can define a passive system as a system which cannot store more energy
than is supplied by some “source”, with the difference between stored energy and
supplied energy, being the dissipated energy.

S
TE S
Hence, it shall be clear that passivity is closely related to the stability of a system, in the

R
AP LS
input-output sense evoked in the Summary. In PBC achieving stability from this
viewpoint is the first goal.
C EO
A fundamental property of passive systems is that, regarding a feedback interconnection
of (other physical) passive systems, passivity is invariant under negative feedback
interconnection. In other words, the feedback interconnection of two passive systems
yields a passive system.
E –

Thus, if the overall energy balance is positive, in the sense that the energy generated by
H

one subsystem, is dissipated by the other one, the closed loop will be stable in an input-
PL O

output sense (see Proposition 2). This property constitutes the basis of passivity-based
control (PBC).
M SC

The term PBC was coined in 1989 in the context of adaptive control of robot
manipulators to define a controller methodology whose aim is to render the closed-loop
SA NE

system passive, seen as a map from an external new input. This objective seemed very
natural within that context, since the robot dynamics defines a passive map from input
torques to output link velocities. As a matter of fact this passivity property is inherent to
U

many other physical systems such as electrical and electromechanical. (See section 4).

Since the aim in PBC is to render the closed loop system passive, the main property
used in PBC is the fact that the interconnection of passive systems is passive.
Conversely, passive systems can be decomposed in passive “subsystems”. Thus, in this
philosophy the controller may be designed as a passive system.

In terms of energy dissipation, the PBC approach may be viewed as an extension of the
so-called energy-shaping plus damping injection technique introduced to solve state-
feedback set (operating) point regulation problems in fully actuated robotic systems
back in 1981. For this particular problem we can concentrate our attention on the
potential energy and the dissipation functions to proceed along two basic stages: firstly,
as energy shaping stage which consists on modifying the potential energy of the system
in such a way that the “new” potential energy function has a global and unique

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

minimum at the desired equilibrium. This is motivated by the well known fact (stated by
Joseph Lagrange in 1788 and proved 50 years later by Dirichlet) that the stable
equilibria of mechanical systems correspond to the minima of the potential energy
function. Secondly, a damping injection stage which consists in modifying the
dissipation properties of the system, to render it strictly passive.

Viewed from the PBC perspective the energy shaping stage accomplishes the objective
of rendering the closed loop system passive with a desired storage function that consists
of the original kinetic energy and the new desired potential energy. The damping
injection reinforces this property to output strict passivity. Finally, Lyapunov
asymptotic stability follows from the input-output stability of the output strictly passive
map provided some dissipation propagation (i.e., detectability) conditions are met. That
is, the system evolves in a way that it reaches the desired set point asymptotically. See
Sections 2.1 and 3.2.

S
TE S
The generality of the PBC allows us to deal with different problems such as output

R
AP LS
feedback and tracking control in a unified way. Moreover, even though here we will
only illustrate the PBC methodology with simple examples of control of EL mechanical
systems, the reader must keep present that, having its origins in electrical circuits, it is
C EO
natural that PBC is most suitable for electrical and electromechanical systems such as
power converters, electrical machines, etc. This will be illustrated through a time-
varying reference tracking control problem in Section 4.4.

2. Passivity: Mathematically Speaking

E –
H

In this section we will introduce the precise definitions of passivity and some important
PL O

theorems on passivity.
M SC

2.1. In a General Input-Output Framework

As it may be clear from the discussion above, when talking about a passive system
SA NE

(operator) one aims at measuring the energy (storage) transformation performed in the
system. The concept of passivity in dynamical systems has its roots in the same concept
used by electrical engineers to characterize elements which consume energy but do not
U

supply it. In this context, the input and output signals have a direct physical meaning,
i.e., current and voltage hence, the (electrical) energy measure of these is evident: it
simply corresponds to the integral of the power over time. However, if we would like to
talk about the passivity property of physical systems of different nature (electrical,
mechanical, chemical, etc), we need a more general concept of measure.

To that end, we must keep in mind that passivity is a property of the system, seen as an
operator which maps inputs into outputs. In this respect, we will find characterizations
and sufficient conditions for passivity, which apply to systems that can be modeled by
rational transfer functions as well as to systems modeled by nonlinear (possibly time-
varying) differential equations.

In this section we will introduce the precise definitions of passivity, which reflect the
fact that passivity is an energy transformation property. We will also extend to the case

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

of nonlinear systems, some of the arguments made before, to sustain the fundamental
properties of passive systems.

Definition 1 ( L2 and L2 norms): The L2 norm of a signal f : \ ≥ 0 → \ , f (t ) is

n n n n n

denoted f (t ) 2T and defined by

(∫ )
1
T 2 2
0
f (t ) dt (1)

S
TE S
and the L 2 norm denoted f (t ) is defined as by lim f (t ) 2T .
2 T →∞

R
AP LS
With this metric we can then define the normed L2e -space:
C EO
Definition 2 ( L2e -space): We say that f : \ ≥ 0 → \ belongs to L2e if and only if
f (t ) 2T
<∞
E –

The definitions above makes sense from a practical viewpoint if we consider the case
H

n
when f (t ) corresponds to power and therefore, the L2 borrows the interpretation of
PL O

energy amount over a time interval.

M SC

n
Now in order to properly define the passivity concept for L2 signals we introduce the
following product, which generalizes the concept of supplied energy discussed above.
SA NE

Definition 3 (Inner product): Let u , y ∈ L2 and T > 0 , then the inner product is
n

defined ∀T > 0 by
U

T
u| y T
:= ∫0 u (t ) y (t )dt . (2)

For illustration we may consider u to be an input voltage supplied to an electrical load

in which case i may be interpreted as the output current. Therefore, inner product gives
a measure on the energy stored in the resistor.

With these tools we can now properly define the following

Definition 4 (Passivity): An operator H : u 6 y is passive if there exists a β ∈ \ such

that

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

u| y T
≥ β. (3)

The number β depends on the initial conditions of the signals. Often, it quantifies the
initial energy stored in the system. This will become clearer when dealing with passivity
of mechanical systems, in section 4.1.

Definition 5 (Output Strict Passivity): An operator H : u 6 y is output strictly

passive if there exists β ∈ \ and δ o > 0 such that

2
u| y T
≥ δo y 2T
+ β. (4)

Definition 6 (Input Strict Passivity): An operator H : u 6 y is input strictly passive

S
TE S
if there exists β ∈ \ and δ i > 0 such that

R
u| y T
AP LS
≥ δi u 2T +
2
β. (5)
C EO
Passive interconnected systems

The following theorems formalize the fact that passivity is sustained for the
interconnection of passive systems.
E –
H

Theorem 1: Consider the input-output system depicted in Figure 1. Let

PL O

e := (e1 , e2 ), u := (u1 , u2 ) and y := ( y1 , y2 ) be in L22en . If ∑1 and ∑ 2 are both passive

M SC

then ∑ : u 6 y is also passive. If ∑1 and ∑ 2 are OSP then ∑ : u 6 y is also OSP.

SA NE
U

Figure 1: Feedback interconnection of passive systems

The theorem below regards a special case of the feedback interconnection depicted in
Figure 1, when u2 ≡ 0 . This structure is particularly important since it is the typical
case of a plant ( ∑1 ) in closed loop with a controller ( ∑ 2 ) . In this case the input

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

u1 plays the role of an external signal to the closed loop. Notice that this input can be in
its turn the output of another passive block. In this way one can build a new passive
system upon a core passive block. Therefore these theorems are fundamental to
passivity-based control.

Theorem 2: Consider the closed loop system of Figure 1, with u2 ≡ 0 .Assume that
∑i : L2ne 6 L2ne , i = 1,2 . Then e2 = y1 ∈ L2ne if either of the following statements is
true:

• If ∑1 : e1 6 y1 is passive and ∑ 2 : y1 6 y2 is ISP or

• If ∑1 : e1 6 y1 is OSP and ∑ 2 : y1 6 y2 is passive.

S
So far we have stated formally and in fair generality, under which conditions a feedback

TE S
interconnection of passive systems yields a passive system. However, it will be also

R
AP LS
useful to know that interconnections not only preserve the passivity properties of the
subsystems but, in certain cases, passivity can be strengthened. To illustrate this idea,
we briefly discuss next, a technique called loop transformation.
C EO
Consider the interconnected system of Figure 1 with only one input, i.e., let u2 ≡ 0 .
Assume that ∑ 2 is ISP and ∑1 is passive. The loop transformation technique will make
evident that since the system ∑ 2 is “more dissipative” than ∑1 (some readers will know
E –
H

that the term “dissipative” and “passive” are mathematically different.

PL O

With an abuse of notation we use here the term dissipative to denote a system which
M SC

dissipates energy in a non recoverable manner, e.g., heat.), by performing the

interconnection ‘some’ of the “dissipation of ∑ 2 is propagated to ∑1 ”.
SA NE

To show this we will use Figure 2, which represents a system equivalent to that of
Figure1 with u2 ≡ 0 , and the following

Fact 1: Assume that the system ∑ 2 is ISP and has finite L2 gain, i.e., there exists
U

∞ > c > 0 such that y2 2T ≤ c e1 2T . Then, the map ∑ 2 is also OSP.

Let us perform a few simple calculations to exhibit the new passivity properties of the
feedback interconnected system of Figure 2. For ∑1′ : e1 6 y1 , using the passivity
property of ∑1 , we have that

e1 | y1 2T
= u1 − y2 + ky1 | y1 = u1 − y2 | y1 + ky1 | y1 2T
2T 2T
≥ β1 + k y1 2T
.

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

S
TE S
R
AP LS
Figure 2: Loop transformed feedback interconnected system

That is, the loop transformation has rendered the map ∑1′ , OSP. The price paid for this
C EO
is that the ISP of ∑ 2 has been “weakened” more precisely,

y1 | y2 − ky1 2T
= y1 | y2 2T
− ky1 | y1 2T
≥ β2 + δi 2 y1 − k y1
E –
H
2T 2T
PL O

hence, we must impose k < δ i 2 .

M SC

It is important to remark at this point that the coefficient k is used only for analysis
hence, there is no loss of generality in restricting it to be k < δ i 2 . Notice also that the
SA NE

physical system has not changed with the loop transformation but only the way we look
at it!

Using the Fact 1 we obtain that the system of Figure 1 with ∑1 passive and ∑ 2 ISP and
U

finite L2 gain is equivalent to the interconnection of an OSP with an OSP and ISP
system.

This observation is sometimes fundamental in the stability analysis of passive systems,

and consequently in passivity –based control, as we will see in section 3.2 and 4.2.

3. Stability of Passive Systems

Before discussing PBC we need to discuss about stability. In particular, the type of
stability which one pursues in PBC, is, in an input-output sense.

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

3.1. L2 -Stability

A “relaxed” definition of L2 stability is that an operator is L2 stable if it maps L2 inputs

into L2 outputs. However, in a more strict sense one may also be interested in
measuring the “amount of stability”. This is given by the so-called L2 gain.

Definition 7 ( L2 -stability): The state space system ∑ is said to be L2 stable with finite
L2 gain if there exists a positive constant γ such that for every initial condition
x0 = x(0) , there exists a finite constant β( x0 ) such that

y (t ) 2T
≤ γ u (t ) 2T
+ β( x0 ) .

S
TE S
Proposition 1: If ∑ : u 6 y is OSP then it has finite L2 -gain.

R
AP LS
Proof. The proof follows straight forward observing that OSP implies the existence of
2
C EO
2
δ > 0 and β ∈ \ such that δ o y ≤ u| y − β + 12 1 u − δo y , therefore
2T T δo
2T
δo 2 2
2
y 2T
≤ 1
2δ o
u 2T
− β . Thus the L2 gain γ ≤ δ1 .
o
E –
H

-
PL O

-
-
M SC

TO ACCESS ALL THE 24 PAGES OF THIS CHAPTER,

SA NE

Click here
U

Bibliography

V. Arnold. (1989) Mathematical Methods of Classical Mechanics. New York :Springer-Verlag .

H. Berghuis and H. Nijmeijer. (1993) A passivity approach to controller-observer design for robots. IEEE
Trans. On Robotics Automat., 9-6:740-755.
C.Byrnes, A. Isidori, and J.C. Willems. (1991) Passivity, feedback equivalence, and the global
stabilization of minimum phase nonlinear systems. IEEE Trans. on Automat. Contr., AC-36(11):1228-
1240.[Section 3 is based on this work]
C.A. Desoer and M. Vidyasagar. (1975) Feedback Systems: Input-Output Properties. 266p. New York
:Academic Press, .[Input-Output stability theory is thoroughly studied. The material of Section 2 is based
on this work]
H. Goldstein. (1974) Classical Mechanics. 672p. Addison Wesley.[An excellent text on mechanics]
D. Hill and P. Moylan. (1976)The stability of nonlinear dissipative systems. IEEE Trans. on Automat.
Contr., Pages 708-711.[Fundamental results on interconnection of passive systems]

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

D. E. Koditschek. Natural motion of robot arms. In Proc. 23rd. IEEE Conf. Decision Contr., Las Vegas,
NV., 1984.[Passivity based control of EL systems and in particular mechanical systems.]
A Loría R. Kelly, R. Ortega, and B. Santibñez. (1996) On output feedback control of Euler-Lagrange
systems under input constraints. IEEE Trans. Automat. Contr. , 42(8):1138-1142.[ Passivity based control
of EL systems and in particular mechanical systems]
A. Loría and H. Nijmeijer. (1998) Nonlinear control systems: (Output feedback) design methods. In John
G. Webster, editor, Wiley encyclopedia of Electrical and Electronic Engineering. John Wiley & Sons IC.
. Article #1025 in Section “Automatic Control”[ Passivity based control of EL systems and in particular
mechanical systems]
A. Loría and H. Nijmeijer (1998) : Output feedback tracking control of Euler-Lagrange systems via
bounded controls. Syst. & Contr. Letters, 33(3):151-163.[ Passivity based control of EL systems and in
particular mechanical systems]
R. Lozano, B. Brogliato, O. Egeland, and B. Maschke.(2000) Dissipative systems analysis and control.
Communications and control Engineering. London : Springer Verlag. [Passivity based control of
nonlinear systems].

S
TE S
A. M. Lyapunov. Problème de la stabilitè de mouvement. (1893) Annales de la facultè de sciences de
Toulouse, 9:203-474, 1907. (Translation from the original published in Comm. Soc. Math. Kharkov 1893,

R
AP LS
reprinted in Ann. Math. Studies 17, Princeton 1949). See also “Stability of motion”, Academic Press: NY
1996..[Liapunov stability was launched]
I. M. Y. Mareels and D. Hill. Monotone stability of nonlinear feedback systems. J. Math. Systems’
C EO
Estimation and Control, 2:275-291, 1992.[ Fundamental results on interconnection of passive systems]
R. Ortega, A. Loría R. Kelly, and L. Praly.(1995) On passivity-based output feedback global stabilization
of Euler-Lagrange systems. Int. J. Robust and Nonlinear Control, Special issue on Control of nonlinear
mechanical systems, 5(4):313-325. (H. Nijmeijer and A. J. van der Schaft eds.).[ Passivity based control
of EL systems and in particular mechanical systems]
E –
H

R. Ortega, A. Loría P.J. Nicklasson, and H. Sira-Ramírez. (1998) Passivity-based Control of Euler-
Lagrange Systems: Mechanical, Electrical and Electromechanical Applications. Communications and
PL O

control Engineering. London : Springer Verlag . ISBN 1-85233-016-3.[ Section 4 is based fully on this
work. ]
M SC

R. Ortega and M. Spong. (1989) Adaptive motion control of rigid robots: A tutorial. Automatica, 25-
6:877-888.[This tutorial introduces the term PBC]
R. Sepulchre, M. Janković, and P. Kokotović. (1997) Constructive nonlinear control. Springer
SA NE

Verlag.[Passivity based control of nonlinear systems]

J. J. Slotine and W. Li. (1988) Adaptive manipulator control: a case study. IEEE Trans. on Automat.
Contr., AC-33:995-1003.[ Passivity based control of EL systems and in particular mechanical systems]
U

M. Takegaki and S. Arimoto. (1981) A new feedback method for dynamic control of manipulators. ASME
J. Dyn. Syst. Meas. Contr., 103:119-125.[The so-called energy-shaping plus damping injection technique
was introduced in the context of robot control]

A. J. van der Schaft.(1999) L2 -Gain and passivity techniques in nonlinear control. Communication and
control Engineering. Heidelberg : Springer Verlag, 2nd edition.[The material of Sections 2 and 3 is based
on this work. Theorems 1 and 2 are from this work.]
J. C. Willems. Dissipative dynamical systems. Part I (1972): General theory. Arch. Rat. Mech . and
Analysis, 45(5):321-351.[The theory of dissipative systems is introduced]
J. C. Willems. Dissipative dynamical systems. Part II (1972): Linear systems with quadratic supply rates.
Arch. Rat. Mech. and Analysis, 45(5):352-393.[ The theory of dissipative systems is introduced]

Biographical Sketches

Antonio Loria was born in Mexico in 1969. He got the BSc degree in Electronic Engineering from the

©Encyclopedia of Life Support Systems (EOLSS)

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Passivity Based Control - Antonio Loría and Henk
Nijmeijer

ITESM, Monterrey, Mexico in 1991. He got the MSc and PhD degrees in Control Engg. from the UTC,
France in 1993 and Nov. 1996 respectively. From December 1996 thru Dec. 1998, he was successively an
associate researcher at Univ. of Twente, The Netherlands; NTNU, Norway and the CCEC of the Univ. of
California at Sta Barbara, USA. Dr. Loria is currently ``Charge de Recherche'', at the the French National
Centre of Scientific Research (CNRS). He is (co)author of more than 75 scientific articles and the book
``Passivity based control of Euler-Lagrange systems'' Springer Verlag, 1998. Antonio Loria is associate
editor of Systems and Control Letters. His research interests include: modeling and control of Euler-
Lagrange systems, stability analysis of nonlinear time-varying systems, biped locomotion, and output
feedback stabilization. Detailed information and publications are available at:
http://public.lss.supelec.fr/perso/loria.

Henk Nijmeijer (1955) obtained his MSc-degree and PhD-degree in Mathematics from the University of
Groningen, Groningen, the Netherlands, in 1979 and 1983, respectively. From 1983 until 2000 he was
affiliated with the Department of Applied Mathematics of the University of Twente, Enschede, the
Netherlands. Since, 1997 he was also part-time affiliated with the Department of Mechanical Engineering
of the Eindhoven University of Technology, Eindhoven, the Netherlands. Since 2000, he is full-time
working in Eindhoven, and chairs the Dynamics and Control section. He has published a large number of
journal and conference papers, and several books, including the 'classical' Nonlinear Dynamical Control

S
TE S
Systems (Springer Verlag, 1990, co-author A.J.van der Schaft). Henk Nijmeijer is editor in chief of the
Journal of Applied Mathematics, corresponding editor of the SIAM Journal on Control and Optimization,

R
AP LS
and board member of the International Journal of Control, Automatica, European Journal of Control,
Journal of Dynamical Control Systems, SACTA, International Journal of Robust and Nonlinear Control,
and the Journal of Applied Mathematics and Computer Science. He is a fellow of the IEEE and was
C EO
awarded in 1987 the IEE Heaviside premium.
E –
H
PL O
M SC
SA NE
U

©Encyclopedia of Life Support Systems (EOLSS)