Some Fundamentals of Stability

Theory
Aaron Greenfield
Outline
I. Introduction + Motivation
II. Definitions
III. Theorems
IV. Techniques for Lyapunov Function
Construction

Basic Notion of Stability
Stability
An important property of dynamic systems
Stability. . .
An insensitivity to small perturbations
Perturbations are modeling errors of system, environment, noise
F=0. OK
Basic Notions of Stability
Stability
An important property of dynamic systems
Stability. . .
An insensitivity to small perturbations
Perturbations are modeling errors of system, environment
F=0. OK
Basic Notions of Stability
Stability
An important property of dynamic systems
Stability. . .
An insensitivity to small perturbations
Perturbations are modeling errors of system, environment
F=0. OK
Stability
An important property of dynamic systems
Stability. . .
An insensitivity to small perturbations
Perturbations are modeling errors of system, environment, unmodeled
noise
F=0. OK
Basic Notions of Stability
Stability
Why might someone in robotics study stability?

Basic Notions of Stability
(1) To ensure acceptable performance of the robot under perturbation
g q C q M + + =

t
Configuration space trajectory
with constraints
Some Notation

|| ||
) , ; (
0 0
x
t x t p
x
e
An isolated equilibrium of an ODE
A solution curve to first-order ODE system
with initial conditions listed
Standard Euclidean Vector Norm
) (x f x =

MANY definitions for related stability concepts

) , ( t x f x =

Definitions
) , , ( t u x f x =

Reduces to above under action of a control

Restrict attention to following classes of differential equations
Autonomous ODE
Non-Autonomous ODE
Stabilizability Question
Definitions Summary Slide
Attractivity
Lyapunov Stability
) (x f x =
0 =
e
x
With Isolated equilibrium at
Defn1.1: Stability of autonomous ODE, isolated equilibrium
Hahn 1967
Slotine, Li
The equilibrium point (or motion) is called stable in the sense
of Lyapunov if:

For all there exists such that

0 > c
0 > o
0 0 0
|| ) , ; ( || t t t x t p > <c
whenever
) ( || ||
0
c o < x
) , ; (
0 0
t x t p
) (x f x =
0 =
e
x
With Isolated equilibrium at
Defn1.1: Stability of autonomous ODE, isolated equilibrium
The equilibrium point (or motion) is called stable in the sense
of Lyapunov if:

For all there exists such that

0 > c
0 > o
0 0 0
|| ) , ; ( || t t t x t p > <c
whenever
) ( || ||
0
c o < x
Notes
(1) If
1 1
c c c >
(2) There can be a but no
min
o
max
o
Lyapunov Stability
(Local Concept)
(Unbounded Solutions)
Lagrange Stability
) (x f x =
0 =
e
x
With Isolated equilibrium at
Defn1.1: Stability of autonomous ODE, isolated equilibrium
The equilibrium point (or motion) is called stable in the sense
of Lyapunov if:

For all there exists such that

0 > c
0 > o
0 0 0
) ( || ) , ; ( || t t t x t p > < o c
whenever
o || ||
0
< x
Lagrange Stability
) (x f x =
0 =
e
x
With Isolated equilibrium at
Defn1.1: Stability of autonomous ODE, isolated equilibrium
The equilibrium point (or motion) is called stable in the sense
of Lyapunov if:

For all there exists such that

0 > c
0 > o
0 0 0
) ( || ) , ; ( || t t t x t p > < o c
whenever
o || ||
0
< x
Lagrange Stable
Lagrange Stability
) (x f x =
0 =
e
x
With Isolated equilibrium at
Defn1.1: Stability of autonomous ODE, isolated equilibrium
The system is Lagrange stable if:


For all there exists such that

0 > c
0 > o
0 0 0
) ( || ) , ; ( || t t t x t p > < o c
whenever
o || ||
0
< x
Notes
(1) Bounded Solutions
(2) Independent Concept
a) Lyapunov, Lagrange
b) Not Lyapunov, Lagrange
c) Lyapunov, Not Lagrange
d) Not Lyapunov, Not Lagrange

(2) Stability completely separate concept
a) Stable, Attractive
b) Unstable, Unattractive
c) Stable, Unattractive
d) Unstable, Attractive

Attractive
) (x f x = 0 =
e
x
With Isolated equilibrium at
Defn 1.2: Attractivity of autonomous ODE,isolated equilibrium
The equilibrium point (or motion) is called attractive if:

There exists an such that

0 > q
e
t
x t x t p =

) , ; ( lim
0 0
whenever q || :||
0 0
< x x
(3) Unstable yet attractive, Vinograd
Notes
(1) Asymptotic concept, no transient notion
Attractivity Example
) (x f x = 0 =
e
x
With Isolated equilibrium at
Defn 1.2: Attractivity of autonomous ODE,isolated equilibrium
0 ; 0 if 0 x
) ) ( 1 )( (
) (
2 2 2 2 2
5 2
= = =
+ + +
+
=
A
y x
y x y x
y x y x
x
0 ; 0 if 0 y
) ) ( 1 )( (
) 2 (
2 2 2 2 2
2
= = =
+ + +

=
A
y x
y x y x
x y y
y
Denominator always positive
y
Switches on x y 2 =
Asymptotic Stability
) (x f x =
0 =
e
x With Isolated equilibrium at
Defn 1.3: Asymptotic stability of autonomous ODE,
isolated equilibrium
q
Asymptotically stable equals both stable and attractive
Defn 1.4: Global Asymptotic stability of autonomous ODE,
isolated equilibrium
Global Asymptotic Stability is both stable and attractive for
[Hahn]
Set Stability
Defn 1.5: Stability of an invariant set M,
autonomous ODE
Notes
M y y x M x e = ||, || inf ) , (
Now consider stability of objects other than isolated equilibrium point
The set M is called stable in the sense of Lyapunov if:


For all there exists such that

0 > c
0 > o
0
t t > < M) (p(t),
whenever
M) , (x
0
<
(1) Attractivity, Asymptotic Stability
are comparably redefined

Invariant-Not entered or exited
(2) Use on limit cycles, for example
[Hahn]
Motion Stability
Defn 1.6: Stability of a motion (trajectory),
autonomous ODE

The motion is stable if:
) , ; (
0
t t p
0
x
Now consider stability of objects other than isolated equilibrium point

For all there exists such that

0 > c
0 > o
0 0 0
|| ) , ; ( ) , ; ( || t t t t p t t p > < c
1 0
x x
whenever
o || || <
0 1
x - x
Notes
(1) Just redefined distance again
(2) Error Coordinate Transform
[Hahn]
Uniform Stability
) , ( t x f x =
0 =
e
x With Isolated equilibrium at
Defn2.1: Stability of non-autonomous ODE,
isolated equilibrium
The equilibrium point (or motion) is called stable (Lyapunov) if:

For all there exists such that

0 > c
0 > o
0 0 0
|| ) , ; ( || t t t x t p > <c
whenever
) , ( || ||
0 0
t x c o <
Defn 2.2: Uniform Stability of non-autonomous ODE,
isolated equilibrium
) , ( || ||
0 0
t x c o < ) ( || ||
0
c o < x

Definitions
) , ( t x f x =
0 =
e
x With Isolated equilibirum at
Defn2.1: Stability of non-autonomous ODE,
isolated equilibrium
Stable, not uniformly stable system
x t t t x ) 2 cos sin ( =
| | | | ) cos 2 ( exp * ) cos 2 ( exp ) (
0 0 0
t t t t x t x + + =
[Dunbar]
Definitions-Wrap Up Slide
Exponential Stability
Input-Output Stability BIBO-BIBS
Stochastic Stability Notions
Stabilizability, Instability, Total
Autonomous ODE Non-Autonomous ODE
Stability of Equilibrium
Lagrange Stability
Attractivity
Asymptotic Stability
Stability of Set
Stability of Motion
Same
Uniform Stability
Not Covered:
Theorems
Restrict attention again to autonomous, non-autonomous ODE
How do we show a specific system has a stability property?
These theorems typically relate existence of a particular function
(Lyapunov) function to a particular stability property
Theorem: If there exists a Lyapunov function,
then some stability property
MANY theorems exist which can be used to prove some stability property
Lyapunov Functions
Lyapunov Functions
0 ) (
0 0 ) (
>
= =
x V
x x V
0 ) ( s
c
c
= f
x
V
x V

Defn 3.1 Lyapunov function for an autonomous system

) (x f x =

) (x V
Positive Definite around origin
For some neighborhood of origin
Defn 3.2 Lyapunov function for an non-autonomous system
) , ( t x f x =

t x V t x V > ) ( ) , (
2
[Slotine, Li]
[Hahn]
0 ) , ( s
c
c
+
c
c
=
t
V
f
x
V
t x V

Dominates Positive Definite Fn

For some neighborhood of origin
Assume
V is continuous in x,t
is also
Note
V

Stability Theorem
) (x f x =
Thm 1.1: Stability of Isolated Equilibrium of Autonomous ODE
An isolated equilibrium of is stable if there exists a Lyapunov Function
for this system
) (x f x =
Proof Sketch 1.1
For all there exists
0 > c
0 > o
0 0 0
|| ) , ; ( || t t t x t p > <c
whenever
o || ||
0
< x
If
exists 0 ) ( ), ( s x V x V

then
(1) Pick Arbitrary Epsilon, Construct Delta
(2) Consider min of V(x) on Vbound
Extreme Value Theorem
c = || || x
o o s = || :|| ) ( max ) ( x x V f
(3) Define function
(4) If continuous, then by IVT
bound
v f < e ) ( : ) , 0 ( o c o
) (o f
(5) Since
c o < < || ) ( || || ||
0
t x x
0 ) ( s x V

Stability proof example

) (x f x =
Thm 1.1: Stability of Isolated Equilibrium of Autonomous ODE
An isolated equilibrium of is stable if there exists a Lyapunov Function
for this system
) (x f x =
Example- Undamped pendulum
0 sin = + u u

|
|
.
|

\
|

=
|
|
.
|

\
|
u
e
e
u
sin

( )
2
2
1
cos 1 ) , ( e u e u + = V
0 * sin ) , ( = + = e e e u e u

V
(1) Propose
(2) Derivative
(Kinetic + Potential)
Asymptotic stability theorem
Thm 1.2: Asymptotic Stability of Isolated Equilibrium of Autonomous ODE
An isolated equilibrium of is asymptotically stable if there exists
a Lyapunov Function for this system with strictly negative time derivative.
) (x f x =
Small Proof Sketch 1.2
Notes
|| || , ) ( x x V
Radial Unbounded, Barbashin Extension
locally 0 ) ( < x V

Local Global
globally 0 ) ( < x V

(1) Stability from prev, Need Attractivity

(2) EVT with

Ball not entered
) (
) (
x V
x V

t x t p > - || ) ; ( :||
0
o o
(3) Construct a sequence of Epsilon balls
l x V x < ) ( :
l
G
Let there be a region be defined by:
Let on
Let there be two more regions E and M:
l
G E M
V E
_ _

); 0 ( :
1

M is largest invariant
set
Lasalle Theorem
) (x f x =

Thm 1.3: Stability of Invariant Set of Autonomous ODE

(Lasalles Theorem)
Small Proof Sketch 1.3
l
G
0 ) ( s x V

Then M is attractive, that is

0 ) ), ( ( lim =

M t p
t

} as ) ; ( and } { : { ) (
0
- = O
n o n n
t p x t p t p x
(1) Define Positive Limit Set
Properties:
Invariant, Non-Empty, ATTRACTIVE!!
(2) Show
E V x = c O

) 0 ( ) (
1
0

Use: and Limit Cycle Stability 0 ) ( s x V

[Lasalle 1975]
Lasalle Theorem example
Lasalles Theorem Example
Example- Damped pendulum
0 sin = + + u u u

b
|
|
.
|

\
|

=
|
|
.
|

\
|
e u
e
e
u
b sin

( )
2
2
1
cos 1 ) , ( e u e u + = V
(1) Propose
(2) Derivative
e e e u e u

+ = * sin ) , ( V
0
2
s = e b
)} 0 , {( ) 0 (
1
= = =

e u V E

) 0 , 0 ( : = = e u M

Asymptotic Stability of Origin
Uniform Stability Theorem
) , ( t x f x = Theorems for Non-Autonomous ODE
Small Proof Sketch 1.4
Stability and Asymptotic Stability remain the same
0 ) , ( s t x V

Stability
0 ) , ( < t x V

Asymptotic Stability
Thm 1.4: Uniform (Stability) Asymptotic Stability of Non-Autonomous ODE,
Isolated Equilibrium point
The equilibrium is uniformly (Stable) asymptotically stable if there exists
A Lyapunov function with and there exists
a function such that:
) 0 ) , ( ( s t x V

0 ) , ( < t x V

) ( ) , (
1
x V t x V s
||) (|| ) , ( ||) (|| x t x V x | o s s Positive Definite and Decrescent
||) ) ( (|| ) , ( ) ( ) ( t x t x V o o | c o > > >
Decrescent
[Slotine,Li]
||) ) ( (|| ) ( t x o c o >
Barbalets Lemma
) , ( t x f x =

Thm 1.5: Barbalets Lemma as used in Stability

(Used for Non-Autonomous ODE)

If there exists a scalar function such that:
(1)
(2)
(3) is uniformly continuous in time
) , ( t x V
0 s V

) , ( t x V

Then
0 ) , ( lim =

t x V
t

e e e u e u

+ = * sin ) , ( V
0
2
s = e b
Barbalet 0 ) ( lim =

t
t
e
bound lower a has V
[Slotine,Li]
Theorems-Wrap Up Slide
Instability Theorems
Converse Theorems
Stabilizability
Kalman-Yacobovich, other Frequency
theorems

Autonomous ODE Non-Autonomous ODE
Lyapunov implies stability
Lyapunov implies a.s
Lasalles Theorem for sets
Same
Uniform Stability
Barbalets Lemma
Not Covered:
Techniques for Lyapunov Construction
Theorems relate function existence with stability
How then to show a Lyapunov function exists? Construct it
In general, Lyapunov function construction is an art.
Special Cases
Linear Time Invariant Systems
Mechanical Systems
Construction for Linear System
Construction for a Linear System
Ax x =

Px x x V
T
= ) (
x P A PA x
PAx x Ax V x V
T T
T
) (
2 ) (
+ =
= V =

(1) Propose
(2) Time Derivative
}

=
0
dt Qe e P
At t A
T
P is symmetric
P is positive definite
Q P A PA
T
= +
If we choose and solve algebraically for P:
0 s Q
As long as A is stable, a solution is known to exist.
Also an explicit representation of the solution exists:
Construction for a Mechanical System
Construction for a Mechanical System
) ( ) ( q g q C q q M + + =

t
q K q q q M q q q V
p
T T
2
1
) (
2
1
) , ( + =
Kinetic Energy
Potential Energy
(1) Propose
(or similar)
(2) Time Derivative
q K q q q M q q q M q q q V
p
T T T

+ + = ) (
2
1
) ( ) , (
[Sciavicco,Siciliano]
g q K q K
d p
+ + = t
If we use PD-controller with gravity compensation
then
q K q q q V
D
T

= ) , (
Asymptotically stable with Lasalle
General Construction Techniques
Pf f x V
T
= ) (
A quadratic form (ellipsoid) of system velocity
Krasofskii
Variable Gradient
[Slotine, Li]
[Hahn]
f
x
f
P P
x
f
f f P f Pf f x V
T
T T T
) * * ( ) (
c
c
+
c
c
= + =

Solve
0 * * s =
c
c
+
c
c
Q
x
f
P P
x
f
T
}
V =
x
Vdx x V
0
) (
Assume a form for the gradient, i.e

=
= V
n
j
j ij i
x a V
1
Solve for negative semi-definite gradient
Integrate and hope for positive definite V
Construction methods for an Arbitrary System ) (x f x =

Construction Wrap-Up Slide

(1) Linear System -> Explicity Solve Lyapunov Equation
(2) Mechanical System -> Try a variant of mechanical energy
(3) Krasovskiis Method
Variable Gradient
Problem specific trial and error
Conclusion
Motivated why stability is an important
concept
Looked at a variety of definitions of various
forms of stability
Looked at theorems relating Lyapunov
functions to these notions of stability
Looked at some methods to construct
Lyapunov functions for particular problems