5th IFAC Workshop on Lagrangian and Hamiltonian Methods

5th IFAC Workshop on Lagrangian and Hamiltonian Methods

5th
for IFAC Workshop on Lagrangian and Hamiltonian Methods
5th Non
5th
for IFACLinear
IFAC
Non
Control
Workshop
Workshop
Linear on
on Lagrangian
Control Lagrangian and
and Hamiltonian
Hamiltonian Methods
Methods
for
July
for Non
4-7,
Non Linear
2015.
Linear Control
Lyon, France
Control Available online at www.sciencedirect.com
for Non
July 4-7,Linear Control
2015. Lyon, France
July 4-7,
July 4-7, 2015.
2015. Lyon,
Lyon, France
France
July 4-7, 2015. Lyon, France
ScienceDirect
IFAC-PapersOnLine 48-13 (2015) 093–098
Energy
Energy shaping
shaping for
for the
the robust
robust stabilization
stabilization
Energy
Energy shaping
shaping for
for the
the robust
robust stabilization
stabilization
of
of a
a wheeled
wheeled inverted
inverted pendulum
pendulum
of
of a wheeled inverted pendulum
a wheeled inverted pendulum
Sergio
Sergio Delgado
Delgado ,,, Paul Paul Kotyczka
Kotyczka
Sergio
Sergio Delgado
Delgado , Paul Paul Kotyczka
Kotyczka
Technische
Technische Universität München
Universität München
Technische
Technische Universität
Boltzmannstr. 15, D-85748
D-85748
Universität München
Garching
München
Boltzmannstr.
Boltzmannstr. 15,
15, D-85748 Garching
Garching
Tel:
Tel: +49-89-289
+49-89-289 15679;
Boltzmannstr.
15679; e-mail: {s.delgado,
15, D-85748
e-mail: {s.delgado, kotyczka}@tum.de.
Garching
kotyczka}@tum.de.
Tel: +49-89-289 15679; e-mail: {s.delgado,
Tel: +49-89-289 15679; e-mail: {s.delgado, kotyczka}@tum.de. kotyczka}@tum.de.
Abstract:
Abstract: The
The paper
paper deals
deals with
with the
the robust
robust energy-based
energy-based stabilization
stabilization of
of aa wheeled
wheeled inverted
inverted
Abstract:
pendulum,
Abstract: The
which
The paper
is
paper an deals with
underactuated,
deals with the
the robust
unstable
robust energy-based
mechanical
energy-based stabilization
system
stabilization of
subject
of aa to
wheeled inverted
nonholonomic
wheeled inverted
pendulum,
pendulum, which is an underactuated, unstable mechanical system subject
subject to
to nonholonomic
nonholonomic
constraints.
pendulum,
constraints. which
The
which
The is
is an
equilibrium
an
equilibrium underactuated,
to
underactuated,
to be
be stabilizedunstable
is
unstable
stabilized is mechanical
characterized
mechanical
characterized system
by
system
by the
the length
subject
length of of the
to driven
nonholonomic
the driven path,
path,
constraints.
the orientation,
constraints. The
The andequilibrium
the
equilibrium pitch to be
be stabilized
to angle. We
stabilized use is characterized
isthe method
characterized of by the
the length
Controlled
by length of
of the
the driven
Lagrangians driven path,
which
path,is
the
the orientation,
orientation, and
and the
the pitch
pitch angle.
angle. We
We use
use the
the method
method of
of Controlled
Controlled Lagrangians
Lagrangians which
which is
is
applied
applied in
the orientation,
in a
a systematic
and theway,
systematic way,
pitch and
and is
angle.
is very
very We intuitive,
use thefor
intuitive, for it
method
it is
is physically
of Controlled
physically motivated.
motivated. After
Lagrangians
After a
a detailed
which
detailedis
applied
presentation in a systematic
of
applied in a systematic the model way,
way,under and is very
and isnonholonomic intuitive, for
very intuitive,constraints, it is physically
we
for it is physically motivated.
provide an
motivated. After
elegant a detailed
Aftersolution
a detailed of
presentation
presentation of
of the
the model under nonholonomic constraints, we provide an elegant solution of
the matching
presentation
the matching the model
ofequations
equationsmodel for
for
under
kinetic
under
kinetic
nonholonomic
and
nonholonomic
and potential
potential
constraints,
energy
constraints,
energy
we
we provide
shaping
shaping for
provide
for the
the
an elegant
elegant solution
anconsidered
considered systems.
solution
systems.
of
of
the matching
Simulations
the matchingshow equations
show the
equations for kinetic
applicability
for kineticand and
and potential
robustness
androbustness
potential of energy
of the
energy shaping
method. for the considered
shaping for the considered systems. systems.
Simulations
Simulations show the the applicability the method.
Simulations
© 2015, IFACshow the applicability
(International applicability
Federationand
and robustness
robustness
of Automatic
of
of the
Control)
method.
theHosting
method. by Elsevier Ltd. All rights reserved.
Keywords:
Keywords: Underactuated
Underactuated mechanical
mechanical systems,
systems, nonholonomic
nonholonomic systems, passivity-based
systems, passivity-based control. control.
Keywords:
Keywords: Underactuated
Underactuated mechanical mechanical systems, systems, nonholonomic
nonholonomic systems, systems, passivity-based
passivity-based control. control.
1. INTRODUCTION
1. INTRODUCTION position controller
position controller using using energy
energy shaping
shaping techniques
techniques for for
1.
1. INTRODUCTION
INTRODUCTION position
wheeled controller
inverted using
pendulum energy
systems. shaping techniques for
position
wheeled controller
inverted pendulum using energy
systems. shaping techniques for
wheeled
wheeled inverted
inverted pendulum
pendulum systems. systems.
The
The wheeled
wheeled inverted
inverted pendulum
pendulum (WIP)
(WIP) –
– and
and its
its commer-
commer- 1.1 Existing
Existing work work
The wheeled 1.1
The
cial wheeled inverted
cial version,
version, the Segway
inverted
the Segwaypendulum
[2015, (WIP)
pendulum
[2015, Jan] –– ––has
(WIP)
Jan]
and
has its
its commer-
gained
andgained commer-inter- 1.1
inter- 1.1 Existing
Existing work work
cial
est
cial forversion,
for human
version, the Segway
assistance
theassistance
Segway [2015,[2015,
and Jan] –
transportation has
Jan] – has gained gained
in the inter-
past
inter- Several control laws laws have
have been
been applied
applied to to the
the WIP,
WIP, mostly
mostly
est
est for human
human assistance and
and transportation
transportation in
in the
the past
past Several control
several
est for years
human due to its
assistance high
and maneuverability
transportation and
in the simple
past Several
using
Several control
linearized
control laws laws
models have
have(seebeen
(see
been Li applied
et al.
applied to the
[2013], WIP,
Ha
to the HaWIP,and mostly
Yuta
mostly
several
several years
years due
due to its high maneuverability and simple using linearized models Li et al. [2013], and Yuta
construction
several
construction due to
(see,
years (see, to
e.
its
e. g.,
g.,
its Lihigh
Li
highet
maneuverability
et al.
al. [2013]). A
maneuverability
[2013]). A WIPWIP and
and–– simple
shown using
simple
shown [1996],
using linearized
Grasser
linearized
[1996], Grasser
models
et al.
models
et al.
(see
[2002]).
(see
[2002]).
Li
Li et
During
et
During
al.
al. [2013],
the last
[2013],
the last
Ha and
decade,
Ha and
decade,
Yuta
how-
Yuta
how-
construction
from the side (see,
the side
construction in Figure
(see, e.
e. g.,
Figure g.,11Li
Li et
et al.
(left) [2013]).
– consists
al. consists
[2013]). of ofA
AaaWIPvertical
WIP –– shown
body [1996],
shown Grasser
ever, researchers
researchers et al.
have [2002]). During
put aa strong
strong the
focus last decade,
on the
the how-
nonlinear
from
from the in (left) – vertical body [1996],
ever, Grasser et al.
have [2002]).
put During the
focus last
on decade, how-
nonlinear
with two
from
with the side
two coaxial
side
coaxial
in
in Figure
driven11wheels
Figure
driven
(left)
(left) ––mounted
wheels consists
consists of
mounted onaathe
of
on
vertical
the body.body
vertical
body. The ever,
body
The modelresearchers
ever,
model controlhave
for control
researchers
for
put
put aa strong
purposes:
have
purposes: Some focus
strong
Some
on
on the
the nonlinear
accessibility
focus
accessibility and con-
con-
nonlinear
and
with
with two
actuation
two coaxial
of both
coaxial driven
wheels
driven wheels
in
wheels the mounted
same
mounted on
direction
on the
the body.
generates
body. The
The model
trollabilityfor control
analysis purposes:
of the WIP Somehas accessibility
been done byand con-
Pathak
actuation
actuation of
of both wheels in the same direction generates model for control
trollability analysis purposes:
of the WIP Somehas accessibility
been done byand con-
Pathak
a forward
forward (or
aaactuation of both
(or both wheels
backward)
wheelsmotion;
backward)
in
in the
the same
motion; opposite
same
opposite
direction
direction
wheel
generates
velocities trollability
wheel generates
velocities et al.
al. [2005]
[2005]
trollability
et
analysis
analysis
and
of
and Nasrallah
of the
Nasrallah
the WIP
WIP et has
et
been
al. [2007].
has
al. [2007].
been done
Based
done
Based
by
by Pathak
on the
the
Pathak
on
forward
alead (or
(or backward)
to aa turning
forward turning motionmotion;
backward) around opposite
motion; the vertical
vertical
opposite wheel
wheelaxis.velocities
Mobile et
velocities al.
analysis [2005] theand
of the Nasrallah
nonlinear system, et al. [2007].
nonlinear Based
control on the
strate-
lead
lead to
to motion around the axis. Mobile et al.
analysis [2005]
of and Nasrallah
nonlinear system, et al. [2007].
nonlinear Based
control on the
strate-
robotic
lead
robotic to aasystems
turning
systems
turningbased motion
based
motion on around
on the WIP
around
the WIPthethe vertical
like
vertical
like the
axis.
axis. Mobile
the intelligent
intelligent Mobiletwo analysis
two haveof
gies have
analysis
gies
the
been
ofbeen nonlinear
nonlinear system,
developed
the developed nonlinear
for Segway-like
system,
for Segway-like
nonlinear control
systems. strate-
Pathak
control Pathak
systems. strate-
robotic
wheeled
robotic systems
road
systems based
vehicle
based B2on
on the
the WIP
presented
WIP like
by
like the
Baloh
the intelligent
and
intelligent Parenttwo
two gies
et al.have
[2005] been developed
present, e. g., for
two Segway-like
different systems.
two-level Pathak
controllers
wheeled
wheeled road vehicle B2 presented by Baloh and Parent gies
et al.have
[2005] been developed
present, e. g., for
two Segway-like
different systems.
two-level Pathak
controllers
[2003], or
wheeled
[2003], orroad
novelvehicle
road
novel and more
vehicle
and
B2
B2 presented
more car-like systems
presented
car-like
by
by Baloh
systems like and
Baloh
like and
the
Parent
the Segway
Segway
Parent based et
et al.
al. [2005]
based thepresent,
on the
[2005] partially
present, e.
e. g., two
two different
feedback
g., two-level
linearized
different model
two-level controllers
for posi-
posi-
controllers
[2003], or novel and more car-like systems like the Segway based on
on the partially
partially feedback
feedback linearized
linearized model
model for
for posi-
Puma
Puma [2015,
[2003], [2015,
or novel Jan]
Jan] are
andare being
more
being developed
car-like
developedsystems to
to be used
likeused
be as
the Segway
as new
new tion
based andon velocity
the partiallycontrol while
feedback maintaining
linearized stable
model for pitch
posi-
Puma [2015, Jan] are tion and velocity control while maintaining stable pitch
personal
Puma
personal urban
[2015,
urban are being
being developed
transportation
Jan]transportation systems.
developed
systems.
to be
be used
Some
to Some used as
as new
institutes
institutes
tion
new dynamics;
tion and
and velocity
dynamics; Nasrallah
velocity
Nasrallah
control
control
et al.while
et al. [2007]maintaining
while
[2007] design in
maintaining
design
stable
stable pitch
in several
several steps
pitch
steps
personal
have
personalalso urban
developed
urban transportation
their
transportationown WIPssystems.
for
systems. Some
research
Some institutes
purposes,
institutes dynamics;
a posture Nasrallah
and velocity et al.
control [2007]
for design
the WIP in several
moving steps
on an
have
have also
also developed their own WIPs for research purposes, dynamics;
a posture Nasrallah
and velocity et al.
control [2007]
for design
the WIP in several
moving steps
on an
e. g.,
have
e. also developed
g., Yamabico
Yamabico
developedKurara,
Kurara,
their own
own WIPs
introduced
theirintroduced WIPsby byforHa
forHa research
and Yuta
research
and Yuta purposes,
[1996], aainclined
purposes,
[1996],
posture
posture
inclined plane.
and
plane.
and velocity
Many
velocity
Many other
control
other
control types
types
for
for the
of
the WIP
modeling
WIP
of modeling
moving
and
moving
and
on an
control
on an
control
e.
org., Yamabico Kurara, introduced by Ha
Ha and Yuta [1996],
only inclined plane.haveMany other types
e.
or
or
JOE,
g.,
JOE,
JOE,
presented
Yamabico
presented
presented
Kurara,by Grasser
by
by
Grasser
introduced
Grasser
et al.
et
et
al.
al.
[2002],
by [2002],
[2002],
and toto
Yuta
to
give
give
give
[1996],
only
only
approaches
inclined
approaches plane.have also been
Many
also been types of
other implemented of modeling
implemented modeling
and
and
and control
and tested:
tested: For aa
control
For
some
or JOE, examples.
presented These
by systems
Grasser et can
al. be
[2002], further
to used
give as
only approaches
very complete
approaches have
have also
overview been
also been of implemented
the existing
implemented and
work tested:
on
and tested: For
For aa
modeling
some
some examples.
examples. These
These systems
systems can
can be
be further
further used
used as
as very complete overview of the existing work on modeling
serviceexamples.
some robots like likeThese
KOBOKER systems(see(see Lee
canLee beand and Jung used
further [2011]). as and very
very complete
and control
control
complete overview
of overview
WIPs until of
until
of thethe existing
2012existing
the readerwork
reader
workis on modeling
isonreferred
referred
modeling to
service
service robots
robots like KOBOKER
KOBOKER (see Lee and Jung
Jung [2011]).
[2011]). of WIPs 2012 the to
service robots like KOBOKER (see Lee and Jung [2011]). and
Chan
and control
et
control al. of
of WIPs
[2013].
WIPs until
until 2012
2012 the
the reader
reader is
is referred
referred to
to
The stabilization
stabilization and and tracking
tracking control
control for for the
the WIP WIP is is ChanChan et
et al.
al. [2013].
The
The
The stabilization
challenging: The
stabilization and
system
and tracking
belongs
tracking control
to the
control for
class
for the
of
the WIP
underac-
WIP is
is Chan
Energy al. [2013].
etshaping [2013].techniques, like like the
the method
method of of Controlled
Controlled
challenging:
challenging: The
The system
system belongs
belongs to
to the
the class
class of
of underac-
underac- Energy shaping techniques,
tuated mechanical
challenging: The system systems, since
belongs the
tothe number
thenumber of
class ofofunderac-control Energy
Lagrangians,
Energy shaping
shaping or techniques,
Interconnection
techniques, like
like the
and
the method
Damping
method of
of Controlled
Assignment
Controlled
tuated
tuated mechanical
mechanical systems, since control
control Lagrangians, or Interconnection and Damping Assignment
inputs is
tuated
inputs is
mechanical
less thansystems,
less than the number
systems,
the
since
number
since of thedegrees
of
the numberofof
number
degrees of
offreedom.
control
freedom.
Lagrangians,
Passivity-Based
Lagrangians, or
or Interconnection
Control (IDA-PBC),
Interconnection(IDA-PBC), and
and Damping
have been
Damping Assignment
been success-
Assignment
inputs is less than the number of degrees of freedom. Passivity-Based
Passivity-Based Control
Control (IDA-PBC), have
have been success-
success-
Furthermore,
inputs is
Furthermore, less the
than
the upward
the
upward position
number
position of
of
of the
degrees
the body
body of represents
freedom.
represents fully used
Passivity-Based for the stabilization
Control (IDA-PBC), of underactuated mecha-
have been success-
Furthermore, the upward position of the body represents fully
fully used
used for
for the
the stabilization
stabilization of
of underactuated
underactuated mecha-
mecha-
an
an unstable
Furthermore,
unstable equilibrium
the upward
equilibrium which which
position needs
needsof to
the
to be
body stabilized
represents
be stabilized by
by nical
fully systems
used for in
thethe past,
stabilization see, e.
ofg., Ortega
underactuated et al. [2002],
mecha-
an unstable
feedback.
an unstable In equilibrium
addition, the
equilibrium thewhich
system
which needs
needsmotionto
to be is stabilized
be restricted by
stabilized by nical
by nical
Chang
systems
systems
et al.
in
in the
the
[2002].
past,
past,
These
see,
see, e.
e.
methods
g.,
g., Ortega
Ortega
are
et
et al.
al.
attractive
[2002],
[2002],
since
feedback.
feedback. In
In addition,
addition, the system
system motion
motion is
is restricted
restricted by nical
Chang systems
et al. in the
[2002]. past,
These see, e.
methods g., Ortega
are et al.
attractive [2002],
since
nonholonomic
feedback. In addition, (nonintegrable)
the system constraints
motion is (Bloch
restricted[2003]).by they Chang
they
Changshape et
shape al.
et al.the [2002].
the energy
[2002]. These
These of methods
the
methodssystem are attractive
but
arebut preserve
attractive since
its
since
nonholonomic
nonholonomic (nonintegrable)
(nonintegrable) constraints
constraints (Bloch
(Bloch [2003]).
[2003]). energy of the system preserve its
These constraints
nonholonomic do not
(nonintegrable) restrict the
constraints configuration
(Bloch space
[2003]). they
physical shape
they shape the
structure, energy
the energy and of
thus,
of thethe system
appear
system but
natural. preserve
The
but preserve idea its
of
its
These
These constraints
constraints do
do not
not restrict
restrict the
the configuration
configuration space
space physical structure, and thus, appear natural. The idea of
Q̃ on
These which the
constraints dynamics
do not evolve,
restrict but
the the motion
configuration direction
space physical
shaping
physical structure,
the energy
structure, and
can
and thus,
also
thus, appear
be
appear natural.
expanded
natural.to The
The idea
mechanical
idea of
of
Q̃ on which the dynamics evolve, but the motion direction shaping
shaping the
the energy
energy can
can also be expanded to mechanical
Q̃
at on
Q̃
at on
a
which
a given
given the
the dynamics
point:
whichpoint: Becauseevolve,
dynamics
Because thebut
of the
evolve,
of
the
the motion
motion direction
rolling-without-slipping
but
rolling-without-slipping direction systems systems subject
shaping subject
the energy to can also
also be
to nonholonomic
nonholonomic expanded
expanded to
constraints:
be constraints: mechanical
Maschke
toMaschke
mechanical and
and
at a given
constraint point:
it is notBecause
possible of the
to rolling-without-slipping
move sideways, and the systems
Van
systems der subject
Schaft
subject to
to nonholonomic
[1994] stabilize
nonholonomic constraints:
nonholonomic Maschke
constraints: Maschkesystems and
by
and
at a given it
constraint point:
is notBecause
possible of to
themoverolling-without-slipping
sideways, and the Van der Schaft [1994] stabilize nonholonomic systems by
constraint
forward it
velocityis not
of possible
the WIP to
and move
its yaw sideways,
rate are and the
directly Van
shaping der
Van der the Schaft
the
Schaft [1994]
potential stabilize
energy.
[1994] energy. nonholonomic
Muralidharan
stabilizeMuralidharan et
nonholonomicetsystems systems
al. by
[2009]
by
constraint
forward it is not
velocity of possible
the WIP to move
and its yaw sideways,
rate are and the shaping
directly potential al. [2009]
forward byvelocity of shaping
stabilize the the potential energy. Muralidharan et
et al.
al. [2009]
given by
forward
given the angular
velocity
the of the
angular WIP
WIP and
velocity
thevelocity of its
and
of theyaw
its
the wheels.
yaw
wheels.
rate
rate Theare
are directly
The goal of
directly
goal of stabilize
shaping the
the pitch dynamics
potential
pitch dynamics
energy. of of the WIP
Muralidharan
the WIP through
through IDA-
[2009]
IDA-
given by stabilize the
the pitch
pitch dynamics
dynamics of of the
the WIPWIP through
through IDA-
given
this by the
this paper
paper is angular
the
is to present
present
angular
to
velocity
velocity
the
of
of the
the design
design ofwheels.
theof aaa robust
wheels. robust The
The goal
goal of
nonlinear
nonlinear of PBC.PBC.
stabilize IDA-
this paper is to present the design of robust nonlinear PBC.
this paper is to present the design of a robust nonlinear PBC.
Copyright
2405-8963 ©© 2015
2015, IFACIFAC(International Federation of Automatic Control) 93 Hosting by Elsevier Ltd. All rights reserved.
Copyright © 2015 IFAC 93
Copyright ©
Peer review under 2015 IFAC
responsibility of International Federation of Automatic 93 Control.
93
Copyright © 2015 IFAC 93
10.1016/j.ifacol.2015.10.220
IFAC LHMNC 2015
94
July 4-7, 2015. Lyon, France Sergio Delgado et al. / IFAC-PapersOnLine 48-13 (2015) 093–098
Nonholonomic systems violate one of the necessary condi-
tions for asymptotic stabilization by smooth state feedback
formulated by Brockett [1983]. Thus, for the asymptotic
stabilization of a desired configuration q ∈ Q̃, a discontinu-
ous or time-varying control law is required (Astolfi [1996]).
In this paper, to avoid this issue, instead of working in the
WIPs six-dimensional configuration space Q̃, we restrict
our analysis to the three dimensional space Q with local
coordinates consisting of the path length, the pitch, and
the yawing angle: ξ = [s α θ]T ∈ Q. The pitch angle is phys-
ically restricted to −π/2 < α < π/2. We design a passivity-
based controller for the stabilization of an equilibrium
ξ ∗ ∈ Q. The controller is thereafter parametrized apply-
ing local linear dynamics assignment (LLDA), a method
used to fix design parameters in nonlinear passivity based
control by making use of the linearized model (Kotyczka
[2013]). Using this approach, prescribed local dynamics (in
terms of the closed-loop eigenvalues) can be achieved.
The passivity-based controller presented in this note can
be systematically computed and leads to an asymptotically
stable equilibrium ξ ∗ ∈ Q with a large domain of attrac-
tion. Since the closed-loop mechanical energy is used as
Lyapunov function, the framework is remarkably intuitive
for it is physically motivated. Moreover, LLDA allows
for transparency concerning parameter tuning. The ap-
plicability, performance, and robustness of the developed
controller is shown with a series of simulations.
Notation: For compactness of notation, the operator
∇x f (x) is used to denote the transposed Jacobian of a
vector-valued function f (x). Additionally, we will use the
notation s(α) = sin α, and c(α) = cos α. When obvious
from the context, arguments are omitted for simplicity.
2. MODELING
In a mechanical system with nonholonomic constraints,
the n-dimensional manifold Q̃ is the configuration space,
its tangent bundle T Q̃ is the velocity phase space and a
smooth (nonintegrable) distribution D ⊂ T Q̃ represents
the constraints. The Lagrangian L is a map L : T Q̃ → R
and is defined as the kinetic energy minus the potential
energy L = T − V . A curve q(t) is said to satisfy the
constraints if q̇(t) ∈ Dq , for all q ∈ Q̃ and all times t.
For k nonholonomic constraints, the admissible velocities
in a point q are thus restricted to a (n−k)-dimensional
subset (Dq ∼ = Rn−k ) of the tangent space Tq Q̃. The
constraint distribution D is assumed to be regular, i. e.,
of constant rank. The widely used Lagrange-d’Alembert
equations (see, e. g., Bloch [2003])
d  IW
(∇q̇ L) − ∇q L = A(q)λ + Fext (1)
dt
describe the dynamics of systems subject to k nonholo-
nomic (Pfaffian) constraints of the form
AT (q)q̇ = 0. (2)
Assuming there are no external forces other than the input
torques τ̃ , (1) results in
M̃ (q)q̈ + C̃(q, q̇)q̇ + ∇q V (q) = τ̃ + A(q)λ,
T
where M̃ = M̃ is the positive definite mass matrix, and
the term C̃ q̇ represents the Coriolis and centripetal forces.
The constraints have been adjoined to the system using
Lagrange multipliers λ ∈ Rk that represent the magnitude
of the constraint forces which oblige the system to satisfy
the constraints. The work done by these forces vanishes as
can be seen by looking at the corresponding power
Pconstr = q̇ T Aλ = λT AT q̇ = 0. (4)
The approach, as explained in the following, is also used,
e. g., by Pathak et al. [2005] for the modeling of the WIP:
Due to the nonholonomic constraints (2), the admissible
velocities at q ∈ Q̃ must be of the form
q̇ = S(q)ν, (5)
with a smooth full rank matrix S satisfying AT S = 0 for
all q ∈ Q̃, and local coordinates of the constrained tangent
space ν ∈ Dq . The admissible velocities at q lie in the
subspace of Tq Q̃ spanned by the columns of S, which is
nothing but the (n−k)-dimensional space Dq . Now, replace
q̇ = Sν and q̈ = S ν̇ + Ṡν in (3), and eliminate the
constraints by pre-multiplying it by S T
 
S T M̃ S ν̇ + S T M̃ Ṡ + C̃S ν + S T ∇q V = S T τ̃ . (6)
The dynamical system represented by (6) can also be
written in the form
ˆ
M̂ ν̇ + Ĉν + S T ∇q V = τ̂ + Jν, (7)
where M̂ = S T M̃ S, and τ̂ = S T τ̃ . Since the matrix
Ĉ is solely defined by the Christoffel symbols of M̂ , the
matching of the systems (6) and (7) requires, in general,
additional gyroscopic forces Jˆν, where Jˆ = −JˆT , which
are mistakently missing in Muralidharan et al. [2009] for
imposing the constraints before taking variations in the
derivation of the equations of motion (see Bloch [2003]).
2.1 The wheeled inverted pendulum (WIP)
Different modeling approaches for WIPs can be found,
e. g., in Pathak et al. [2005], Delgado et al. [2015], Nas-
rallah et al. [2007]. The dynamic parameters needed for
the modeling of the WIP are listed below in Table 1
with the values used for the simulations. Figure 1 shows
mB body mass 1 kg
mW wheel mass 0.5 kg
r wheel radius 0.05 m
b distance from the wheel axis to
the body’s center of mass 0.08 m
d half of the wheel distance 0.05 m
IB body’s moment of inertia
IBxx around x-axis 1e-5 kg m2
IByy around y-axis 9e-4 kg m2
IBzz around z-axis 4e-4 kg m2
wheel’s moment of inertia
IWyy around y-axis 1e-8 kg m2
IWzz around z-axis 1e-6 kg m2
g gravity constant 10 m/s2
Table 1. System parameters
a simple scheme of the wheeled inverted pendulum. Let
Q̃ = R2 × S1 × S1 × S1 × S1 be the configuration space and
(3)
define local coordinates q = (x, y, θ, α, ϕl , ϕr ) ∈ Q̃. The
coordinates ϕl and ϕr represent the absolute rotation of
the left and right wheel, respectively. The equations
94
 IFAC LHMNC 2015
July 4-7, 2015. Lyon, France Sergio Delgado et al. / IFAC-PapersOnLine 48-13 (2015) 093–098 95

yaw angle
u1 = τr + τl
α
d
u2 = (τr − τl ), (11)
α mB , IB , b r
such that the new inputs u1 and u2 represent the resulting
z
r, mw , IW torque for the forward and the turning motion, respec-
body
tively. The input vector can then be written as
/r 0 � �
�1 �
wheels u1
x y τ̂ = Gu = −1 0 . (12)
θ u2
plane 0 1
2d In order to obtain a model of the WIP suitable for energy-
based control it is helpful to feedback linearize the yawing
Fig. 1. The wheeled inverted pendulum dynamics, for the terms involved in the computation get
simpler. Choose
− s(θ) c(θ) 0 0 0 0
� �
T
u1 = w1 ,
A q̇ = c(θ) s(θ) d 0 −r 0 q̇ = 0 (8)
c(θ) s(θ) −d 0 0 −r u2 = Iθ (α)w2 + 2c4 α̇θ̇ s(α) c(α) + c2 v θ̇ s(α)
represent the rolling-without-slipping constraints of the to get finally the simplified model
wheels. The velocities in a specific configuration q ∈ Q̃ M ν̇ + (C − J)ν + ∇ξ V = Gw, (13)
are thus restricted to  used for controller design. Here, the corresponding matri-
c(θ) 0 0
 ces and vectors are
c1 c2 c(α) 0 0
� � � �
 s(θ) 0 0   
 v M = c2 c(α) c3 0 , ∇ξ V = −c2 g s(α) ,
 0 0 1   α̇ 

q̇ = Sν =  , (9) 0 0 1 0
 0 1 0  θ̇

 1/r 0 d/r   
/r 0 0 −c2 α̇ s(α) −c2 θ̇ s(α)
�1 �
1/r 0 −d/r
G = −1 0 , C − J =  0 0 −c4 θ̇ s(α) c(α) .
where v is the forward velocity of the WIP. With this 0 1 0 0 0
matrix S, the equations of motion are of the form (7) with (14)
�
c1 c2 c(α) 0
� �
0
� Note that the yawing dynamics is not completely decou-
M̂ = c2 c(α) c3 0 T
, S ∇q V = −c2 g s(α) , pled, since a turning motion still affects the pitch and
0 0 Iθ (α) 0 forward motion.
3. ENERGY-BASED CONTROLLER DESIGN
 
/r(τr + τl ) 0 0 c2 θ̇ s(α)
�1 �
τ̂ = −τr − τl , Jˆ =  0 0 0 , (10)
d/r (τr − τl ) This section presents the methodology for the design of the
−c2 θ̇ s(α) 0 0
energy-based controller. Since the procedures IDA-PBC
where and Controlled Lagrangians are equivalent (see Blanken-
Iθ (α) = c4 s2 (α) + c5 , stein et al. [2002], Chang et al. [2002]), the following con-
and troller design can be done in both frameworks analogously.
IW We will put the focus on the Lagrangian case, for velocities
c1 = mB + 2mW + 2 2yy , c2 = mB b, are more intuitive than momenta.
r
c3 = mB b2 + IByy , c4 = IBxx + mB b2 − IBzz , As stated in the introduction, the objective is to design
IW d2 in a systematical way a controller which stabilizes an
c5 = IBzz + 2 yy2 + 2mW d2 + 2IWzz . admissible equilibrium ξ ∗ ∈ Q in the reduced space 2 . An
r
equilibrium is called admissible if G⊥ ∇ξ V |ξ∗ = 0, where
The n−k equations of motion (7) together with the recon-
struction equation (9) describe the motion of the WIP in V is the potential energy of the uncontrolled system. We
the space (q, ν). For a simpler analysis and control synthe- formulate the conditions for the desired closed-loop equi-
librium ξ ∗ to be (asymptotically) stable. The closed-loop
sis, define reduced local coordinates ξ = [s α θ]T ∈ Q ⊂ Q̃, system is parametrized using LLDA to achieve prescribed
such that 1 ξ˙ = ν. In the remaining of the paper we will local dynamics in terms of the closed-loop eigenvalues.
restrict the analysis to the configuration space Q. This is
possible, since S T ∇q V = ∇ξ V , for the potential forces act 3.1 Matching equations
directly on the admissible space D.
The goal of the Controlled Lagrangians procedure is to
2.2 Input and feedback transformation transform (13) by static state feedback w = w(ξ, ν) into a
Lagrangian closed-loop system. Let
1
The control inputs are the motor torques on the right and Lc (ξ, ν) = ν T Mc (ξ)ν − Vc (ξ) (15)
on the left wheel, τr and τl , respectively. These inputs 2
can, however, be transformed into more natural quantities 2 A given configuration q ∗ ∈ Q̃ in the original configuration space
for the control of the WIP. Apply the following input cannot be asymptotically stabilized using the energy-based controller
transformation since Brockett’s necessary condition for asymptotic stabilization is
not met, in which case only convergence to a (non-intuitive) invariant
1 The variable s defines the path length. set can be shown.

95
IFAC LHMNC 2015
96
July 4-7, 2015. Lyon, France Sergio Delgado et al. / IFAC-PapersOnLine 48-13 (2015) 093–098

be the desired closed-loop Lagrangian with mass matrix how to solve the three matching equations independently,
Mc (ξ) = McT (ξ) and potential energy Vc (ξ), and let us which is sufficient to satisfy (22).
consider the Euler-Lagrange equations of motion for the
target system with dissipation (and additional gyroscopic 3.2 Shaping the kinetic energy
forces 3 Chang et al. [2002])
d One can show that the upward equilibrium of the WIP
(∇ν Lc ) − ∇ξ Lc = (Jc − Rc )ν. (16)
dt cannot be stabilized simply by shaping the potential
The target system dynamics evolving on Q can be rewrit- energy, i. e., solving (23b) for an appropriate potential
ten as energy Vc (ξ) and keeping Mc = M . It is thus necessary to
Mc ν̇ + Cc ν + ∇ξ Vc = (Jc − Rc )ν, (17) also shape the kinetic energy by solving (23a) for a positive
or equivalently definite closed-loop matrix Mc �= M . The skew-symmetric
ν̇ = −Mc−1 Cc ν − Mc−1 ∇Vc + Mc−1 (Jc − Rc )ν, (18) matrix Jc is free, and thus, a further design parameter.
Assuming that Jc and Cc are linear in the velocities, and
where the matrix Jc = Jc (ξ, ν) (linear in ν) is skew that the kinetic shaping input is of the form
symmetric, and the closed-loop damping matrix Rc (ξ) is
symmetric. wke = F T (ξ, ν)ν, (25)
3×2
Proposition 1. The equilibrium (ξ ∗ , 0) of the system (17) with F (ξ, ν) ∈ R also linear in the velocities, more
is stable if Mc (ξ) > 0 in a neighborhood Ω of ξ ∗ , modest sufficient conditions for matching are obtained. We
Vc (ξ) has a strict minimum at ξ ∗ , and Rc ≥ 0 in Ω. require
−M Mc−1 (Cc − Jc ) + C − J − GF T ν = 0,
� �
The equilibrium is asymptotically stable if the system is (26)
pervasively damped.
or equivalently, since (26) has to be satisfied for all ν ∈ R3 :
Proof. Consider the closed-loop mechanical-type energy Jc = Cc + Mc M −1 (J − C + GF T ). (27)
as Lyapunov function
Recalling that Ṁc = CcT
+ Cc , the skew symmetry of (27),
1
Ec = ν T Mc ν + Vc . (19) Jc + JcT = 0, can be rewritten as
2
The time derivative of Ec along the trajectories of (17) is Mc M −1 (J −C +GF T )+(J −C +GF T )T M −1 Mc + Ṁc = 0.
(28)
Ėc = −ν T Rc ν, (20) The matrix F in (28) is the velocity feedback matrix in
where it has been used the fact that Ṁc = Cc + CcT . the kinetic energy shaping control law. To extract the
Stability of the equilibrium is shown for Rc ≥ 0. If the conditions that have to be satisfied independently from
damping is pervasive, the largest invariant set under the control, we pre-multiply (28) by G⊥ M Mc−1 and post-
closed-loop dynamics (17) contained in multiply it by Mc−1 M GT⊥ , where G⊥ = [r 1 0] is a full rank
� � left annihilator of G, i. e., G⊥ G = 0. Note that the matrix
(ξ, ν) ∈ Q×R3 | Ėc = 0 (21) equation (28) is symmetric, so is the projected equation
equals the equilibrium (ξ ∗ , 0). Asymptotic stability follows G (J − C)M̄ M + M M̄ (J − C)T GT = G M M̄˙ M GT .
� �
⊥ c c ⊥ ⊥ c ⊥
from La Salle’s invariance principle. An estimate of the (29)
domain of attraction is given by the largest bounded level
set of Ec in Ω. Thus, shaping the kinetic energy only requires the solution
M̄c of this differential equation for M̄c = Mc−1 . One
In order to formulate conditions, under which it is possible possible solution for (29) is
to match both, the system (13) and the desired Euler-  
Lagrange system (17), first replace the target dynamics γk3 φ1 (α) + g
k1 − 0 
(18) in the systems equations of motion (13) to get
 γφ2 (α)
−1
 2

Mc =   − γk3 φ1 (α) + g k3 φ1 (α)  , (30)
−M Mc−1 Cc ν − M Mc−1 ∇ξ Vc + M Mc−1 (Jc − Rc )ν 0 
 γφ2 (α) φ22 (α) 
+ (C − J)ν + ∇ξ V = Gw. (22) 0 0 k2
We require to find an input w, which solves (22), for
the closed-loop system to take the desired form (17). with φ1 (α) = c1 r + c2 c(α), φ2 (α) = c3 + c2 r c(α), and
Splitting the equations in terms of the dependency on constant positive parameters k1 , k2 , k3 , and γ, which are
the velocities ν leads to the matching equations of the chosen such that Mc > 0 in −π/2 < α < π/2. The kinetic
potential (independent from ν) and kinetic (quadratic in energy shaping control (25) can be now derived by pre-
ν) energy, and of the dissipation, which consists of the multiplying (27) by GT M Mc−1
wke = (GT G)−1 GT M Mc−1 (Jc − Cc ) + (C − J) ν. (31)
� �
terms linear in ν. The resulting set of equations
−M Mc−1 (Cc − Jc )ν + (C − J)ν = Gwke , (23a) The matrices Cc and Jc can be easily calculated from
−1
−M Mc ∇ξ Vc + ∇ξ V = Gwpe , (23b) the matrix Mc , and premultiplying (27) by G⊥ M Mc−1 ,
−1 respectively. The matrix Jc takes the form
−M Mc Rc ν = Gwdi , (23c)  
0 −f2 v − f3 α̇ 0
determines the components of the control law
Jc =  f2 v + f3 α̇ 0 −f1 θ̇  (32)
w = wke + wpe + wdi (24)
0 f1 θ̇ 0
related to the shaping of the kinetic and potential energy,
and to damping injection. In the following we demonstrate for some functions 4 fi (ξ).
3 The matrix Jc serves as additional design parameter. 4 The explicit form of the functions fi (ξ) is omitted for brevity.

96
 IFAC LHMNC 2015
July 4-7, 2015. Lyon, France Sergio Delgado et al. / IFAC-PapersOnLine 48-13 (2015) 093–098 97
3.3 Shaping the potential energy
With the new mass matrix of the closed-loop system Mc ,
we can proceed to shape the potential energy by solving
(23b). The corresponding projected matching equation is
G⊥ (∇ξ V − M Mc−1 ∇ξ Vc ) = 0,
which represents a set of linear first order PDEs and can be
easily solved using a computer algebra system. The closed-
loop potential energy takes the form
Vc (ξ) = γ ln(φ1 (α))(r2 c1 − c3 ) − rc2 c(α)
+ Π1 (Φ(s, α)) + Π2 (θ),
where Π1 (Φ(s, α)) is a free function of the homogeneous
solution
γ 2
Φ(s, α) = s − rα + (k1 − k3 )(c3 α + c2 r s(α))
g The resulting closed-loop yawing dynamics are of the form
 
c 3 − c1 r 2
(c2 − c1 r)(1− c(α))
+ 2 2 arctan
c1 r2 − c22 c21 r2 − c22 s(α)
and Π2 (θ) is a free function of θ. Both, Π1 and Π2 , need
to be chosen such that Vc (ξ) has an isolated minimum at
ξ = ξ ∗ . The potential energy shaping control is
wpe = (GT G)−1 GT (∇ξ V − M Mc−1 ∇ξ Vc ). (36)
3.4 Damping injection and control law
To achieve asymptotic stability of the equilibrium (ξ, ν) =
(ξ ∗ , 0), it is necessary to add (pervasive) damping accord-
ing to Proposition 1, for which we need the solution of
(23c) for a dissipation matrix Rc ≥ 0, such that any
possible system motions elicit energy dissipation. First,
define Rc = Mc M −1 R̆M −1 Mc , such that (23c) becomes
−R̆M −1 Mc ν = Gwdi .
T where the model uncertainties are denoted by ∆. Using
Choose the damping matrix as R̆ = GKdi G , for Kdi =
diag(kd,1 , kd,2 ) > 0. We add damping by choosing
wdi = −Kdi GT M −1 Mc ν. (38)
Proposition 2. Consider the equations of motion (13).
Assume there is a matrix Mc (ξ) > 0 and a scalar function
Vc (ξ) which verify (29) and (33), where the function Vc is
such that ξ ∗ = arg min Vc . Then, the closed-loop system
(13) with input w according to (24), with (31), (36), and
(38), has an (asymptotically) stable equilibrium (ξ, ν) =
(ξ ∗ , 0) for Kdi > 0.
Proof. The solution of (23) is sufficient to meet the
requirement for matching (22). Since G is not an invertible
matrix, the equations (23) cannot be trivially solved. Pre-
multiplying (23) by the full rank matrix

G⊥
 γ, and kd,1 , and the function
(39)
GT
splits the matching equations (23) into non-actuated and
fully actuated parts. The fully actuated part leads straight-
forwardly to (24) according to (31), (36), and (38), respec-
tively. The non-actuated part of (23c) is trivially solved
for a damping matrix of the form R̆ = GKdi GT . The PDE
(33) represents the part of the potential matching equation
which is not dependent on the input. It is clear from (27)
and (28) that the non-actuated part of (23a) is equivalent
to (29). The control law (24) composed of the parts cor-
responding to the potential and kinetic energy shaping,
and the damping injection, renders (13) the closed-loop
Lagrangian system (17). Stability of the desired equilib-
rium (ξ ∗ , 0) follows from Proposition 1, since Mc (ξ) > 0,
ξ ∗ = arg minVc , and Rc ≥ 0. Asymptotic stability can be
(33)
shown invoking La Salle’s invariance principle.
3.5 Some remarks on the parameter choice
Since the yawing dynamics of the closed-loop system (17)
(34) is fully actuated, it can be parametrized independently and
arbitrarily. Choose, e. g., the function
1
Π2 (θ) = kp (θ − θ∗ )2 , kp > 0. (40)
k2 θ̈ = −kp (θ − θ∗ ) − kd,2 θ̇ + f1 (ξ)θ̇ α̇, (41)
,
where the term quadratic in the velocities arises from
(35) (32). The closed-loop yawing dynamics (41) can be
parametrized similar to a PD-controller by the choice of
k2 , kp , and kd,2 to achieve desired local behavior. For
the parametrization of the remaining dynamics we apply
LLDA: The 4 free parameters k1 , k3 , γ, and kd,1 , and the
free function Π1 (s, α) are chosen such that the linearized
closed-loop system has desired eigenvalues at the equilib-
rium (ξ ∗ , 0). The procedure results in an asymptotically
stable closed-loop system with desired local dynamics and
a large domain of attraction.
3.6 Robustness
In order to check the robustness of the controller, let us
consider the plant
(M + ∆M )ν̇ + (C − J + ∆C − ∆J)ν + ∇ξ (V + ∆V ) = Gw,
(37) (42)
the controller (24) results in a closed-loop system
(Mc +∆M )ν̇+(Cc −Jc +∆C−∆J)ν+∇ξ (Vc +∆V ) = −Rc ν.
(43)
Since the real system is of mechanical nature, the matrix
˙ ) = ∆C + ∆C T holds.
∆J is skew symmetric, and (∆M
According to Proposition 1, the closed-loop system has an
(asymptotically) stable equilibrium (ξ ∗ , 0) if Mc +∆M > 0
and Vc + ∆V has a strict minimum at ξ ∗ .
4. SIMULATIONS
The yawing dynamics has been parametrized by the choice
of k2 , kp , and kd,2 , such that, locally, it has closed-loop
eigenvalues {−1, −6.2}. The remaining parameters k1 , k3 ,
1
Π1 (Φ(s, α)) = µ (Φ(s − s∗ , α))
2
2
are chosen such that the linearized closed-loop system has
eigenvalues {−1, −2, −3, −6}. Figure 2 shows two level sets
of the Lyapunov function Ec in the plane
(ξ, ν) ∈ Q×R3 | ν = 0, θ = 0 .
 
(44)
The level set of interest is limited by the pitch angle
|α| < π/2. The simulations have been run for the initial
condition α0 = 1.5 rad, θ0 = 2 rad and s0 = ṡ0 = α̇0 =
97
IFAC LHMNC 2015
98
July 4-7, 2015. Lyon, France Sergio Delgado et al. / IFAC-PapersOnLine 48-13 (2015) 093–098

π/2 In Proc. of the Conf. on computational intelligence,

0.116
−2 0.1 robotics, and autonomous systems, 1–7.
0 16
1 7·1 7· Blankenstein, G., Ortega, R., and Van Der Schaft, A.J.
10 −
2 (2002). The matching conditions of controlled la-
α in rad

grangians and ida-passivity based control. Int. J. of

0
Control, 75(9), 645–665.

0.116
0. 7
11 · 1 −
Bloch, A.M. (2003). Nonholonomic mechanics and control.
6 0 2 Springer.
−1 −2
10 Brockett, R.W. (1983). Asymptotic stability and feedback
7·
−π/2 0.116 stabilization. In Differential Geometric Control Theory.
−3 −2 −1 0 1 2 3
Chan, R.P.M., Stol, K.A., and Halkyard, C.R. (2013).
s in m Review of modelling and control of two-wheeled robots.
Fig. 2. Level sets of Ec for θ = 0, ν = 0. Annual Reviews in Control, 37(1), 89–103.
Chang, D.E., Bloch, A.M., Leonard, N.E., Marsden, J.E.,
θ̇0 = 0, and for s∗ = 1 m. For the disturbed model we and Woolsey, C.A. (2002). The equivalence of controlled
have chosen matrices M̂ and Ĉ in (10) to be 1.5 times the lagrangian and controlled hamiltonian systems. ESAIM:
nominal value. The simulation results are shown below in Control, Opt. and Calculus of Variations, 8, 393–422.
Figure 3 and Figure 4. Delgado, S., Gajbhiye, S., and Banavar, R.N. (2015).
Reduced equations of motion for a wheeled inverted
Position in plane Path length pendulum. In Proc. of the 8th Vienna Int. Conf. on
1.5 Mathematical Modelling, 328–333.
Grasser, F., D’Arrigo, A., Colombi, S., and Rufer, A.C.
1 1 (2002). Joe: a mobile, inverted pendulum. IEEE Trans.
y in m

s in m

on Industrial Electronics, 49(1), 107–114.

0.5
nominal Ha, Y.S. and Yuta, S. (1996). Trajectory tracking con-
disturbed trol for navigation of the inverse pendulum type self-
0 0
contained mobile robot. Robotics and autonomous sys-
-0.25 0 0.25 0 2 4 6 8
tems, 17(1), 65–80.
Kotyczka, P. (2013). Local linear dynamics assignment in
x in m t in s ida-pbc. Automatica, 49(4), 1037–1044.
Fig. 3. Response of the path length s for the nominal Lee, S. and Jung, S. (2011). Novel design and control
(solid) and the disturbed model (dashdotted). of a home service mobile robot for korean floor-living
life style: Koboker. In Proc. of the 8th Int. Conf. on
Pitch angle Yawing angle Ubiquitous Robots and Ambient Intelligence, 863–867.
Li, Z., Yang, C., and Fan, L. (2013). Advanced control of
2 2 wheeled inverted pendulum systems. Springer.
Maschke, B.M. and Van der Schaft, A.J. (1994). A
α in rad

θ in rad

1 nominal hamiltonian approach to stabilization of nonholonomic

0 Decision and Control, volume 3, 2950–2950.
−1
0 2 4 6 8 0
t in s
Fig. 4. Response of the pitch and yawing angle for the nom-
inal (solid) and the disturbed model (dashdotted).
5. FURTHER WORK
The methodology presented here can be extended to
the speed control without any further computations as
shown in an extended version of this paper submitted
to Automatica. Further, the stabilization of the position
given by the cartesian coordinates x and y has been solved
by other authors applying a coordinate transformation. We
plan to redesign our approach to achieve that task.
REFERENCES
Astolfi, A. (1996). Discontinuous control of nonholonomic
systems. Systems & Control Letters, 27, 37–45.
Baloh, M. and Parent, M. (2003). Modeling and model
verification of an intelligent self-balancing two-wheeled
vehicle for an autonomous urban transportation system.
disturbed 1 mechanical systems. In Proc. of the IEEE Conf. on
Muralidharan, V., Ravichandran, M.T., and Mahindrakar,
0 A.D. (2009). Extending interconnection and damping
2 4 6 8
assignment passivity-based control (ida-pbc) to under-
t in s actuated mechanical systems with nonholonomic pfaf-
fian constraints: The mobile inverted pendulum robot.
In Proc. of the 48th IEEE Conf. on Decision and Control
& 28th Chinese Control Conf., 6305–6310.
Nasrallah, D.S., Michalska, H., and Angeles, J. (2007).
Controllability and posture control of a wheeled pen-
dulum moving on an inclined plane. IEEE Trans. on
Robotics, 23(3), 564–577.
Ortega, R., Spong, M.W., Gómez-Estern, F., and Blanken-
stein, G. (2002). Stabilization of a class of underactuated
mechanical systems via interconnection and damping
assignment. IEEE Trans. on Aut. Contr., 47(8), 1218–
1233.
Pathak, K., Franch, J., and Agrawal, S.K. (2005). Velocity
and position control of a wheeled inverted pendulum by
partial feedback linearization. IEEE Trans. on Robotics,
21(3), 505–513.
Puma [online]. http://www.segway.com/puma.
Segway [online]. http://www.segway.com.

98