Vous êtes sur la page 1sur 7

botics &

Ro

Advances in Robotics
n
Dhaouadi and Hatab, Adv Robot Autom 2013, 2:2

Au
Adv nces i

tomation
DOI: 10.4172/2168-9695.1000107

& Automation
a

ISSN: 2168-9695

Research Article Open Access

Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and


Newton-Euler Methodologies: A Unified Framework
Rached Dhaouadi* and Ahmad Abu Hatab
College of Engineering, American University of Sharjah, Sharjah, UAE

Abstract
This paper presents a unified dynamic modeling framework for differential-drive mobile robots (DDMR). Two
formulations for mobile robot dynamics are developed; one is based on Lagrangian mechanics, and the other on
Newton-Euler mechanics. Major difficulties experienced when modeling non-holonomic systems in both methods are
illustrated and design procedures are outlined. It is shown that the two formulations are mathematically equivalent
providing a check on their consistency. The presented work leads to an improved understanding of differential-
drive mobile robot dynamics, which will assist engineering students and researchers in the modeling and design of
suitable controllers for DDMR navigation and trajectory tracking.

Keywords: Differential-drive; Mobile robot; Dynamics; Modeling; interact with each other through gears, springs, and frictional elements.
Lagrange; Newton-Euler Therefore, we need to take into account all of these forces. It is clear that
the Newtonian approach includes a few practical difficulties since in
Introduction most cases these forces are not easily quantifiable.
In recent years, there has been a considerable interest in the area The methodology developed by Lagrange overcomes these problems
of mobile robotics and educational technologies [1-7]. For control by expressing the forces in terms of the energies in the system, i.e., the
engineers and researchers, there is a wealth of literature dealing kinetic energy and the potential energy, which are scalar quantities
with wheeled mobile robots (WMR) control and their applications. easily expressible in terms of the system coordinates. The derivation of
However, while the subject of kinematic modeling of WMR is the Lagrange equations requires also that the generalized coordinates
well documented and easily understood by students, the subject of be independent.
dynamic modeling of WMR has not been addressed adequately in the
literature. The dynamics of WMR are highly nonlinear and involve The Lagrangian approach usually provides a powerful and versatile
non-holonomic constraints which makes difficult their modeling method for the formulation of the equations of motion for holonomic
and analysis especially for new engineering students starting their systems. However, for non-holonomic systems, the usual method is to
research in this field. Therefore, a detailed and accurate dynamic model introduce the motion constraint equations into the dynamic equations
describing the WMR motion need to be developed to offer students using the additional Lagrange multipliers. These multipliers are not
a general framework for simulation analysis and model based control constants and are usually functions of all the generalized coordinates
system design. and often of time as well. They represent a set of unknowns whose
values should be obtained as a part of the solution. To solve this
In the case of a differential drive mobile robot (DDMR), for computational complexity, additional methods have been suggested to
example, there is no textbook available that investigates thoroughly remove the presence of the multipliers from the dynamic equations of
the dynamic modeling approach taking into consideration the non- the given system [21,22].
holonomic constraints in a step by step procedure. The analysis is
available mainly in journals, conference papers, and technical reports The focus of this paper is to derive simple and well-structured
[8]. Moreover, the material presented differs from one paper to dynamic equations of the DDMR taking into account the non-
another with different variables and reference frames used, and various holonomic constraints. First, the Lagrange formulation is presented.
assumptions. In addition, some papers present different results for the Coordinates transformation is used to cancel the Lagrange multipliers
same DDMR used, which adds to the confusion of dynamic modeling. to obtain well-structured equations. Second, the Newton-Euler method
is used to derive the dynamic equations of the DDMR. Major difficulties
For the case of DDMR, the methods used are either the Lagrangian experienced in using both methods are illustrated and procedures are
approach [9-15] or the Newton-Euler approach [16-19]. Other outlined to offer a systematic approach to the dynamic modeling of
formalisms such as the Kane’s method have been also suggested as
viable approaches of modeling DDMR [20]. Therefore, it is not clear
for new engineering students and researchers which concept to use and *Corresponding author: Rached Dhaouadi, College of Engineering, American
which method offers a better physical insight on the dynamic behavior University of Sharjah, P.O. Box 26666, Sharjah, UAE, E-mail: rdhaouadi@aus.edu
of the system and the effect of the non-holonomic constraints. Also, it
Received June 19, 2013; Accepted September 16, 2013; Published September
is not clear if both methods will lead to the same final dynamic model. 19, 2013
In the Newton Euler method, one has to take into account two Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive
kinds of forces applied to a system: the given forces and the constraint Mobile Robots using Lagrange and Newton-Euler Methodologies: A Unified
Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107
forces. The given forces include the externally impressed forces by
the actuators while the constraint forces are the forces of interaction Copyright: © 2013 Dhaouadi R, et al. This is an open-access article distributed
under the terms of the Creative Commons Attribution License, which permits
between the robot platform and ground through the wheels. Moreover,
unrestricted use, distribution, and reproduction in any medium, provided the
in a system with interconnected elements, the components may original author and source are credited.

Adv Robot Autom


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107
Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and Newton-Euler Methodologies: A
Unified Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107

Page 2 of 7

DDMR with no major mathematical complexity. It is shown that both This transformation will enable also the handling of motion
methods reach equivalent dynamic equations for the mobile robot between frames.
providing a check on their consistency.
X I = R (θ ) X r (4)
Coordinate Systems It will be seen in the next section that equation (4) is very important
In order to describe the position of the WMR in his environment, in deriving the DDMR kinematic and dynamic models as it describes
two different coordinate systems (frames) need to be defined. the relationship between the velocities in the Inertial Frame and the
Robot Frame.
1. Inertial Coordinate System: This coordinate system is a global
frame which is fixed in the environment or plane in which the WMR Kinematic Constraints of the Differential-Drive Robot
moves in. Moreover, this frame is considered as the reference frame
The motion of a differential-drive mobile robot is characterized by
and is denoted as {XI ,YI} .
two non-holonomic constraint equations, which are obtained by two
2. Robot Coordinate System: This coordinate system is a local main assumptions:
frame attached to the WMR, and thus, moving with it. This frame is
• No lateral slip motion: This constraint simply means that the
denoted as {Xr,Yr}.
robot can move only in a curved motion (forward and backward) but
The two defined frames are shown in Figure 1. The origin of the not sideward. In the robot frame, this condition means that the velocity
robot frame is defined to be the mid-point A on the axis between the of the center-point A is zero along the lateral axis:
wheels. The center of mass C of the robot is assumed to be on the axis
of symmetry, at a distance d from the origin A. y ar = 0 (5)
Using the orthogonal rotation matrix R(θ), the velocity in the
As shown in Figure 1, the robot position and orientation in the
inertial frame gives
Inertial Frame can be defined as
− xa sin θ + y a cos θ =
0 (6)
 xa 
q =  ya 
I
(1) • Pure rolling constraint:
 θ  The pure rolling constraint represents the fact that each wheel
maintains a one contact point P with the ground as shown in Figure
The important issue that needs to be explained at this stage is the
2. There is no slipping of the wheel in its longitudinal axis (xr) and no
mapping between these two frames. The position of any point on the
skidding in its orthogonal axis (yr). The velocities of the contact points
robot can be defined in the robot frame and the inertial frame as follows.
in the robot frame are related to the wheel velocities by:
 xr   xI 
 r  
Let X =  y  , and X =  y I  and be the coordinates of the given point
r I v pR = Rϕ R
 (7)
θ 
 
r θ I 
   v pL = Rϕ L
in the robot frame and inertial frame, respectively. In the inertial frame, these velocities can be calculated as a function
of the velocities of the robot center-point A:
Then, the two coordinates are related by the following
transformation:  x pR= xa + Lθ cos θ (8)

 y pR= y a + Lθ sin θ
X I = R (θ ) X r (2)
 x pL= xa + Lθ cos θ
Where R(θ) is the orthogonal rotation matrix  (9)

 y pL= y a + Lθ sin θ
cos θ − sin θ 0
R (θ ) =  sin θ cos θ 0  (3) Using the rotation matrix R(θ), the rolling constraint equations are
formulated as follows:
 0 0 1 
x pR cos θ + y pR sin θ =Rϕ R
(10)
x pL cos θ + y pL sin θ =
Rϕ L
Using the contact points velocities from equation (x,y) and

Figure 1: Differential Drive Mobile Robot (DDMR). Figure 2: Pure Rolling Motion Constraint.

J Adv Robot Automat


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107
Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and Newton-Euler Methodologies: A
Unified Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107

Page 3 of 7

substituting in (x ,y), the three constraint equations can be written in DDMR. Another alternative form for the kinematic model can be
the following matrix form: obtained by representing the DDMR velocities in terms of the linear
and angular velocities of DDMR in the Robot frame.
Λ ( q ) q =
0 (11)
 xar  cos θ 0
Where  r  sin θ v
= I
q =y a   0    (20)
 − sin θ cos θ 0 0 0  ω 
 θ   0 1 
=Λ ( q )  cos θ sin θ L − R 0  (12)  
 cos θ sin θ − L 0 − R  Dynamic Modeling of the DDMR
and
Dynamics is the study of the motion of a mechanical system
T
q =  xa y a θ ϕ R ϕ L  (13) taking into consideration the different forces that affect its motion
unlike kinematics where the forces are not taken into consideration.
vR = Rϕ R The dynamic model of the DDMR is essential for simulation analysis
 (14)
of the DDMR motion and for the design of various motion control
 vL = Rϕ L
algorithms.
The above constraints matrix Λ ( q ) will be used in the next section
for the DDMR dynamic modeling. A non-holonomic DDMR with n generalized coordinates (q1,q2,…
,qn) and subject to m constraints can be described by the following
Kinematic Model equations of motion:
M ( q ) q + V ( q, q ) q + F ( q ) + G ( q =
) + τ d B ( q )τ − ΛT ( q )
λ
Kinematic modeling is the study of the motion of mechanical (21)
systems without considering the forces that affect the motion. For the
where:
DDMR, the main purpose of kinematic modeling is to represent the
robot velocities as a function of the driving wheels velocities along with M(q) an nxn symmetric positive definite inertia matrix, V ( q, q ) is
the geometric parameters of the robot. the centripetal and coriolis matrix, F ( q ) is the surface friction matrix,
G(q) is the gravitational vector, τ d is the vector of bounded unknown
The linear velocity of each driving wheel in the Robot Frame is
disturbances including unstructured unmodeled dynamics, B(q) is the
therefore, the linear velocity of the DDMR in the Robot Frame is the
input matrix, τ is the input vector, ΛT ( q ) is the matrix associated
average of the linear velocities of the two wheels
with the kinematic constraints, and λ is the Lagrange multipliers vector
=v
vR + vL
= R R
(ϕ + ϕL ) (15) [21,22].
2 2
and the angular velocity of the DDMR is Lagrange dynamic approach


vR − vL
= R R
(ϕ − ϕL ) (16)
Lagrange dynamic approach is a very powerful method for
formulating the equations of motion of mechanical systems. This
2L 2
method, which was introduced by Lagrange, is used to systematically
The DDMRs velocities in the robot frame can now be represented derive the equations of motion by considering the kinetic and potential
in terms of the center-point A velocities in the robot frame as follows: energies of the given system.
 r (ϕR + ϕL ) The Lagrange equation can be written in the following form:
 xa = R
2
 d  ∂L  ∂L
 + = F − Λ ( q )
T λ
 y ar = 0 (17)  (22)
 dt  ∂qi  ∂qi
θ= ω= R (ϕ R + ϕ L )
 2L Where L=T-V is the Lagrangian function, T, is the kinetic energy of
the system, V is the potential energy of the system, qi are the generalized
Thus
coordinates, F is the generalized force vector, Λ is the constraints
R R  matrix, and λ is the vector of Lagrange multipliers associated with the
 xar   2 2   constraints.
 r   ϕ R 
y
 a  = 0 0  ϕ  (18)
The first step in deriving the dynamic model using the Lagrange
 θ   R R   L
   −  approach is to find the kinetic and potential energies that govern the
 2L 2L  motion of the DDMR. Furthermore, since the DDMR is moving in the
The DDMR velocities can be obtained also in the inertial frame as XI-YI plane, the potential energy of the DDMR is considered to be zero.
follows:
For the DDMR, the generalized coordinates are selected as
R R 
 cos θ cos θ  q = [ xa ya θ ϕ R ϕ L ]
T
(23)
 xa   2
r 2

 r R R ϕ  The kinetic energies of the DDMR is the sum of the kinetic energy
q I =
= y a  sin θ sin θ   R  (19)
2 2  ϕ L  of the robot platform without wheels plus the kinetic energies of the

θ  
 
 R R  wheels and actuators.

 2 L 2 L  The kinetic energy of the robot platform is
Equation (19) represents the forward kinematic model of the 1 1
=Tc mc vc2 + I cθ 2 (24)
2 2

J Adv Robot Automat


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107
Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and Newton-Euler Methodologies: A
Unified Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107

Page 4 of 7

While the kinetic energy of the right and left wheel is


M ( q ) q + V ( q=
, q ) q B ( q )τ − ΛT ( q )
λ
(38)
1 1 1
TwR = 2
mwvwR + I mθ 2 + I wϕ R2 (25) Where
2 2 2
 m 0 −md sin θ 0 0
1 1 1 
TwL = 2
mwvwL + I mθ 2 + I wϕ L2 (26)
 0 m md cos θ 0 0 
2 2 2
M ( q ) =  −md sin θ md cos θ I 0 0 ,
where, mc is the mass of the DDMR without the driving wheels and  
 0 0 0 0
actuators (DC motors), mw is the mass of each driving wheel (with
actuator), Ic is the moment of inertia of the DDMR about the vertical  0 0 0 0 I w 
axis through the center of mass, Iw is the moment of inertia of each
driving wheel with a motor about the wheel axis, and Im is the moment
0 −mdθ cos θ 0 0 0
  
of inertia of each driving wheel with a motor about the wheel diameter.
0 −mdθ sin θ 0 0 0
All velocities will be first expressed as a function of the generalized V ( q, q ) = 0 0 0 0 0
coordinates using the general velocity equation in the inertial frame.  
2 2 2
0 0 0 0 0
v= x + y (27) 0
i i i
 0 0 0 0 
The Xi and Yi components of the center of mass and wheels can be
obtained in terms of the generalized coordinates as follow
0 0   − sin θ cos θ cos θ   λ1 
 x=
c xa + d cos θ 0 0   cos θ sin θ sin θ  λ 
 (28)    2
 y= ya + d sin θ  
c =
B ( q ) = 0 0  , and Λ T
( q ) λ  0 L − L  ×  λ3 
     
 xwR= xa + L sin θ  0 −R 0  λ4 
 (29) 1 0 
 ywR= ya + L cos θ 0 1  
 0 0 − R   λ5 

 xwL= xa − L sin θ Next, the system described by equation (38) is transformed into
 (30) an alternative form which is more convenient for the purpose of
 ywL= ya + L cos θ control and simulation. The main aim is to eliminate the constraint
term Λ ( q ) λ in equation (88) since the Lagrange multipliers λi are
T
Using equations (24)-(26) along with equations (27- (30), the total
kinetic energy of the DDMR is
unknown. This is done first by defining the reduced vector
1 1 1
T
= m ( xa2 + y a2 ) − mc dθ ( y a cos θ − xa sin θ ) + I w (ϕ R2 + ϕ L2 ) + Iθ 2 ϕ 
2 2 2 η =  R  (39)
(31) ϕ L 
Next, by expressing the generalized coordinates velocities using the
where the following new parameters are introduced
forward kinematic model (19). Then we have
m
= mc + 2mw is the total mass of the robot, I =+
I c mc d 2 + 2mw L2 + 2 I m
 R cos θ R cos θ 
and is the total equivalent inertia.  xa   R sin θ
 y   R sin θ 
Using equation (22) along with the Lagrangian function, L=T the
 a 1 R R  ϕ R 
equations of motion of the DDMR are given by 
θ  =  −    (40)
mxa − mdθsin θ − mdθ 2 cos θ =
C1 (32)   2 L L  ϕ L 
ϕ R   2 0 
mya − mdθ cos θ − mdθ 2 sin θ =
C2 (33) ϕ L   0 2 
Iθ − mdxa sin θ + mdy
a c os θ =
C3 (34) 
This can be written in the form
I wϕ=
R τ R + C4 (35)
q = S ( q )η (41)
I wϕ=
L τ L + C5 (36)
It can be verified that the transformation matrix S(q) is in the null
where (C1, C2, C3, C4, C5), are coefficients related to the kinematic space of the constraint matrix Λ(q). Therefore we have
constraints, which can be written in terms of the Lagrange multipliers
vector λ and the kinematic constraints matrix Λ introduced in section 3. S T ( q ) ΛT ( q ) =
0 (42)
 C1  Next, taking the time derivative of equation (41) gives
C 
 2 =q S ( q )η + S ( q )η (43)
Λ (q) =
T
 C3  (37)
  Substituting equations (41) and (43) in the main equation (38) we
 C4  obtain
C5  M ( q )  S ( q )η + S ( q )η  + V ( q, q )  S (=
q )η  B ( q )τ − ΛT ( q ) λ
Now, the obtained equations of motion (32)-(36) can be represented (44)
in the general form given by equation (21) as

J Adv Robot Automat


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107
Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and Newton-Euler Methodologies: A
Unified Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107

Page 5 of 7

Next, rearranging the equation and multiplying both sides by leads


to
S T ( q ) M ( q ) S ( q )η + S T ( q )  M ( q ) S ( q ) + V ( q, q ) S ( q )  η
= S T ( q ) B ( q )τ − S T ( q ) ΛT ( q ) λ (45)
where the last term is identically zero. Now defining the new matrices

M (q) = ST (q) M (q) S (q)

V S T ( q ) M ( q ) S ( q ) + S T ( q ) V ( q, q ) S ( q ) ,
B = ST (q) B (q)
The dynamic equations are reduced to the form
M ( q )η + V ( q, q )η =
B ( q )τ (46)
Where
 R2 R2 
 I w + 4 L2 ( mL + I ) 2 (
mL2 − I ) 
2 Figure 3: Robot free body diagram for Newtonian dynamic modeling.
4 L
M (q) =  
 R2 R2 We start the derivation by representing the robot position using
 4 L2 ( mL 2
− I ) w 4L2 (
I + mL 2
+ I ) polar coordinates. Assuming that the robot is a rigid body, its polar
coordinates in the inertial frame can be represented using a complex
 R2  vector
 0 mc dθ 
2 L 1 0 
V ( q, q ) =  , B (q)   rˆ = re jθ (48)
0 1 
2
 R  

 2 L c m dθ 0  Differentiating the above position vector with respect to time will
give us the velocity and acceleration of the robot in the inertial frame.
Equation (46) shows that the DDMR dynamics are expressed only
as a function of the right and left wheel angular velocities (ϕ R , ϕ L ) , the =rˆ re
 jθ + jrθe jθ (49)

rˆ =re jθ + 2 jrθe jθ + jrθe jθ − rθ 2 e jθ
 (50)
robot angular velocity θ and the driving motor torques (τ R ,τ L ) . The
equations of motion (46) can be also transformed into an alternative Simplifying and writing the velocity and acceleration terms in
form which is represented by the linear and angular velocities (v,w)of radial and tangential terms, we have
the DDMR. Using the kinematic model equations (15) and (16), it can  π
j θ + 
be easily shown that the model equations (46) can be rearranged in the =rˆ [ r] e jθ +  rθ  e  2 (51)
following compact form  π
j θ+

rˆ =
 
r − rθ 2  e jθ +  2rθ + rθ e  2 
  (52)
  2I w  1
  m + R 2  v − mc d ω =R (τ R + τ L )
2

   The radial and tangential velocity and acceleration terms are


 (47) defined as

 I+ 2 L2
 L

I w ω
 + mc ω
d v = (τ R τ L )

R2  R vu = r (53)

Newton-Euler approach v = rθ


w
(54)

The first and most important step in Newton-Euler dynamic r − rθ


au =  2
(55)
modeling is to draw the free body diagram of the system and to analyze aw 2rθ + rθ
= (56)
the forces acting on it. The free body diagram of the differential drive
From the above four equations, we can write the following relations
mobile robot is shown in Figure 3. Using the robot local frame {xr, yr}, between the radial and tangential components of the robot velocity and
the following notations are introduced. acceleration
(vu,vw) represents the velocity of the vehicle center of mass C in the a=
u vu − vwθ (57)
local frame; vu is the longitudinal velocity and vw is the lateral velocity;
a=
w vw − vuθ (58)
(au,aw) represent the acceleration of the vehicle's center of mass C;
( Fu , Fu ) are the longitudinal forces exerted on the vehicle by the left
R L
The above equations (57) and (58) are the fundamental acceleration
equations that can be also obtained using the theorem of motion of a
and right wheels; ( Fw , Fw ) are the lateral forces exerted on the vehicle
R L
rigid body in a rotating reference frame [21,22].
by the left and right wheels; θ is the orientation of the robot; ω is the
angular velocity; m is the mass of the robot; and is the yaw moment of The next step is to write the Newton’s second law of motion in the
inertia with respect to the center of mass. robot frame and find the relationship between the forces, torques, and
accelerations. The DDMR exhibits two types of motion: translations in
As it can be seen from the above free body diagram, the only forces the radial and tangential directions, and rotation around the vertical
acting on the robot are actuator forces acting on the robot wheels. axis at the center of mass. Let M be the total mass of the robot including

J Adv Robot Automat


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107
Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and Newton-Euler Methodologies: A
Unified Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107

Page 6 of 7

the wheels and actuators and J the moment of inertia with respect to easily rewrite the general dynamic equations (71) in terms of the wheels
the center of mass. Then the dynamic equations are rotational velocities and actuator torques. The leads to the following
formulation
Ma=
u FuL + FuR (59)  R ( Md 2 + J ) MR   R ( Md 2 + J ) MR 
 +  ϕ
 + − +  ϕL
Ma
= w FwL − FwR (60)  4 L2 4 
R
 4 L2 4 
 
Jθ = ( FuR − FuL ) L + ( FwR − FwL ) d (61)  MdR 2  2  MdR 2  1
− ϕ + 
2  L
ϕ ϕ =
2  R L
τ R (74)
Substituting the acceleration terms from (57) and (58) we get  4 L   4 L  R
FuL + FuR  R ( Md 2 + J ) MR   R ( Md 2 + J ) MR 
vu vwθ +
= (62)  +  ϕL +  − +  ϕR
M 
4L2
4 
 
4 L2 4 

F − FwR
−vuθ + wL
vw = (63)  MdR 2  2  MdR 2  1
M − 2  R
ϕ +  ϕ ϕ =
2  R L
τ L (75)
 4L   4L  R
L d
θ= ( FuR − FuL ) + ( FwR − FwL ) (64) The above equations are also equivalent to those derived using the
J J Lagrangian approach as given by equation (46).
The absence of slipping (pure rolling) in the longitudinal direction
Figure 4 shows the dynamic model of the DDMR representing
and no sliding in the lateral direction creates independence between the
the equations of motion (69) and (70). This model shows clearly the
longitudinal, lateral and angular velocities and simplifies the dynamic
coupling between the motor torques, the linear and angular velocities
equations. These non-holonomic constraints are taken into account
of the robot, and the wheels velocities. This model can be adequately
by defining the velocity of the center-point A in the local frame and
used for DDMR simulation and analysis.
forcing it to be zero. Using the transformation matrix R(θ), we first find
the velocity of the center of mass C in the inertial frame as Actuator Modeling
 xc  cos θ − sin θ   vu  The dc motors which are generally used to drive the wheels of a
=   × (65)
 y c   sin θ cos θ  vw  differential drive mobile robot system are considered to be the servo
actuators. In an armature-controlled dc motor which is the case for
Next, using equation (28), we can find the velocity of the center- our DDMR system, the armature voltage va is used as the control input
point A in the inertial frame. It can then be shown that the lateral while keeping the conditions in the field circuit constant. In particular,
velocity of point A in the local frame is vw − dθ . Therefore, in the for a permanent-magnet dc motor, we have the following equations for
absence of lateral slippage we have vw = dθ (66) the armature circuit
Next, substituting (66) in (62), (63), and combining with (64) we  di
obtain va = Ra ia + La a + ea
dt
1 
dθ 2 + ( FuL + FuR )
vu = (67)  ea = K bωm (76)
M  τ m = K t ia
L Mdv 
=θ ( FuR − FuL ) − 2 u θ (68)  τ = Nτ m
Md 2 + J Md + J
The above two equations are the dynamic equations of the robot where, ia is the armature current,(Ra,La) is the resistance and inductance
considering the non-holonomic constraints. These equations can of the armature winding respectively, ea is the back emf, wm is the rotor
now easily be transformed to show the actuator torques applied to the angular speed , τm is the motor torque, (Kt,Kb) are the torque constant
wheels similar to the notations used in the Lagrangian approach. and back emf constant respectively, N is the gear ratio, and is τ the
1 output torque applied to the wheel.
Mvuθ − Mdθ 2 = (τ R + τ L ) (69)
R Since in the DDMR the motors are mechanically coupled to the
robot wheels through the gears, the mechanical equations of motion
( Md + J )θ + Mdvuθ = RL (τ R − τ L )
2
(70) of the motors are linked directly with the mechanical dynamics of the
Next, these two equations can be written in matrix form as follows DDMR. Therefore each dc motor will have
M 0  vu   0 − Mdθ  vu  1  1 1  τ R  ωmR = Nϕ R
 0 Md 2 + J   θ  +    =     (77)
     Mdθ 0   θ  R  L − L  τ L 
(71) ωmL = Nϕ L
The total dynamic equations of the DDMR with the actuators
As it can be observed, equation (62) is similar to equation (47),
which was obtained using the Lagrangian approach. Note that in the
Newton-Euler approach the mass and inertias of the wheels were not
taken into consideration, and the robot is considered as one rigid body.
Therefore both formulations are equivalent if the inertia and mass
parameters are defined as
M = mc (72)
J = Ic (73)
Next, using the forward kinematics equations (15) and (16), we can Figure 4: DDMR Dynamic Model.

J Adv Robot Automat


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107
Citation: Dhaouadi R, Hatab AA (2013) Dynamic Modelling of Differential-Drive Mobile Robots using Lagrange and Newton-Euler Methodologies: A
Unified Framework. Adv Robot Autom 2: 107. doi: 10.4172/2168-9695.1000107

Page 7 of 7

5. Fang Y, Liu X, Zhang X (2012) Adaptive Active Visual Servoing of Nonholonomic


Mobile Robots. IEEE Trans. Ind. Electron 59: 486-497.

6. Huang HP, Yan JY, Hu Cheng T (2010) Development and Fuzzy Control of a
Pipe Inspection Robot. IEEE Trans. Ind. Electron 57: 1088-1095.

7. Li THS, Yeh YC, Da Wu J, Hsiao MY, Chen CY (2010) Multifunctional Intelligent


Autonomous Parking Controllers for Carlike Mobile Robots. IEEE Trans. Ind.
Electron 57: 1687-1700

8. Campion G, Bastin G, Novel BD (1996) Structural properties and classification


of kinematic and dynamic models of wheeled mobile robot. IEEE Transactions
on Automatic Control 12: 47-62.

9. Fukao T, Nakagawa H, Adachi N (2000) Adaptive Tracking Control of a


Nonholonomic Mobile Robot .IEEE Transaction on Robotics and Automation
16: 609-615.

Figure 5: DDMR dynamic model with actuators. 10. Hou ZG, Zou AM, Cheng L, Tan M (2009) Adaptive Control of an Electrically
Driven Nonholonomic Mobile Robot Via Back stepping and Fuzzy Approach.
IEEE Transaction on Control Systems Technology 17: 803-815.
are obtained by combining equation (76) for each motor with the
mechanical dynamics of the DDMR. Additional torque disturbances 11. Fierro R,Lewis FL (1997) Control of a nonholonomic mobile robot: back
stepping kinematics into dynamics. Journal of Robotic Systems 14: 149-163.
acting on the wheels can be included as additive terms to the motor
torques. Figure 5 shows a block diagram representation of the overall 12. Yamamoto Y, Yun X (1992) On Feedback Linearization of Mobile Robots.
system. The forward kinematic model (19) can be added as a cascade Technical Report No. MS- CIS-92-45, Philadelphia, PA.

to the dynamic model to form a complete model for simulation and 13. Yamamoto Y,Yun X (1992) Coordinating Locomotion and Manipulation of a
analysis of the DDMR. Mobile Manipulator Technical Report No. MS-CIS-92-18 Philadelphia, PA.

14. Sarkar N, Yun X, Kumar V (1994) Control of mechanical systems with rolling
Conclusion constraints: Application to dynamic control of mobile robots Int. J. Robot. Res
13: 55-69.
We have presented a detailed derivation of the dynamic model of a
15. Yun X,Yamamoto Y (1993)Internal dynamics of a wheeled mobile robot. IEEE/
differential-drive mobile robot using the Lagrange and Newton-Euler
RSJ International Conference on Intelligent Robots and System (IROS'93) 2:
methods. They were shown to be mathematically equivalent providing 1288-1294.
a check on their consistency. The equations of motion of DC motors
16. DeSantis RM (1995) Modeling and Path-tracking Control of a mobile Wheeled
actuators were also added to form the complete dynamic model of the Robot with a Differential Drive. Robotica 13: 401-410.
DDMR. The insight gained in this study will assist engineering students
17. de Vries TJA, van Heteren C, Huttenhuis L(1999) Modelling and control of a
and researchers in the modeling and design of suitable controllers for fast moving, highly maneuverable wheelchair. Proceedings of the International
DDMR navigation and trajectory tracking. Bio mechatronics Workshop 6: 110-115.
References 18. Albagul A, Wahyudi A (2004) Dynamic Modelling and Adaptive Traction Control
for Mobile Robots. International Journal of Advanced Robotic Systems 1: 149-154.
1. Mitchell R, Warwick K, Browne WN, Gasson MN, Wyatt J (2010) Engaging
Robots: Innovative Outreach for Attracting Cybernetics Students. IEEE 19. Thanjavur K, Rajagopalan R (1997) Ease of dynamic modelling of wheeled
Transactions on Education 53: 105-113 mobile robots (WMRs) using Kane's approach. IEEE International Conference
on Robotics and Automation 4 : 2926-2931
2. Dhaouadi R, Sleiman M (2011) Development of a modular mobile robot platform
for motion-control education. IEEE Industrial Electronics Magazine 5: 35-45. 20. Thanjavur K, Rajagopalan R (1997) Ease of dynamic modelling of wheeled
mobile robots (WMRs) using Kane's approach. IEEE International Conference
3. Shibata T, Murakami T (2012) Power-Assist Control of Pushing Task by on Robotics and Automation 4: 2926-2931
Repulsive Compliance Control in Electric Wheelchair. IEEE Trans. Ind. Electron
59: 511-520. 21. Neǐmark JI, Fufaev NA (1972) Dynamics of Nonholonomic Systems:
Translations of Mathematical Monographs. American mathematical Society
4. Yang SX, Zhu A, Yuan G, Meng MQH (2012) A Bioinspired Neurodynamics-
Based Approach to Tracking Control of Mobile Robots. IEEE Trans. Ind. 22. Bloch A, Crouch P, Baillieul J, Marsden J (2003) Nonholonomic Mechanics and
Electron 59: 3211-3220. Control.

J Adv Robot Automat


ISSN: 2168-9695 ARA, an open access journal Volume 2 • Issue 2 • 1000107