You are on page 1of 5

Dynamic Modeling And Inverse Dynamic Control

Of Mobile Robot
AZAZA Awatef BEN HAMED Mouna
Research Unit of Photovoltaic, Research Unit of Photovoltaic,
Wind and Geothermal Systems, Wind and Geothermal Systems,
National Engineer School of Gabès, National Engineer School of Gabès,
University of Gabès, University of Gabès,
Rue Omar Ibn Elkhattab, Rue Omar Ibn Elkhattab,
Zrig, Gabès, 6029, Tunisia Zrig, Gabès, 6029, Tunisia
Email: atoufaazaza@hotmail.fr Email: benhamed2209@yahoo.fr

Abstract—This paper presents a dynamic modeling and control robust control method, inverse dynamics control and other
studies of mobile robot. The robot is mainly built all around methods.Then, the adaptive control may have some advan-
three wheels and platform. A mathematical model of unicycle tages. For instance the hugeness of calculation capability and
mobile robot is determined by kinematic and dynamic model.
The dynamic modelling equations are based on lagrangian its complication. But the robust control can make the system
formulation. The motion control strategy is based on the inverse unstable [14] and [15]. This paper discussed the dynamic
of dynamic control. This leads to accurate tracking of trajectory. model in the form of a second-order vector differential equa-
The goal to reach in this paper is to improve the performances tion. The dynamics model is based on the Lagrange formula
of inverse dynamic control of mobile robot. The validity of the because it can easily be applied. But the control strategies
proposed controller is demonstrated by the simulation of two
wheels mobile robots case. using the inverse dynamics control can avoid the nonlinear
Keywords: modeling dynamic, inverse dynamic control, system, decoupling and the realization of tracking trajectory
mobile robot. of the system[3].

I. I NTRODUCTION II. K INEMATICS AND DYNAMICS M ODEL OF MOBILE


ROBOT
Mobile robots can be utilized in many applications such
as industry, medicine and agriculture, which is capable of A. Kinematics Model
autonomous motion. Then, it can be found in many config- In this paper we focused on a mobile robot that has
urations. In mobile robots, there are unicycle robots, tricycle, two independent wheels and single freewheel to ensure the
car-like, omnidirectional robots and spherical mobile robot. stability of the system. The coordinate of mobile robot is
Mobile robots are very complicated in mathematical statics given by (Fig. 1):
and dynamics model, electronics, control theory and manufac- The position of the mobile robot wheels, in this environment
turing. Many advanced researches are interested in the area of
modeling and control of mobile robotics and their localization
[2]. In mathematical model, there are many methods. This
enables us to calculate dynamics models such as formula
lagrangian, Newton-Euler methodologies, Kanes method and
other formulations. In the Newton-Euler method, the dynamic
model can be based on two kinds of forces. The first, is given
forces and the second, constraint forces. The given forces
contain the external forces which are applied to the actuators.
But the constraint forces are the forces of interaction between
the robot platform and ground through the wheels [1]. The
methodology lagrange is the derivation of the kinetic energy
and the potential. The kinetic energy is expressed in function
of the speed of rotation, speed of translation and the forces
exercised on the system [1].
In fact, the dynamic model of mobile robot is generally non- *
linear, a complex form and multivariable system.To overcome Fig. 1. Coordinate of mobile robot.
these problems, there are several control techniques that can
solve this problem, such as: the indirect adaptive control, has two different coordinate systems (inertial coordinate

978-1-5090-6287-4/17/$31.00 ©2017 IEEE


system, robot coordinate system).
R1 = (O, X, ⃗ Y⃗ , Z):
⃗ inertial coordinate system is a global
reference frame.
⃗ robot coordinate system is moving
R2 = (D, ⃗et, ⃗eθ, Z):
frame.
Where:
q1 and q2 two coordinates that define the position of the
origin D of the mobile robot.
   
x et
   
   

q1 =  y   and q2 =  eθ 
 (1)
Fig. 2. Rolling contact Characterization.

   
θ θ Then:
The two coordinates q1 and q2 are related by the transforma- − sin(θ)ẋ + cos(θ)ẏ = 0 (11)
tion given with (2), (3) and (4) :
{ 2) The second constraint has no lateral slip motion: The
et = x cos(θ) + y sin(θ) wheel must not drag orthogonally to the axis of the
(2)
eθ = −x sin(θ) + y cos(θ) wheel. So the movement is along the tangential axis
{ (ėθ= 0) [1], [4], [9], [10] and [13].
x = cos(θ) × et − sin(θ) × eθ
(3) {
y = sin(θ) × et + cos(θ) × eθ
cos(θ)ẋ + rφ̇1 cos(θ) cos(θ) = 0
  (12)
cos(θ) − sin(θ) 0 sin(θ)ẏ + rφ̇1 sin(θ) sin(θ) = 0
 
  Then:

q1 =  sin(θ) cos(θ) 0  × q2 (4)
 cos(θ)ẋ + sin(θ)ẏ = −rφ̇1 (13)
 
0 0 1 VA = rφ̇1 = CA × θ̇ + VC = BA × θ̇ + VB (14)
With:
VA = BA × θ̇ + VB = rφ̇1 = 2l × θ̇ + rφ̇2 (15)
 
cos(θ) − sin(θ) 0
  Then:
  r
θ̇ = (φ̇1 − φ̇2 ) (16)
R(θ) = 
 sin(θ) cos(θ) 0 
 (5) 2l
 
1 VB = rφ̇2 = CB × θ̇ + VC (17)
0 0
1 r
R(θ): Orthogonal rotation matrix according to the axis Z [1]. VC = (VA + VB ) = (φ̇2 + φ̇1 ) = ėt (18)
The movement of mobile robot is described by two constraint 2 2
equations, which are obtained by two the following assump-
   r r

tions: ėt 2 2  
1) The first constraint pure rolling: The contact between the     φ̇1
   
wheels and the ground during locomotion is punctual. 
q̇2 =  ėθ  
= 0 0 ×

 (19)
So the speed of the wheel is relative to the ground at     φ̇2
the point of contact, it is equal to zero. The wheels are θ̇ r
2l
− 2l
r

non-deformable with a constant radius r. The contact


between the wheels and the ground is given by (Fig. 2):
The speed of the wheel is relative to the ground at the Where:
point P is given by the equation (6):
XD = Xc + d cos(θ), YD = Yc + d sin(θ) (20)
⃗P = V
V ⃗A + ω ⃗ = ⃗0
⃗ ˆAP (6)
XA = Xc + l sin(θ), YA = Yc − l cos(θ) (21)
⃗ + ẏ Y
(ẋX ⃗ − cos(θ)Y
⃗ ) + φ̇1 (sin(θ)X ⃗ )ˆ(−rZ)
⃗ = ⃗0 (7)
⃗ + (ẏ + rφ̇1 sin(θ))Y
⃗ = ⃗0 XB = Xc − l sin(θ), YB = Yc + l cos(θ) (22)
(ẋ + rφ̇1 cos(θ))X (8)
{ ẊD = Ẋc − dθ̇ sin(θ), ẎD = Ẏc + dθ̇ cos(θ) (23)
ẋ + rφ̇1 cos(θ) = 0
(9)
ẏ + rφ̇1 sin(θ) = 0 ẊA = Ẋc + lθ̇ cos(θ), ẎA = Ẏc + lθ̇ sin(θ) (24)
{
sin(θ)ẋ + rφ̇1 sin(θ) cos(θ) = 0
cos(θ)ẏ + rφ̇1 cos(θ) sin(θ) = 0
(10) ẊB = Ẋc − lθ̇ cos(θ), ẎB = Ẏc − lθ̇ sin(θ) (25)
B. Dynamics Model τ : is the load torque.
To determine the dynamic model of the system, we must H(q): is the mobile robot inertia matrix.
apply the formula of Lagrange. It can be written by the K(q) : is the matrix of centrifugal and coriolis torques.
equation (26): With :
 
r2 r2
d dl dl IW + 4l 2
2 (mt l + It ) (m t l 2
− I t )
 
4l 2
( )− = Qi − τi (26) H (q) =  
dt dq̇i dqi 2
r
(m l 2
− I ) r2 2
4l2 t t I W + 4l2 (m t l + I t )
With:
  (38)
l: Lagrangian of the system. r2
0 m d d θ̇
Lagrangian equation is given by the equation (27): 2l
K(q̇, q) =   (39)
l =E−U (27) r2
2l md dθ̇ 0
E: Total kinetic energy of the system.  
φ̇1
U : Total energy potential of the system. q̇ =   (40)
Let’s consider a moving rigid body (U = 0) φ̇2
Then, l is given by (28)
III. I NVERSE DYNAMICS CONTROL OF MOBILE ROBOT
l=E (28)
The aim of inverse dynamics control is to make the system
The mobile robot is composed of four bodies: two motor- linear and decoupled and also ensure the desired trajectory
ized wheels (left wheel, right wheel), a freewheeling and a tracking.The dynamic of mobile robot is given by (37).We
platform. The kinetic energy for each corresponding body is choose the acceleration as a control input given by (41):
presented by (29)-(32)[1], [12]and [15]:
H(q) × u1 + K(q̇, q)q̇ = τ (q) (41)
Ecrouelibre = 0 (29)
1 1 1 u1 Can now be written by the following expression:
2
EcrA = mr (ẊA + ẎA2 ) + IW φ̇21 + Im θ̇2 (30)
2 2 2 u1 (t) = −K0 q(t) − K1 q̇(t) + R(t) (42)
1 2 1 1
EcrB = mr (ẊB + ẎB2 ) + IW φ̇22 + Im θ̇2 (31) K1 and K0 Are constant, diagonal and positive matrix
2 2 2
1 1
Ecplatf orme = md (Ẋd2 + Ẏd2 ) + Id θ̇2 (32) K1 = diag {2wi ξi } (43)
2 2
And { }
We add equations (29), (30), (31) and (32) the following
expression is obtained: K0 = diag wi2 (44)

Ectotale = 12 md (Ẋd2 + Ẏd2 ) + 21 Id θ̇2 + 12 mr (ẊA


2
+ ẎA2 ) In the following expression R(t) is chosen as the reference
input:
+ 12 mr (ẊB
2
+ ẎB2 ) + 12 IW φ̇21 + 12 IW φ̇22 + Im θ̇2 R(t) = q̈d (t) + K1 q̇d (t) + K0 qd (t) (45)
(33)
The total kinetic energy can be written as: By combining the equation (41) with (37) the following
expression of the tracking error e = q-qd is described by :
1 1 1 1
Ec = mt (Ẋc2 + Ẏc2 ) + It θ̇2 + IW φ̇21 + IW φ̇22 (34)
2 2 2 2 ë + K1 ė + K0 e = 0 (46)
With: The closed loop system is identical to a system of a second
mt = md + 2mr (35) order. The closed-loop system is globally decoupled and
And asymptotically stable provided that K1 and K0 are symmetric
It = md d2 + 2mr l2 + Id + 2Im (36) matrix [7], [8] and [14].

Where mt is the total vehicle mass, md is the mass of the IV. SIMULATION RESULT
platform without actuators, mr is the mass of the wheel with In order to demonstrate the performance of the inverse
actuators, Id is the moment of inertia of the platform about dynamics controller from the simulation results in matlab
the vertical axis through the center of mass, Iw is the moment simulink we suppose the desired trajectories of the wheels
of inertia of wheel with a motor about the wheel axis and positions are opted to be q1 = sin(θ)×t2 (rad) and q2 =
Im is the moment of inertia of wheel with a motor about the cos(θ)×exp(− t) (rad).
wheel diameter [1]. A numerical simulation was made out on a model of unicycle
The matrix form of the system can be expressed as: mobile robot. [ ]
18 0
H(q)q̈ + K(q̇, q)q̇ = τ (q) (37) K0 = (47)
0 18
And [ ]
23 0
K1 = (48)
0 23
The figures Fig. 3 - 6 show that the real trajectory of position
and speed of the robot is following the desired trajectory
perfectly.The figures Fig. 8-9 shows the actual trajectory of
position of the left and the right wheel of the unicycle in the
presence of external disturbances of the model dynamic F1 =
sin(−100 × t) × t5 (rad) and F2 = 105 × cos(t) × t2 (rad) that
not could track the reference trajectory very accurately and
perfectly.This explains that The controller based on inverse
dynamic has not good performances and not robustness when
the modeling dynamic of mobile robot including a disturbance
vector.
The mobile robot parameters are given by table I
Fig. 4. Trajectory of position of right wheel.
TABLE I
TABLE OF MOBILE ROBOT PARAMETRS

parameter lift-wheel right-wheel platform


mr (kg) 0.3 0.3 -
md (kg) - - 0.45
Iw (Kgm2 ) 0.12 0.12 -
Id (Kgm2 ) - - 0.2
Im (Kgm2 ) 0.143 0.143 -
r (m) 0.125 0.125 -
l (m) - - 0.2
d (m) - - 0.15

Fig. 5. Trajectory of speed of left wheel.

Fig. 3. Trajectory of position of left wheel.

Fig. 6. Trajectory of speed of right wheel.


But when the modeling dynamic of unicycle robot has the
external perturbation, the performances of proposed control
are not excellent.So we need in our search to find other more
robust methods to avoid the disturbance effect on the system.
Then we are going to deal with the experimental tests in
the purpose of evaluating the performances of the conceived
system and its control law by applying new methods.
R EFERENCES
[1] R. Dhaouadi and A. Abu Hatab, Dynamic modelling of differential-drive
mobile robots using lagrange and newton-euler methodologies: a unified
framework, Advances in Robotics & Automation, 2013.
[2] T. Mac, C. Copot, R. Keyser, T. Tran and T. Vu, Mimo fuzzy control
for autonomous mobile robot, Department of Electrical Energy, Systems
Fig. 7. Trajectory of perturbation of left and right wheel. and Automation, Ghent University, School of Mechanical Engineering,
Hanoi University of Science and Technology (HUST), vol. 4, pp. 277–
282, February 2016.
[3] A. Zubizarreta, M. Marcosa, I. Cabanes and C. Pinto, A procedure
to evaluate Extended Computed Torque Control configurations in the
StewartGough platform, Robotics and Autonomous Systems, vol. 59,
pp. 770–781, 2011.
[4] F. Salem, Dynamic and kinematic models and control for differential
drive mobile robots, International Journal of Current Engineering and
Technology, vol. 3, pp. 253–263, num. 2, 2013.
[5] M. Patil, Modelling and simulation of dc drive using PI and PID
controller, International Journal Of Innovative Research In Electrical,
Electronics, Instrumentation And Control Engineering, vol. 2, num. 12,
2014.
[6] X. Zhang, Y. Fang and N. Sun , Visual servoing of mobile robots for
posture stabilization: from theory to experiments, International Journal
of Robust and Nonlinear Control, Industrial Electrical and Electronic
Engineering SanatkadeheSabze Pasargad, vol. 25, pp. 1–15, num. 1, 2015.
[7] F. Piltan, F. Bayat and B. Boroomand, Design Error-Based Linear Model-
Free Evaluation Performance Computed Torque Controller, International
Fig. 8. Trajectory of position of left wheel. Journal of Robotics and Automation, Industrial Electrical and Electronic
Engineering SanatkadeheSabze, vol. 3, pp. 151–166, 2012.
[8] D. Choi and J. HoOh, Active Suspension for a Rapid Mobile Robot Using
Cartesian Computed Torque Control, Journal Intellgent Robot System,
2014.
[9] J. Jiang, B. Yao, J. Guo and Q. Wei Chen, Modeling and Nonlinear Com-
puted Torque Control of Ship-mounted Mobile Satellite Communication
System, International Journal of Automation and Computing, pp. 459–
466, 2012.
[10] I. Zohar, A. Ailon and R. Rabinovici, Mobile robot characterized by
dynamic and kinematic equations and actuator dynamics: Trajectory
tracking and related application,Robotics and Autonomous Systems,
vol. 59, pp. 343-353, 2011.
[11] A. Zelei, L. Kovcs and G. Stpn, Computed torque control of an under-
actuated service robot platform modeled by natural coordinates, Commun
Nonlinear Science and Numerical Simulation, pp. 2205-2217, 2011.
[12] T. DungLe, H. JunKang, Y. SooSuh and Y. ShickRo, An on line self-gain
tuning method using neural networks for nonlinear PD computed torque
controller of a 2-dof parallel manipulator, Neurocomputing, vol. 116,
Fig. 9. Trajectory of position of right wheel. pp. 53-61, 2013.
[13] J. Cerkala and A. Jadlovsk, Nonholonomic mobile robot with differential
chassis mathematical modelling and implementation in simulink with
friction in dynamics, Department of Cybernetics and Articial Intelligence,
V. C ONCLUSION Faculty of Electrical Engineering and Informatics, vol. 15, 2015.
[14] Y. Li, L. Ding and G. Liu, Attitude-based Dynamic and Kinematic
In this paper, the modeling kinematic and dynamic of Models for Wheels of Mobile Robot on Deformable Slope, Robotics and
unicycle mobile robot is discussed. So the kinematic model Autonomous Systems, 2015.
of mobile robot motion was established using constraint pure [15] T. Hoang Tran, V. Tinh Nguyen, M. Tuan Pham and T. Cat Pham,
Trajectory Tracking Control of a Mobile Robot by computed torque
rolling and constraint with no lateral slip motion and the method with on-line learning neural network, 10th IEEE International
dynamics model was developed using the formula lagrangian. Conference on Control and Automation (ICCA) Hangzhou, China, 2013.
The problem of decoupling and nonlinear model dynamics for
unicycle-type vehicles is solved by using the inverse dynamic
control. The simulation result of control we have already
proposed in this paper has good responses which allow to
achieve the trajectory tracking when there is no disturbances .

You might also like