See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225622849

Robust adaptive control for mobile manipulators

Article  in  International Journal of Automation and Computing · February 2011

DOI: 10.1007/s11633-010-0548-y

CITATIONS READS

32 172

3 authors:

Mohamed Boukattaya Tarak Damak

National School of Engineering of Sfax, Tunisia University of Sfax
30 PUBLICATIONS   312 CITATIONS    82 PUBLICATIONS   1,002 CITATIONS   

SEE PROFILE SEE PROFILE

Mohamed Jallouli
University of Sfax
60 PUBLICATIONS   360 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Parallel robot View project

Mobile robot control View project

All content following this page was uploaded by Mohamed Boukattaya on 07 January 2014.

The user has requested enhancement of the downloaded file.

International Journal of Automation and Computing 8(1), February 2011, 8-13
DOI: 10.1007/s11633-010-0548-y

Robust Adaptive Control for Mobile Manipulators

Mohamed Boukattaya1 Tarak Damak1 Mohamed Jallouli2
1
Control of Industrial Process Unit, National School of Engineers, University of Sfax, BP 1173, Sfax 3038, Tunisia
2
Intelligent Control Design and Optimization of Complex Systems Unit, National School of Engineers, University of Sfax,
BP 1173, Sfax 3038, Tunisia

Abstract: This paper addresses the trajectory tracking control of a nonholonomic wheeled mobile manipulator with parameter
uncertainties and disturbances. The proposed algorithm adopts a robust adaptive control strategy where parametric uncertainties are
compensated by adaptive update techniques and the disturbances are suppressed. A kinematic controller is first designed to make the
robot follow a desired end-effector and platform trajectories in task space coordinates simultaneously. Then, an adaptive control scheme
is proposed, which ensures that the trajectories are accurately tracked even in the presence of external disturbances and uncertainties.
The system stability and the convergence of tracking errors to zero are rigorously proven using Lyapunov theory. Simulations results
are given to illustrate the effectiveness of the proposed robust adaptive control law in comparison with a sliding mode controller.

Keywords: Mobile manipulator, trajectory tracking, robust adaptive control, sliding mode control, uncertainties and disturbances.

1 Introduction tive hybrid control scheme was proposed, which combines

the kinematics of the mobile platform and the unified dy-
Mobile manipulators refer to robotic manipulators (or namic model of the mobile manipulator. This controller
arms) mounted on mobile platforms (or vehicles). Such consisted of a kinematics part for the mobile platform and
systems combine the advantages of mobile platforms and a dynamic part for the robot arm. For further considera-
robotic arms and reduce their drawbacks. The mobile plat- tion of unmodeled dynamics and external disturbances, a
form extends the arm0 s workspace, whereas the arm offers radial basis function neural-network (RBFNN) was adopted
much operational functionality. Applications for such sys- in the adaptive controller. An unified model and robust NN
tems could be found in mining, construction, forestry, plan- controller was also investigated in [11], where a computed
etary exploration, and military[1−3] . With the assumption torque method and radial basis function NN were used to
of known dynamics, much research has been carried out. In approximate the unstructured or structured uncertainties
[4], a nonlinear feedback control for the mobile manipula- of the mobile manipulator. A minimal controller synthe-
tor was developed to compensate for the dynamic interac- sis adaptive control strategy for controlling the attitude of
tion between the mobile platform and the arm to achieve a rigid body satellite was adopted in [12]. The controller
tracking performance. In [5], a coordination and control of could achieve excellent closed-loop control despite the pres-
mobile manipulators were presented with two basic task- ence of plant parameter variations, external disturbances,
oriented controls: end-effector task control and platform dynamic coupling within the plant, and plant nonlinear-
self posture control. In [6], a control method based on an ities. In [13], a hybrid sliding-mode fuzzy neural network
extended Jacobian transpose was proposed to compensate (HSMFNN) controller was presented. The proposed control
for dynamic interactions between the manipulator and plat- law consisted of a kinematic velocity part for the control of
form. In [7], a computed torque control was developed, such the mobile platform, the onboard arm separately, and a ro-
as a car-like mobile manipulator and a differentially driven bust tracking control system based on HSMFNN to ensure
mobile manipulator. Most previous approaches require a the velocity tracking ability under dynamic uncertainties.
precise knowledge of the dynamics of the system and ig- A robust global stabilization problem of a class of uncer-
nore external disturbance. These issues make the proposed tain nonlinear systems with input unmodeled dynamics was
schemes inappropriate for realistic applications. To handle considered in [14]. The proposed controller can achieve the
these difficulties, adaptive schemes were investigated to deal tracking performance using a dynamic output compensator.
with mobile manipulators with unknown parameters and An output-feedback controller was adopted in [15] to control
disturbances. In [8], adaptive neural network (NN) based a high-order nonlinear system with unmodeled dynamics.
controls were proposed for the joint-space position control By introducing a new rescaling transformation, adopting
of a mobile manipulator. Each NN controller output com- an effective reduced-order observer, and choosing an inge-
prised a linear control term and an NN compensation term nious Lyapunov function and appropriate design parame-
for the online estimation of unknown nonlinear dynamics ters, the proposed algorithm maintains the desired control
caused by parameter uncertainty and disturbances. In [9], performance and reduces the control effort. Most of the pre-
an NN based adaptive controller using radial basis function vious research work designed the controller in generalized
network (RBFN) was proposed for a mobile manipulator coordinates and used a complicated and a computationally
to track the given trajectories. An additional control in- expensive learning algorithm. In order to overcome these
put was added to overcome the approximation error due difficulties, an adaptive control in task space coordinates
to the inexact approximation by RBFN. In [10], an adap- will be proposed based on the advantage of the robot re-
gressor dynamics that expresses the highly nonlinear robot
Manuscript received October 23, 2009; revised April 6, 2010
 M. Boukattaya et al. / Robust Adaptive Control for Mobile Manipulators 9

dynamics in a linear form in terms of the known and un- For the mobile platform, the kinematic equation relating
known robot parameters. The proposed adaptive algorithm linear velocity of F to the wheel velocities is
does not rely on precise prior knowledge of dynamics pa- r 
rameters, and it can suppress disturbances and modeling à ! r
(bC0 + (d + La )S0 ) (bC0 − (d + La )S0 )
errors caused by parameters uncertainties. This paper is ẋF 2b 2b 
= r r ×
organized as follows. Section 2 is devoted to kinematic and ẏF (bS0 − (d + La )C0 ) (bS0 + (d + La )C0 )
dynamic modeling of the mobile manipulator with nonholo- 2b 2b
à !
nomic constraints. Section 3 presents the design of the ro- θ̇R
bust adaptive controller. Section 4 presents computer sim- (1)
θ̇L
ulation results to illustrate the effectiveness of the proposed
theory. Conclusions are formulated in Section 5. where θ̇R and θ̇L are the angular velocities of the right and
Notations. left wheels, respectively. C0 = cos(ϕ) and S0 = sin(ϕ).
O: The intersection of the platform axis of symmetry The linear velocity of the end-effector is found using the
with the driving wheel axis. fact that its base velocity is known and given by (1). There-
C: The center of mass of the platform. fore, the end-effector velocity is written as
F : The manipulator base point.
E: The end-effector of the manipulator. Ã ! Ã !
ẋE ẋF
ϕ: The heading angle. = +
d: Distance between O and C. ẏE ẏF
à !à !à !
La : Distance between C and F . C0 −S0 J11 J12 θ̇1 + ϕ̇
2b: Distance between the two wheels. (2)
S0 C0 J21 J22 θ̇2
L1 and L2 : The lengths of arm 1 and arm 2, respectively.
Lcm1 and Lcm2 : The locations of the center of mass of
where Jij (i, j = 1, 2) are the elements of the fixed-base
arm 1 and arm 2, respectively. mw , mc , m1 , and m2 : The
Jacobian of the manipulator employed, given by J11 =
masses of the wheel, platform, arm 1, and arm 2, respec-
−L1 S1 − L2 S12 , J12 = −L2 S12 , J21 = L1 C1 + L2 C12 , and
tively.
J22 = L2 C12 . θ1 and θ2 are the joint variables of the ma-
Iw , Ic , I1 , and I2 : The moments of inertia about our
nipulator, with the notations Ci = cos(θi ), Si = sin(θi ),
centers of masses of the wheels, platform, arm 1, and arm
Cij = cos(θi + θj ), and Sij = sin(θi + θj ).
2, respectively.
Combining (1) and (2), the forward differential kinemat-
ics of the mobile manipulator is obtained as
2 Modeling of a mobile manipulator    
ẋE C0 −S0 0 0
2.1 Kinematic modeling  ẏ   S C0 0 0 
 E   0 
 = ×
 ẋF   0 0 C0 −S0 
Consider the mobile manipulator system depicted in
Fig. 1. ẏF 0 0 S0 C0
 r r r r 
− J11 + J11 J11 J12  
 r 2 2b 2 2b  θ̇R
− (d + La + J21 ) r (d + La + J21 ) J21 J22  θ̇ 
 2b 2b  L
 r r  
 0 0  θ̇1 
 2 2 
  θ̇
(d + La )r (d + La )r 2
− 0 0
2b 2b
(3)
which can be expressed in the following form:

ẋ = (RJa )υ = Jυ. (4)

Its time derivative is

˙ + J υ̇.
ẍ = Jυ (5)

2.2 Dynamic modeling

The dynamics of a mobile manipulator subject to non-
holomonic constraints can be obtained using the Lagrangian
approach in the following form:

H(q)q̈ + V (q, q̇) + AT (q)λ = E(q)(τ + τd ). (6)

The m nonholomonic constraints can be expressed as

Fig. 1 Mobile manipulator system on a differentially-driven
platform[16] A(q)q̇ = 0 (7)
10 International Journal of Automation and Computing 8(1), February 2011

where q = [q1 · · · qn ]T ∈ Rn is the generalized coordinates, For any υ, υ̇ ∈ Rp , where θ ∈ Rl is the vector of uncertain
H(q) ∈ Rn×n is a symmetric, positive definite inertia ma- (or unknown) parameters and Y ∈ Rp×l is the so-called
trix, V (q, q̇) ∈ Rn×1 represents the vector of centripetal “regressor” matrix that contains known function.
and Coriolis forces terms, A(q) ∈ Rm×n is the constraint The uncertain parameters considered for our mobile ma-
matrix, λ ∈ Rm×1 is the Lagrange multipliers that denote nipulator are the mass and the moment of inertia of the
the vector of the constraint forces, E(q) ∈ Rn×(n−m) is the system. Therefore, we have
input transformation matrix, τ ∈ R(n−m)×1 is the vector of
input torques, and τd ∈ R(n−m)×1 is the vector of unknown θ = [mc , mw , m1 , m2 , Ic , Iw , I1 , I2 ]T (14)
external disturbance. and
In order to eliminate the constraint force λ, let S(q) be a
full rank matrix (p = n − m) formed by a set of smooth and Y (q, q̇, υ, v̇) =
linearly independent vector fields spanning the null space  
of A(q), i.e., S T (q)AT (q) = 0. From (7), we can find a joint Y11 Y12 Y13 Y14 Y15 Y16 Y17 Y18
 Y 
velocity input vector υ = [θ̇R , θ̇L , θ̇1 , θ̇2 ]T for all  21 Y22 Y23 Y24 Y25 Y26 Y27 Y28 
 . (15)
 Y31 Y32 Y33 Y34 Y35 Y36 Y37 Y38 
q̇ = S(q)υ (8) Y41 Y42 Y43 Y44 Y45 Y46 Y47 Y48
where q = [xc , yc , ϕ, θR , θL , θ1 , θ2 ]T is the generalized coor- The terms of the regressor matrix in (15) are
dinates for the considered mobile manipulator system and
the matrix S(q) is defined as ∂D1 ∂D2 ∂D3 ∂D4
Y1i = , Y2i = , Y3i = , Y4i = (16)
 r  ∂θi ∂θi ∂θi ∂θi
r
(bC0 + dS0 ) (bC0 − dS0 ) 0 0 where
 2b 2b 
 r r 
 (bS0 − dC0 ) (bS0 + dC0 ) 0 0
r
X r
X
 2b 2b  D1 = H̄1i υ̇i + V̄1i υi + Ḡ1
 
S(q)= r r . (9)
 − 0 0
i=1 i=1
 2b 2b  Xr Xr
 
 0 0 1 0 D2 = H̄2i υ̇i + V̄2i υi + Ḡ2
0 0 0 1 i=1 i=1
(17)
Xr Xr

Substituting (8) and its derivative into (6) and left multi- D3 = H̄3i υ̇i + V̄3i υi + Ḡ3
plying by S T , the dynamic model can be reduced as follows: i=1 i=1
Xr Xr

H̄(q)υ̇ + V̄ (q, q̇)υ + Ḡ(q) = τ̄ + τ̄d (10) D4 = H̄4i υ̇i + V̄4i υi + Ḡ4 .
i=1 i=1

where H̄ = S T HS, V̄ = S T H Ṡ, Ḡ = S T V , τ̄ = S T Eτ , and

τ̄d = S T Eτd .
3 Adaptive controller design
We note that (10) reflects only the feasible motions al- Suppose that the desired trajectories in the task space
lowable by satisfying the constraints. are described by xd and ẋd . In order to track the desired
We assume that the external disturbances can be ex- trajectories, we define the sliding surface:
pressed as
τ̄d = F τm (11) s = ė + σe (18)

where F is a known matrix. where e = x − xd is the tracking error and σ is a constant

Using (4) and (5), the reduced dynamic system (10) can positive definite diagonal matrix. Let us define the reference
be further written in the task space as state as

Hx (q)ẍ + Vx (q, q̇)ẋ + Gx (q) = τx + τdx (12) ẋr = ẋ − s = ẋd − σe

(19)
ẍr = ẍ − ṡ = ẍd − σ ė.
−T −1 −T
where Hx = J H̄J , Vx = J (V̄ − H̄J J)J ˙−1 −1
, Gx =
J −T Ḡ, τx = J −T τ̄ , and τdx = J −T τ̄d . We now define a robust control input based on the sliding
The reduced dynamic system (10) possess a number of mode theory. Thus, we extend the sliding mode control law
important properties that facilitate analysis and control sys- given in [15], and applied for robotic manipulators in joint
tem design. space coordinates to mobile manipulators in the task space
Property 1. The inertia matrix H̄(q) is symmetric and coordinates:
positive definite, i.e., H̄ T = H̄ > 0.
˙ τ̄ = J T (Hx ẍr + Vx ẋr + Gx − Ks1 s − Ks2 sign(s)) (20)
Property 2. H̄(q) − 2V̄ (q, q̇) is skew symmetric, i.e.,
T ˙ where Ks1 and Ks2 are constant positive definite diagonal
x (H̄ − 2V̄ )x = 0, ∀x ∈ Rp .
Property 3. The left-hand side expression of (10) can matrices. The proposed control input consists of a contin-
be linearly parameterized in terms of the robot system pa- uous nominal control part, and a discontinuous switching
rameters as follows: control part to make the system invariant to parametric un-
certainty and external disturbances. The switching control
H̄(q)υ̇ + V̄ (q, q̇)υ + Ḡ(q) = Y (q, q̇, υ, υ̇)θ. (13) part causes the chattering problem. To reduce chattering,
 M. Boukattaya et al. / Robust Adaptive Control for Mobile Manipulators 11

the switching function sign(s) can be replaced by a con- as time approaches to infinite, which completes the proof.
tinuous approximation in the neighbourhood of the sliding The proposed adaptive control system is shown in Fig. 2.
surface, such as sigmoid-like function. A complete stability
analysis of the error system is given in [17].
However, the existence of external disturbances and un-
certainties influences the performance of the proposed slid-
ing mode controller. To handle these difficulties, the fol-
lowing adaptive control law is considered:
τ̄ = Y (q, q̇, υc , υ̇c )θ̂ − F τ̂m − Kev (21)
with parameter update laws:
˙
θ̂ = −ΓY T (q, q̇, υc , υ̇c )ev (22)
Fig. 2 Structure of the proposed controller
and
τ̂˙m = P F T ev (23)
where Γ and P are constant positive definite diagonal ma-
trices. ev = υ − υc is the feedforward error and υc is the 4 Simulation results
reference outer loop control design which drives the system
directly in the task space coordinates and can be obtained Let us consider the mobile manipulator system shown in
using the inverse kinematics of (5) as Fig. 1. In order to verify the effectiveness of the proposed
adaptive robust control, we compare the sliding mode con-
˙
υ̇c = J −1 (xc − Jυ) (24) troller given by (18)–(20) with the proposed adaptive con-
troller given by (21)–(24).
where xc is given by the following proportional-
The initial conditions are given as follows:
derivative (PD) controller:
xc = ẍd + Kd (ẋd − ẋ) + Kp (xd − x) (25) π π π
q(0) = [0, 0, , 0, 0, , − ]T
2 4 2
where Kp and Kd are proportional and derivative gains.
In the following, we will analyze the stability of the sys- and
tem.
Substiting (21) into (10), we have the closed-loop error υ(0) = [0, 0, 0, 0]T .
dynamics:
H̄ ėv + V̄ ev + Ḡ = Y θ̃ − F τ̃m (26) Let the desired trajectory be
where θ̃ = θ − θ̂ and τ̃m = τm − τ̂m . If an appropriate    
adaptive update law for θ̂ and τ̂m can be selected, we may xEd 0.2t + 0.3
easily prove the convergence of the tracking errors to zero  yEd   0.5 + 0.25sin(0.2πt) 
   
and the system stability. xd (t) =  = .
 xF d   0.2t 
Let us define a Lyapunov function candidate as yF d 0
1 T 1 1 T −1
L= ev H̄ev + θ̃T Γ−1 θ̃ + τ̃m P τ̃m . (27)
2 2 2 It consists of a sinusoidal path for the end-effector and a
The time derivative of L can be computed as straight line for the mobile platform.
We assume the following external disturbances applied
1 T ˙ ˙ T −1 ˙
L̇ = eT
v H̄ ėv + ev H̄ev + θ̃T Γ−1 θ̃ + τ̃m P τ̃m . (28) at each joint of the system as
2
Using the closed-loop error dynamics (26), the time deriva- τ̄d = [1.5sin(3t), 1.5cos(3t), 3sin(1.5t), 3cos(1.5t)]T .
tive of (28) becomes
L̇ = eT
v (Y θ̃ − F τ̃m − V̄ ev − Ḡ − Kev )+
The control gains used in the simulation were selected
1 T ˙ as Ks1 = 30I4×4 , Ks2 = 30I4×4 , σ = 15I4×4 , Γ = I4×4 ,
˙ T −1 ˙
ev H̄ev + θ̃T Γ−1 θ̃ + τ̃m P τ̃m . (29) P = I4×4 , Kp = 50, Kd = 50, and K = diag(0.1, 0.1, 1, 1).
2
The tracking performances of each control scheme are il-
Since (H̄˙ − 2V̄ ) is skew symmetric, the above equation be- lustrated in Figs. 3–5 and Figs. 6–8, respectively. From the
comes comparison of both controls, it can be seen that the tracking
results of the sliding mode control are not satisfactory and
−1 ˙
L̇ = eT T
v Kev + θ̃ (Γ θ̃ − Y T ėv ) + τ̃m
T
(P −1 τ̃˙m − F T ėv ). the tracking errors fluctuate greatly in comparison with the
(30) adaptive robust control schemes, which attain good con-
trol performance, and the tracking error is much smaller
˙
With selected update law θ̂ = −ΓY T (q, q̇, υc , υ̇c )ev and because of the adaptive mechanism. The simulation result
τ̂˙m = P F ev , L̇ 6 0. Thus, based on the Lyapunov stabil-
T
thus verifies the effectiveness of the proposed control in the
ity system, it can be concluded that ev converges to zero presence of external disturbances.
12 International Journal of Automation and Computing 8(1), February 2011

Fig. 3 Trajectory tracking with sliding mode control Fig. 6 Trajectory tracking with adaptive control

Fig. 7 Trajectory tracking errors for the end-effector with adap-

tive control

Fig. 4 Trajectory tracking errors for the end-effector with slid-

ing mode control

Fig. 8 Trajectory tracking errors for the platform with adaptive

control

5 Conclusion
In this paper, a robust adaptive controller for mobile ma-
nipulator system in the presence of parametric uncertain-
Fig. 5 Trajectory tracking errors for the platform with sliding ties and external disturbances was proposed. The proposed
mode control control strategy was designed to drive simultaneously in
 M. Boukattaya et al. / Robust Adaptive Control for Mobile Manipulators 13
View publication stats

the task space desired end-effector and platform trajecto- [14] M. Z. Hou, A. G. Wu, G. R. Duan. Robust output feed-
ries without violating the nonholonomic constraints. The back control for a class of nonlinear systems with input
unknown parameters and the external disturbances were unmodeled dynamics. International Journal of Automation
and Computing, vol. 5, no. 3, pp. 307–312, 2008.
estimated by using an update law in the adaptive control
[15] N. D, F. N. Hu, X. Y. An improved control algorithm for
scheme. The effectiveness of the proposed controller was high-order nonlinear systems with unmodelled dynamics.
verified by simulation. Future work will concentrate on the International Journal of Automation and Computing, vol. 6,
real implementation of the proposed controller to our mo- no. 3, pp. 234–239, 2009.
bile manipulator system.
[16] G. White. Simultaneous Motion and Interaction Force Con-
trol of a Nonholonomic Mobile Manipulator, Ph. D. disser-
References tation, University of New York at Buffalo, USA, 2006.

[1] V. Pavlov, A. Timofeyev. Construction and stabilization [17] Y. Guo, P. Y. Woo. Adaptive fuzzy sliding mode control
of programmed movements of a mobile robot-manipulator. for robotic manipulators. In Proceedings of the 42nd IEEE
Engineering Cybernetics, vol. 14, no. 6, pp. 70–79, 1976. Conference on Decision and Control, IEEE, Maui, USA,
[2] J. H. Chung, S. A. Velinsky. Modeling and control of a vol. 3, pp. 2174–2179, 2003.
mobile manipulator. Robotica, vol. 16, no. 6, pp. 607–613,
1998.
[3] K. Watanabe, K. Sato, K. Izumi, Y. Kunitake. Analysis and Mohamed Boukattaya received the
control for an omnidirectional mobile manipulator. Journal electromechanical engineering diploma
of Intelligent and Robotic Systems, vol. 27, no. 1–2, pp. 3– from the National School of Engineers,
20, 2000. University of Sfax, Tunisia in 2002. From
[4] Y. Yamamoto, X. P. Yun. Effect of the dynamic interac- the same school, he received the master
tion on coordinated control of mobile manipulators. IEEE degree in automatic and industrial infor-
Transactions on Robotics and Automation, vol. 12, no. 5, matics in 2006. He is currently a researcher
pp. 816–824, 1996. in the Control of Industrial Process Unit
[5] O. Khatib. Mobile manipulation: The robotic assis- at the same school.
tant. Robotics and Autonomous Systems, vol. 26, no. 2–3, His research interests include robot con-
pp. 157–183, 1999. trol and simulation, especially in mobile manipulators, mobile
robotics, and redundant systems.
[6] N. Hootsmanns, S. Dubowsky. The motion control of ma-
E-mail: mohamed.boukattaya@gmail.com (Corresponding au-
nipulators on mobile vehicles. In Proceedings of IEEE In-
thor)
ternational Conference on Robotics and Automatics, IEEE,
Sacramento, USA, vol. 3, pp. 2336–2341, 1991.
[7] E. Padopoulos, J. Poulakakis. Planning and model- Tarak Damak received the electri-
based control for mobile manipulators. In Proceedings of cal engineering diploma from the National
IEEE/RSJ International Conference on Intelligent Robots School of Engineers, University of Sfax,
and Systems, IEEE, Takamatsu, Japan, 2000. Tunisia in 1989, the D. E. A degree in auto-
[8] S. Lin, A. A. Goldenberg. Neural-network control of mo- matic control from the Institut National des
bile manipulators. IEEE Transactions on Neural Networks, Sciences Appliquées de Toulouse-France in
vol. 12, no. 5, pp. 1121–1133, 2001. 1990, the Ph. D. degree from the University
[9] C. Y. Lee, I. K. Jeong, I. H. Lee, J. J. Lee. Motion control Paul Sabatier de Toulouse-France in 1994.
of mobile manipulator based on neural networks and error In 2006, he obtained the University Habil-
compensation. In Proceedings of IEEE International Con- itation from the National School of Engi-
ference on Robotics and Automation, IEEE, New Orleans, neers, University of Sfax. He is currently a professor at the same
USA, pp. 4627–4632, 2004. school.
His research interests include distributed parameter systems,
[10] X. M. Tan, D. B. Zhao, J. Q. Yi, D. Xu. Adaptive hy-
sliding mode control and observers, and adaptive nonlinear con-
brid control for omnidirectional mobile manipulators using
trol.
neural-network. In Proceedings of American Control Con-
E-mail: tarak.damak@enis.rnu.tn
ference, IEEE, Seattle, Washington, USA, pp. 5174–5179,
2008.
[11] X. M. Tan, D. B. Zhao, J. Q. Yi, Z. G. Hou, D. Xu. Uni- Mohamed Jallouli received the D. E. A
fied model and robust neural-network control of omnidirec- degree from University of Valenciennes,
tional mobile manipulators. In Proceedings of the 6th IEEE France in 1986 in automatics and the Ph. D.
International Conference on Cognitive Informatics, IEEE, degree from University Paris XII, France in
California, USA, pp. 411–418, 2007. 1991, in robotics engineering. He is cur-
[12] T. T. Arif. Adaptive control of rigid body satellite. Interna- rently an assistant professor of electric and
tional Journal of Automation and Computing, vol. 5, no. 3, computer engineering at National School of
pp. 296–306, 2008. Engineers, University of Sfax, Tunisia.
His research interests include the imple-
[13] M. B. Cheng, C. C. Tsai. Hybrid sliding-mode fuzzy neural mentation of intelligent methods (neural
network tracking control for a wheeled mobile manipula- network, fuzzy logic, and genetic algorithm) in robotic and vision
tor. In Proceedings of IEEE International Conference on system as well as in multisensory data fusion mobile bases.
Industrial Technology, IEEE, Hong Kong, PRC, pp. 944– E-mail: mohamed.jallouli@enis.rnu.tn
949, 2005.