Recursive Backstepping Stabilization of A Wheeled Mobile Robot

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Mnif, F. / Recursive Backstepping Stabilization of a Wheeled Mobile Robot, pp.

287 - 294, International Journal of


Advanced Robotic Systems, Volume 1, Number 4 (2004), ISSN 1729-8806

Recursive Backstepping Stabilization of a


Wheeled Mobile Robot

Faical Mnif
Sultan Qaboos University, Department of Electrical and Computer Engineering, Oman.
mnif@squ.edu.om,
Intelligent Control, Design and Optimization of Complex Systems Research Unit, Tunisia.
faical.mnif@insat.rnu.tn.

Abstract: This research is aimed to the development of a dynamic control to enhance the performance of the existing
dynamic controllers for mobile robots. System dynamics of the car-like robot with nonholonomic constraints were
employed. A Backstepping approach for the design of discontinuous state feedback controller is used for the design of
the controller. It is shown that the origin of the closed loop system can be made stable in the sense of Lyapunov. The
control design is made on the basis of a suitable Lyapunov function candidate. The effectiveness of the proposed
approach is tested through simulation on a car-like vehicle mobile robot.
Key-Words: backstepping approach, mobile robot, chained form systems, nonholonomic systems.

1. Introduction control is ineffective, even locally, and innovative design


is required.
Wheeled mobile robots (WMRs) are increasingly present The trajectory tracking problem of WMRs was globally
in industrial and service robotics, particularly when solved in (Samson & Ait Abderrahim, 1991) by using
flexible motion capabilities are required on reasonably nonlinear feedback control, and independently in (De
smooth ground and surfaces (Scraft & Schmier, 1998). Luca & Benedetto, 1993) and (D’arendra & Bastin,
Several mobility configurations can be found in these 1995) through the use of dynamic feedback linearization.
applications. The most common are the tricycle and the Recursive backstepping control schemes for chained
car-like drive. Kinematics study of several forms of WMRs have been also addressed by several
configurations of WMRs can be found in (Alexander & authors (Jiang & Nijimar, 1999), (Tayebi et al. 1997).
Maddocks, 1989). It can be shown that the dynamic equations of a car-like
Beside the relevance in industrial applications, the vehicle mobile robot can be written in chained form as:
problem of autonomous motion planning and control of .
x1 = ν 1
WMRs has attracted the interest of researchers in view of
.
its theoretical challenges. In particular, these systems are
x2 = ν 2
typically examples of nonholonomic mechanical systems
.
(Neimark & Fufaev, 1992).
x 3 = x2ν 1 (1)
In the absence of workspace obstacles, the basic motion
tasks assigned to a WMR may be reduced to moving
between two postures and following a given trajectory. .
From a control viewpoint, the peculiar nature of x n = x n −1ν 1
nonholonomic kinematics makes the control problem
easier than the first; in fact, it is known (Campion et al. where x = ( x1 , x2 ,…, xn )T ∈ ℜ n denotes the state vector
1991), that feedback stabilization at a given posture
cannot be achieved via smooth time invariant control. and ν = (ν 1 ,ν 2 )T ∈ ℜ2 is the control input vector. Such a
This indicates that the problem is really nonlinear; linear class of nonlinear systems can not be stabilized via

287
continuous based time-invariant systems (Brockett,1993) W1 (q)q1 + W2 (q)q 2 = 0 (6)
which motivated the search of other stabilizing controls
for this type of systems. Kolmanovsky and Mc-Clamroch Assumption 2
(Klamanovsky & Mc Clamroch, 1995) stated the art of We assume that the system generalized mass matrix M(q)
existing solution for nonholonomic systems. In the , F(q) and W(q) are all independent of q1.
recent last few years many authors exhibit a particular Note that M (q) can be directly obtained from the
attention to the design of discontinuous controller for
chained systems, (Tayebi et al, 1997), (Tanner & relation L(q, q ) = 12 qT M (q )q − V (q ) , where V(q) is the
Kyriakopoulos, 2002) potential energy of the system.
In this paper we propose a systematic backstepping based
procedure for the design of a discontinuous time- Assumption 3
invariant controller for nonholonomic chained forms Assume the potential energy of the system V(q) is in the
with application to nonholonomic mobile robot systems. form
It is shown that backstepping control for chained form
V (q ) = K vT q1 + U (q 2 ) (7)
systems can guarantee the boundedness of the whole
state and makes the origin of the closed-loop system where k v ∈ ℜ n − m is a constant vector and U is a scalar
exponentially attractive provided that the initial states function.
belong to the set defined by: Under the above assumptions, the dynamics of the
Ω 0 : Ω 0 = {( x1 (0), x2 (0), … , x n (0)) T ∈ ℜ n / x1 (0) ≠ 0} (2) underactuated nonholonomic system (4)-(5) can be
The proposed control scheme is applied to the dynamics expressed as
of a car-like vehicle mobile robot which investigation d ∂L ∂L
and reduction are presented exhaustively in section 2. − = W T (q 2 )λ + F (q 2 )u (8)
dt ∂q ∂q
2. Dynamic Model and Reduction of Nonholonomic
Mechanical Systems W1 (q)q1 + W2 (q)q 2 = 0 (9)

Consider a Lagrangian system with n-dimensional To eliminate λ from (8), one can multiply both sides of
configuration vector q, a force matrix F (q) and m (m < the forced Euler-Lagrange equation in (5) by a matrix
n) nonholonomic first order constraints defined by: A(q) that annihilates W T(q), i.e. A(q)W T(q) =
0. For doing so, let us define
W T (q )q = 0 , (3)
ω 12 (q2 ) = −W1−1 (q2 )W2 (q2 ) (10)
T
that are non-integrable, i.e. there is no h(t ) : h = W (q )q then
and where W ∈ ℜ n × n is the constraint equation, then the q = D(q 2 )q 2 (11)
dynamics of the nonholonomic system can be written as where
d ∂L(q, q ) ∂L(q, q ) ⎡ω (q )⎤
− = W T (q )λ + F (q )u (4) D(q 2 ) = ⎢ 12 2 ⎥ (12)
dt ∂q ∂q ⎣ I n−m ⎦
W (q)q = 0 (5)
where I n − m is the n − m identity matrix. By direct
Where L(q, q ) is the system Lagrangian, u ∈ ℜl is the calculation, it can be readily shown that
control input, such that l = rank( F (q)) , and λ ∈ ℜ m is A(q2 ) = D T (q2 ) annihilates WT(q2). This eventually
the vector of Lagrange multipliers. Since m ≥ 1 then we leads to the following reduction theorem for
have necessarily l ≤ n − 1 . The term W T λ represents the underactuated nonholonomic systems with symmetry.
effect of the constrained forces. This is based on the
principle of virtual forces which states that the constraint Theorem 1
forces do not work on motions allowed by constraints. Consider the underactuated mechanical control system
The system (4)-(5) is called a mechanical system with with nonholonomic constraints and symmetry in (8)-(9).
first order nonholonomic constraints. Nonholonomic Then (8)-(9) with (2n + m) first-order equations can be
reduced to a system of (2n - m) first-order equations in
systems in the form (4)-(5) with F (q) ≡ 0 are called
the following cascade form.
Caplygin systems.
q x = ω r (q r )q r (13)
Assumption 1
Assume W (q) has full row rank. Then q can be M r (q r )q r + C r ( q r , q r )q r + G r ( q r ) = Fr ( q r )u (14)
partitioned as (q1,q2)T such that
T
W (q) = (W1(q), W2(q)) where W1(q) is an invertible where
matrix. Therefore the constraint equation (3) can be
rewritten as (q x , q r ) = (q1 , q 2 ), ω r (q r ) = ω 12 (q 2 ),

288
and by taking q r = q 2 , this last equation can be rewritten as
M r (q r ) = D (q 2 ) T M (q 2 ) D (q 2 )
G r (q r ) = ω rT (q r )k v + ∇ qr U (q r ) (15) M r (q r )q r + C r (q r , q r )q r + G r (q r ) = Fr (q r )u (20)
where
C r (q r , q r ) := D T (q 2 )C (q 2 , q) D(q 2 ) + D T (q 2 ) M (q 2 ) D(q 2 , q 2 )
Fr (q r ) = D T (q 2 ) F (q 2 ) M r (q r ) := D T (q 2 ) M (q 2 ) D(q 2 ) (21)
T
C r (qr , q r ) := D (q2 )C (q2 , q) D(q2 )
Moreover, if V(q) = U(q2) i.e. k v = 0 in (7), the reduced
+ D T (q2 ) M (q2 ) D(q2 , q2 ) (22)
system is a well defined Lagrangian system with
configuration vector qr = q2 and the Lagrangian
function To establish that (20) is in fact equivalent to the forced
1 Euler-Lagrange equation for the reduced system with the
L r (q r , q r ) = q rT M r (q r )q r − U (q r ) (16) Lagrangian function L r (q r , q r ) and the force
2
matrix Fr (q r ) , we need to prove that
which satisfies a forced Euler-Lagrange equation in C r (q r , q r ) satisfies M r = C r + C rT . By direct
cascade with the constraint equation as the following
calculation we have
q x = ω r (q r )q r
M r = D T MD + D T MD + D T MD
d ∂Lr ∂L r
− = Fr (q r )u (17) = D T (C + C T ) D + D T MD + D T MD
dt ∂q r ∂q r
= ( D T CD + D T MD) + ( D T C T D + D T MD)
In addition, if l + m < n or l + m = n , then the reduced = C r + C rT
system with configuration vector qr is an underactuated
(or fully actuated) mechanical system.
and the result follows.
Proof. The forced Euler-Lagrange equation in (8) can be
rewritten as 3. The Car-like Vehicle

In this section, we address reduction of the dynamic


M (q 2 )q + C (q 2 , q)q + G (q ) = W T (q 2 )λ + F (q 2 )u (18) model of a car-like vehicle as shown in Figure 1. The
Where C (q 2 , q ) satisfies dynamic model of a car is an example of underactuated
M = C + CT nonholonomic systems with five degrees of freedom, two
control inputs and two velocity constraints. Let
and G (q ) such that
q = ( x, y, θ ,ψ , φ ) denote the configuration vector of the
G (q ) = ∇ q V ( q) .
system. (x, y) denote the position of the center of the
Multiplying both sides of the last equation by axle between the two rear wheels, θ is the orientation of
the car body with respect to x-axis, ψ is the angle of
A(q 2 ) = D T (q 2 ) eliminates λ and gives
rotation of each wheel and φ is the steering angle with
respect the car body. The distance between the rear and
D T (q 2 ) M (q 2 )q + D T (q 2 )C (q 2 , q)q + G r (q) = Fr (q 2 )u
front wheels axles is denoted by l and r denotes the
(19) radius of the wheels
where Fr (q 2 ) = D T (q 2 ) F (q 2 ) and

G r (q 2 ) = D T (q 2 )G (q) = ω 12
T
(q2 ) [ I ⎢ 1 ]
⎡∇ q V ( q ) ⎤

⎣∇ q2 V (q )⎦
T
= ω 12 (q 2 )k + ∇ q2 U (q 2 ) =: G r (q 2 )

on the other hand, taking the derivative of


q = D(q 2 )q 2 implies
q = D (q 2 )q 2 + D (q 2 , q 2 )q 2

Substituting q in (19), we get


Fig.1. A car-like vehicle
T T
D (q2 ) M (q2 ) D(q2 )q2 + [ D (q2 )C (q2 , q) D(q2 )
The velocity constraints of the front and rear wheels are
+ D T (q2 ) M (q2 ) D(q2 , q2 )]q2 + G r (q2 ) = Fr (q2 )u given by

289
d d ml 2
sin(θ + φ ) ( x + l cos θ ) − cos(θ + φ ) ( x + l sin θ ) = 0 where J θ = J b + 4J v + , J h' = 2J h (2 + tan 2 (φ ))
dt dt (23)
4
sin(θ ) x − cos(θ ) y = 0 The differential-algebraic equations of motion of the
or dynamic car are such as
sin(θ + φ ) x − cos(θ + φ ) y − l cos φθ = 0
(24)
sin(θ ) x − cos(θ ) y = 0 d ∂L ∂L ⎡λ ⎤ ⎡0 ⎤ ⎡u ⎤
− = W T (q ) ⎢ 1 ⎥ + ⎢ 3×2 ⎥ ⎢ 1 ⎥ (29)
dt ∂q ∂q ⎣λ2 ⎦ ⎣ I 2×2 ⎦ ⎣u 2 ⎦
These two constraints can be rewritten as
W (q )q = 0 where W = (W1 , W 2 ) is partitioned according
where (λ1 , λ 2 ) ∈ ℜ × ℜ are the Lagrange multipliers and
to q x = ( x, y ) and q r = (θ ,ψ , φ ) . The matrices W1 and
(u1 , u 2 ) ∈ ℜ × ℜ are the torques applied to the rear
W2 are given by
wheels and the steering wheel respectively. The
dynamics of the car in (29) can be reduced to the cascade
⎡sin(θ + φ ) − cos(θ + φ )⎤ of the constraint equation and a reduced Lagrangian
W1 (θ , φ ) = ⎢
⎣ sin(θ ) − cos(θ ) ⎥⎦ system with configuration vector q = (θ ,ψ , φ ) as
and
⎡− l cos φ 0 0⎤ M r (q r )q r + C r (q r , q r )q r + G r (q r ) = Fr (q r )u (30)
W 2 (θ , φ ) = ⎢ (25)
⎣ 0 0 0⎥⎦
where
Note that W1 is not invertible at φ = 0 . M r = D T MD
Define now C r = D T CD + D T MD
Fr = D T F
⎡α 0 0⎤
ω 12 (θ , φ ) = −W1−1W2 = ⎢ 1
⎣α 2 0 0⎥⎦ By direct calculation and after simplification we get
where
⎡0 0 ⎤
l cos θ l sin θ Fr = ⎢⎢1 0⎥⎥
α 1 (θ , φ ) = , α 2 (θ , φ ) = .
tan φ tan φ ⎢⎣0 1⎥⎦
and
And D(q) can then be determined from ω12 (θ , φ ) as ⎡ J θ (φ ) 0 2J v ⎤
⎢ 2 ⎥ (31)
M r = M r (φ ) = ⎢ 0 2 J h (2 + tan (φ )) 0 ⎥
⎡α 1 0 0⎤ ⎢ 2J v 0 2 J v ⎥⎦
⎢α ⎣
⎢ 2 0 0⎥⎥ with
D(q ) = D(θ , φ ) = ⎢ 1 0 0⎥ (26) ml 2
⎢ ⎥ J θ (φ ) = J θ +
⎢0 1 0⎥ tan 2 (φ )
⎢⎣ 0 0 1⎥⎦
Clearly, the reduced Lagrangian is itself underactuated
The Lagrangian of the dynamic car is given by with three degrees of freedom (θ , φ ,ψ ) and two controls.
In addition, (θ ,ψ ) are the external variables and φ is the
1 l l
L= m[( x − sin(θ )θ ) 2 + ( y + cos(θ )θ ) 2 ] shape variable of the car. The dynamics of the actuated
2 2 2 (27)
1 2 1 1 2 1 2
variables (θ ,ψ ) of the reduced system can be linearized
+ ( J b + 2J v )θ + [2J h (1 + )]ψ + 2J v (θ + φ )
2 2 cos 2 (φ ) 2 as
ψ =τ1
where m is the mass of the car, Jb is the inertia of the (32)
body, Jh is the inertia of each wheel along the horizontal φ =τ 2
axes and Jv is the inertia of each wheel in the vertical
axes. The Lagrangian can be expressed as with τ 1 and τ 2 the new control inputs. Equation (32)
can be obtained using an explicit collocated change of
⎡ m 0 − ml / 4 sin θ 0 0 ⎤
⎢ variable in the form
0 ml / 4 cos θ mc 0 0 ⎥⎥
1 ⎢
L = q T ⎢− ml / 4 sin θ ml / 4 cos θ Jθ 0 2J v ⎥ q
2 ⎢ ⎥ u = η (φ )τ + β (φ , q r )
⎢ 0 0 0 J h' 0 ⎥

⎣ 0 0 2J v 0 2J v ⎥⎦ Where τ = [τ 1 τ 2 ]T , β a scalar function, and

290
⎡2J h (2 + tan 2 (φ )) 0 ⎤ 4. Backstepping control of the car-like vehicle
⎢ ⎥
η (φ ) = ⎢ 2J tan (φ ) ⎥ (33)
2
0 2J v (1 − 2 v In order to apply the backstepping approach procedure,
⎢ ml + J θ tan 2 (φ ) ⎥⎦
⎣ let us consider the following change of coordinates
z i = x n −i +1 , 1 ≤ i ≤ 4 . System (39) becomes
η (φ ) is well defined and positive definite for all φ . .
z 1 = z 2ν 1
.
From the first constraint equation, we can solve for θ to z 2 = z 3ν 1
get (40)
.
θ = ( x cos θ + y sin θ ) tan φ (34) z3 =ν 2
.
and the overall dynamics of the car can be written as z 4 =ν1
x = ψr cos θ
y = ψr sin θ . To guarantee the exponential convergence of z 4 , we use
r a state feedback such that
θ = ψ tan θ (35)
l
ν 1 = −k 4z4 ,
ψ =τ1
φ =τ2 where k 4 is a positive constant. System (40) becomes
.
where r is the radius of the wheel. z 1 = −k 4 z 4 z 2
After normalization of the units of (x, y) by r, and taking .
z 2 = −k 4 z 4 z 3 (41)
ψ = ω1 z3 = ν 2
we get .
x = ω 1r cos θ z 4 = −k 4 z 4
y = ω 1r sin θ
(36) Now the problem consists of finding a control law
r
θ = ω1 tan θ v 2 such that, if the initial state belongs to the set
l
φ = ω2
{ }
Ω = (z1 (0), z 2 (0), z 3 (0), z 4 (0) )T ∈ R 4 / z 4 (0) = x1 (0) ≠ 0 The
whole state of the closed loop system remains bounded
where
and converges exponentially to zero. The design
ω1 = τ 1 , ω 2 = τ 2 (37) procedure is to be done in 3 steps.

Applying the change of coordinates and control as: Step1


Consider the first equation of the system (41), where z 2
x1 = x is viewed as a virtual control variable, and consider the
tan φ Lyapunov function candidate
x2 =
l cos 3 θ 1 2
V1 = z1 (42)
x3 = tan θ 2k 4
(38) the time derivative of (42) becomes
x4 = y
.
v 1 = ω 1φ cos θ V 1 = −z1z2z4 (43)
1 + tan 2 φ 3 tan θ tan 2 φ
v2 = ω2 + 2 Using the following virtual control
l cos 3 θ l cos 3 θ
transforms finally the nonholonomic system (38) into a z
η1 = k 1 1 (44)
chained form-type system of the form z4
where k1 is a positive constant, and substitute z2 by
x1 = ν 1
η1 . Equation (43) becomes
x2 = ν 2
(39)
x 3 = x2ν 1
V1 = −k 1 z12
x 4 = x 3ν 1
Step 2
T 4
where ( x1 , … , x 4 ) ∈ ℜ are the states of the system We introduce a new variable ε 2 = z2 − η1 , which
and (ν 1 ,ν 2 ) ∈ ℜ 2 are the control inputs. represents the deviation between y 2 and the virtual

291
control η1 and consider the first two equations of (41) V 3 = V 2 − k3ε 32
where z 2 is substituted by ε 2 + η1 and z 3 is viewed as
= −(k1z12 + k2ε 12 + k3ε 22 ) (55)
a virtual control variable, then
which is negative semi-definite function.
z1 = −k 4 (ε 2 + α 1 ) Now one can easily conclude that conclude that
(45)
ε 2 = −k 4 z 3 z 4 − α 1 ( z1 , ε 2 , ε 3 ) T is bounded and tends to zero when t

Using the Lyapunov function candidate approaches the infinity. Therefore zi → ηi −1 , i = 2,3
when t → ∞ . To guarantee the boundedness and the
1 2 1 2 1 2 convergence to zero of ( z 2 , z 3 ) T , one must ensure the
V 2 =V1 + ε2 = z1 + ε2 (46)
2k 4 2k 4 2k 4 boundedness and convergence to zero of (η1 , η 2 )T .
z1
The time derivative of (41) yields Since η1 = k1 , and that z 4 is assumed to be different
z4
⎛ η ⎞ from zero, then η1 is bounded and converges to zero as
V2 = − z1 z 4 (ε 2 + η1 ) − ⎜⎜ z 3 z 4 + 1 ⎟⎟ε 2 (47)
⎝ k4 ⎠ .
k ε η
⎛ . ⎞ t → ∞ . On the other hand, η 2 = − z1 + 2 2 − 1 ,
. ⎜ η ⎟ z4 k4 z4
= V 1 − ε 2 ⎜ z3zn + 1 ⎟ − z1zn ε 2 (48)
⎜ k4 ⎟ this function is bounded and converges to zero if η1 is
⎝ ⎠
also bounded and converges to zero. η1 can be directly
By choosing the virtual control law η 2 which substitutes calculated as
z1z4 − z4z1
. η1 = k1 (56)
ε2 η1 z 24
z3 as η 2 = −z1 + k2 − , (49)
z4 k 4z 4
which is also bounded and converges to zero when
such that k 2 > 0 , equation (48) becomes t→∞ .
Finally we conclude that if z 4 (0) = x1 (0) ≠ 0 then
V 2 = V 1 − k2ε 22 = −k1z12 − k2ε 22 (50) i) The whole state remains in Ω since z 4 and then x 1
decays to zero
Step 3 ii) The state trajectory of the closed loop system is
Likewise in step 2 we introduce a new variable ε 3 such bounded and converges to zero
that: Now we can summarize the previous procedure in the
ε 3 = z3 − η 2 (51) following theorem
and using the Lyapunov function candidate
Theorem 1
1 2 Consider the system (39) with the control law
V 3 =V 2 + ε3 (52)
2k 4
⎧− k 4 x1 for x1 (0) ≠ 0
ν1 = ⎨ *
Evaluating the time derivative of (52), gives ⎩ν 1 for x1 (0) = 0

⎧⎪ .
for x1 (0) ≠ 0
ε3 k ε x − k
ν2 = ⎨ 4 2 1 4 3 3 k ε + η 2
V 3 = V 2 − k 4ε 2ε 3z4 + (ν 2 − η 2 ) (53) for x1 (0) = 0
k4 ⎪⎩ ν 2*

Under the following control law v 2 in (41), defined where ν 1* ∈ ℜ /{0}, ν 2* ∈ ℜ /{0} are two constant
over Ω , as values and where
.
ν 2 = −η 2 − k 4k3ε 3 + k 4z4ε 2 (54) z ε η
η1 = k1 1 and η 2 = − z1 + k 2 2 − 1
z4 z 4 k4 z4
where k 3 is a positive parameter and η 2 can be
evaluated from the expression of η 2 , equation (53) with k i > 0 , i = 1, ,4
becomes then

292
i) the whole state z = (z1 , … , z4 )T and equivalently θ and φ. It is clear that the discontinuous backstepping
approach presented in this paper guarantees the
x = (x1 , … , x 4 )T is bounded and decays to zero when convergence and the stabilization of all the states of the
t →∞, system.
ii) the control law is well defined for all t ≥ 0 . 2

1.8
5. Simulation Results
1.6

The technique presented in this paper is tested on the 1.4

system 1.2
x = v cos θ

y (m )
1
y = v sin θ
0.8
v
θ = tan θ 0.6
l
0.4
φ =ω
0.2

where v is the constant linear velocity of the car and l is 0


0 2 4 6 8 10 12 14 16 18 20
the distance between the front and rear wheels axles. t (sec)

Results are depicted in Figures 2-6. The Initial conditions Fig. 4. Time history of the variable y.
are taken as 1.4

x(0) = [ x0 , y 0 , θ 0 , φ 0 ]T = [1.5,2, π4 ,− π2 ]T 1.2

and the following gains are used 1

0.8
k1 = 3, k 2 = 2 , k 3 = 0.5 and k 4 = 0.5 .
theta (rad)

0.6

x-y
5 0.4

4.5
0.2
4

3.5
0
0 2 4 6 8 10 12 14 16 18 20
3
t (sec)
2.5
Fig. 5 Time history of the variable θ
y

1.5

1
0.2
0.5

0 0

-0.5 -0.2
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4
x
-0.4
Fig. 2. Path to the origin of the mobile robot
-0.6
1.5
phi (rad)

-0.8

-1

-1.2
1
-1.4
x (m )

-1.6

-1.8
0.5 0 2 4 6 8 10 12 14 16 18 20
t (sec)

Fig. 6 Time history of the variable φ.

0
0 2 4 6 8 10 12 14 16 18 20 6. Conclusions
t(sec)

Fig. 3 Time history of the variable x. This paper presents a novel technique for backstepping
discontinuous control with application to the stabilization
Figure 2 shows the path traversed by the mobile robot. of a nonholonomic mobile robot. The methodology is
Figures 3-6 show the time behavior of the variable, x, y, not restricted, though to nonholonomic systems but can

293
be applied to a broad class of strictly feedback nonholonomic mobile robots. Proc of the IEEE
discontinuous systems. The proposed control scheme is ICRA, pp. 3948-3953, 2002.
smooth everywhere except at x1 = 0 .The discontinuity Tayebi, A. Tadjine, M. and Rachid, A. (1997),
involved in the control is not very restrictive since it Discontinuous control design for the stabilization of
occurs just for x1 (0) = 0 . The simulation results made on nonholonomic systems in chained form using the
backstepping approach. Proc of the 36th IEEE CDC,
a car-like mobile robot demonstrate the validity of the
pp. 3089-3090.
proposed control.

7. References

Alexander, J.C., and Maddocks, J.H., (1989), On the


Kinematics of wheeled mobile robot, International
Journal of Robotics Research. 8, 5, pp. 15-27.
d’Andrea-Novel, B., Bastin, Campion, G. (1995),
Control of nonholonomic wheeled mobile robots by
state feedback linearization. International Journal of
Robotics Research, 14, 6, pp. 543-559.
Brockett, R.W. 1993: Asymptotic stability and feedback
stabilization. Progress in Math., vol. 27, Birkhauser,
pp. 181-208, 1993.
Campion, G., d’Andrea-Novel, B, and Bastin, G. (1991),
Modeling and state feedback control of
nonholonomic mechanical systems. Proc. of the 30th
IEEE Conference on Decision and Control,
Brighton, UK, pp. 1184-1189.
De Luca, A. and Di Benedetto, M.D. (1993), Control of
nonholonomic systems via dynamic compensation.
Kybernetica, 29, 6.
Jiang, Z-P, Nijimer, H., (1999), A recursive technique for
tracking control of nonholonomic systems in chained
form. IEEE Transactions on Automatic Control, 44,
2, pp. 265-279.
Klomanovsky, I. and Mc-Clamroch, N.H.: 1995
Developments in nonholonomic control problems,
IEEE Control System magazine. 15, 6, pp. 20-36.
Mnif, F., (2003), On the reduction and Control of a class
of nonholonomic underactuated systems”, Journal
of Electrical Engineering, 54, 1-2, pp. 22-29.
Mnif, F., (2004), An Adaptive Control Scheme for
Nonholonomic Mobile Robot with Parametric
Uncertainty, 10th International Symposium on
Robotics and Applications - World Automation
Congress, WAC-ISORA 2004, Spain.
Neimark,, J.I., and Fufaev, F.A., (1972), Dynamics of
Nonholonomic Systems. American mathematical
Society, Providence, RI.
Schraft, R.D. and Schmierer, G. 1998: Serviceroboter.
Springer Verlag.
Samson, C., and Ait-Abderrahim,K, (1991), Feedback
control of a nonholonomic wheeled cart in Cartesian
space, Proc. of the IEEE International Conference on
Robotics and Automation, Sacramento, CA, pp.
1136-1141.
Tanner, H. G. and Kyriakopoulos, K. (2002),
Discontinuous Backstepping for stabilization of

294

You might also like