International Journal of Research in Engineering and Science (IJRES)

ISSN (Online): 2320-9364, ISSN (Print): 2320-9356

www.ijres.org Volume 3 Issue 10 October. 2015 PP.36-42

Semi-global Leaderless Consensus with Input Saturation

Constraints via Adaptive Protocols
Zhiyang Wu1, Jian Li1, Fang Song1,2
1(Laboratory of Intelligent Control and Robotics, Shanghai University of Engineering Science, Shanghai,
201620, P. R. China)
2( State Key Laboratory of Robotics and System, Harbin Institute of Technology, Harbin, 150001, P. R. China )
ABSTRACT: In this paper, the problems of semi-global leaderless consensus for continuous-time multi-agent
systems under input saturation restraints are investigated. We consider the leaderless consensus problems
under fixed undirected communication topology. New necessary synthesis conditions are established for
achieving semi-global leaderless consensus via the distributed adaptive protocols, which is designed by utilizing
low gain feedback tactics and the relative state measurements of the neighboring agents. Finally, simulation
results illustrate the theoretic developments.
Keywords -Semi-global consensus; Multi-agent systems; Inputsaturation; Adaptive protocols

I.

INTRODUCTION

During the last few decades, consensus has been became a very hot research topic inmany scientific fields,
such as computer science, physics, biology,and system control. Roughly speaking, consensus means the
agreementof agents on the joint status value via utilizing local interactionfrom the neighbouring agents. In
recent years, extensive research attentionhas been paid into the consensus problem of multi-agent systems[1-5].
It should be noted that, although thepower of each individual agent is quite limited, the wholeinteraction
network system may accomplish complex tasks in acoordinated fashion. Therefore, the consensus problem has a
widerange of potential applications, to mention just a few, includingcooperation of unmanned air vehicles
(UAVs) in [6],distributed sensor networks in [7] and [8],attitude alignment for a flock of satellites in
[9],flocking control in [10] and [11], congestion control under communication topology [13].Interested readers
are referred to the recent survey articles [14], [15] and the references citedtherein for details.It is known that
consensus has been studied for several decades in computer science. Broadlyspeaking, there are two classes of
consensus, that is, the leaderless consensus and the leaderfollowing one. Usually, the multi-agent system
dynamic models are described by double-integratordynamics or single-integrator kinematics [13]-[17]. In recent
literature, the general linear dynamicmodel has been considered to address more complicated system dynamics
in engineering practice.For instance, in [18] the authors studied the consensus problems of both leaderless and
leader-following cases in the situation of fixed interaction networks. In [19] the consensus problemsin the case
of leader-following were discussed under a switching communication network. Theleaderless nonlinear
consensus subject to parametric uncertainties was achieved by developingadaptive continuous finite-time
distributed protocols in [20]. The consensus of multi agents withsecond-order, linear and Lipschitz nonlinear
dynamics via utilizing the adaptive algorithms werestudied in [21] and [22], respectively. A distributed observer
based algorithm was proposed in [23]for leader-following consensus of networked multiple agents systems.
Indeed, up to now, the investigation on consensus has many comprehensive achievements.However, in
practice networked multi-agent systems may have more complicated dynamics. Theindustrial applications of
consensus are limited by all kinds of physical conditions, such as the inputsaturation constraints. Till now, more
and more research activities has paid close attention tothe consensus problems for networked multiple agents
systems under input saturation constraintsbecause of its unique characteristic and practical applications. In [24],
the authors investigated theglobal consensus of linear networked multiple agents dynamics with input saturation
by distributedobserver-based algorithm over fixed and switching interaction networks. For discrete-time
systems,the global consensus problem under input saturation constraints was addressed in [25]. In [26] Suet al.
proposed a linear low gain feedback algorithm to reach semi-global consensus for networkedmulti agents
dynamics under the assumption of agents are linear asymptotically null controllablewith bounded controls
(ANCBC) [27]. While in recent paper [28], the semi-global consensus oflinear dynamics in the case of input
saturation was studied by utilizing observer-based algorithmsand the low gain feedback technique, under the
assumptions that the communication networkswere jointly connected or connected and each agent was
detectable. With the input saturationrestricts, the semi-global synchronization problem for multiple agents with
linear systems wasstudied in [29].
In the present paper, we revisit the problem of leaderless consensus of linear multi agentsdynamics with
input saturation constraints. New semi-global consensus conditions are establishedby using distributed adaptive

www.ijres.org

36 | Page

Semi-global Leaderless Consensus with Input Saturation Constraints via Adaptive Protocols
protocols. Compared with some existing results, the distinctivepoints of the paper can be expressed as the
following sentences. Under the assumption thatthe agents are ANCBC, without realizing any information of the
networks, all agents achievean agreement by utilizing the distributed adaptive protocols and using the low gain
feedbacktechnique over both fixed and switching interaction networks. Also, we provide simulation resultson
two numerical examples to verify the theoretical developments.
The remainder of this paper is arranged as follows. Preliminaries and notations are introducedin Section 2.
In Section 3, based on the distributed adaptive protocols, we address the problemof semi-global leaderless
consensus under fixed undirected graph.The simulation results on two numericalexamples are given in Section 4
to verify the theoretic developments. Finally, a brief conclusionis drawn in Section 5.

II.

PROBLEM BACKGROUND AND PRELIMINARIES

In this paper, we suppose that the information exchange between agents and the leader is described by a

fixed undirected graph G (V , E, A), with the set of vertices V 1, 2,..., N whose elements represent
agents of the group, the set of undirected edges E V V represent neighboring relations among agents, and a
weighted adjacency matrix A [aij ] R

nn is defined as a 0 if and only if

( j, i) E ; otherwise,
ij

aij 0 . Here, we also assume that aij a ji for all i 1, 2,..., n .

For an undirected graph G , a Laplacian matrix L [l ] Rnn of graph G is defined as
ij
with the degree matrix

D diag d1, , dN is defined as i th diagonal elements

N
j 1, j i

L D A,
aij (t ) . Define

B0 diag (b10 (t ), b20 (t ),, bN 0 (t )) as bi 0 0 if the leader is available by agent i and bi 0 0 otherwise.
The eigenvalues of L can be ordered as 1 2 N . Then, 1 0 with its corresponding right
eigenvector 1 [1,1,...,1] R . Moreover, 2 0 if G is a connected graph.
T

Assume that M is a matrix, the notation M T denotes its transpose matrix. A symmetric positive matrix
M indicates that all its eigenvalues are positive, a positive semi-definite M indicates that all its eigenvalues
are non-negative. The symbol is the Kronecker product. I n denotes the identity matrix of dimension

N N.
Lemma 1 [4]:Denote L as the Laplacian of a fixed undirected graph G include N agents, denote L as
the Laplacian of the graph consisting of N agents and one leader, bi 0 0 if there is a spanning tree with the
leader as the root vertex.
Lemma 2 [26]: For any

R mm and i , i R m , i 1, 2,..., N

N N
1 N N
T
a
(
t
)(

)
(

)(

))

aij (t ) iT ( ( )( i j ))

ij
i
j
i
j
2 i 1 j 1
i 1 j 1

In this paper, we consider an MAS of N identical continuous-time agents subject to input saturation
constraints, moving in an n -dimensional Euclidean space. The dynamic of each agent is modeled by

xi Axi B (ui )

yi Cxi , i 1, 2,, N

(2)

where xi R , yi R are the state of agent i and the measurement output of agent i respectively,
n

ui R m
as

is

the

control

: Rm Rm

input,

(ui ) sgn(uij ) min uij ,

with

is

a
scalar

saturation

function

value

defined

0.

For

ui (u1 , u2 , , um ) T , (ui ) ( (u1 ), (u2 ), , (um )) T .

To address the leaderless consensus, the average state of

1
xi
N

x
i 1

N agents is defined as follows

, i 1, 2,..., N

www.ijres.org

(3)

37 | Page

Semi-global Leaderless Consensus with Input Saturation Constraints via Adaptive Protocols
Definition 1. To address the semi-globalleaderless consensus problem, it means that, for any prior given
compact set R , we can design a distributed control law uiby only using the partial information of agent i
from neighbors, such that all the agents can reach the same state asymptotically, i.e.,
n

lim xi (t ) x j (t ) 0 , i 1, 2,, N
t

with any initial condition xi (0) , i 0,1, 2,, N

In order to employ the low gain feedback technique, we need to suppose that the pair (A,B)satisfies the
following Assumption 1. For more details, the reader can refer to the book [27].
Assumption 1: The pair ( A, B) is ANCBC and each eigenvalue of the matrix A is on theimaginary axis
simple.
Lemma 3[29]: Under Assumption 2. A unique solution P ( ) is derived by the following parametric ARE.

AT P( ) P( ) A P( ) BB T P( ) P( )
where

(4)

0 is a positive scalar. Furthermore, we have lim P 0.

0

Lemma 4[29]: Let Assumption 1 hold, for any 0 and

(0,1] , the unique solution in Lemma 3 is

P( ) W ( ) which satisfies a Lyapunov function

1

W (A

I n )T ( A

I n )W BB T

(5)

Assumption 2:The graph G is connected at all time.

III.

MAIN RESULTS

This section investigates the semi-global leaderless consensus with input saturation under a fixedinteraction
network topology. The adaptive control algorithm will be employed to address the problem.Now, motivated by
the distributed consensus protocol with an adaptive control law in [22], we are ready to design a distributed
consensus algorithm as follows:
T
ui K aij mij xi x j m ij ij aij xi x j K xi x j
N

(6)

j 1

ij ji are the plus quantities, mij is the adaptive coupling weights, the feedbackmatrix K
T
T
and K are designed as K B P ( ) and K P ( ) BB P ( ) respectively, P ( ) is theunique
where

solution that can be obtained from the algebra Riccati equation (4).
By introducing the above contents, now we propose the results as follows.
Theorem1. Suppose that Assumptions 1 and 2 are satisfied. Then the multi-agent system(2) of N agents achieve
semi-global leaderless consensus under the distributed control law (6).
Proof:

Let us define

i 1T , 2T ,..., NT

with

i xi xi . Then, we have the following equations

i xi xi A i B (ui )

1
N

B
i 1

(ui )

T
m ij ij aij i j K i j

(7)

Now, for the multi-agent system (2), let us consider a common Lyapunov function candidate
N

i 1

i 1 j 1, j i

V iT P i
The matrix

ij

(8)

2 ij

P is the unique solution from the parametric algebra Riccati equation (6), is a

positive parameter to be designed. By way of convenient notation, define c be a positive number such that

www.ijres.org

38 | Page

Semi-global Leaderless Consensus with Input Saturation Constraints via Adaptive Protocols
2

N
mij

c
sup
i P

2ij
0,1, xi 0 ,i 1,2,..., N i 1
j 1, j i

Such a c exists since

(9)

lim P 0 . Let

is bounded and

c : R : V c
Nn

and

(10)

(0,1] be such that, for each (0, ] , e c implies that

*

B T P( ) mij aij xi x j
N

, i 1, 2,..., N

j 1

(11)

(0, * ] , for any e c , the derivative of V can be evaluated as

Therefore, by Lemma 3, consider any

V iT P( ) i iT P( )i
N

i 1

i 1

i 1 j 1

mij

ij

m ij

iT ( P( ) A AT P( )) i 2 iT P ( ) BBT P ( ) mij aij i j

N

i 1

i 1

j 1

ij

ij

i 1 j 1

ij aij i j P BBT P i j
T

(12)

i R , i R , i 1, 2,..., N , we have
1 N N
aij (t )( i j )T ( i j ) T ( L(t ) I m )
2 i 1 j 1
m

By Lemma 2, for any

where

1 2 N

and 1 2 N

. Therefore, we can continue (12) as follows:

V iT P( ) i iT P( )i 2 aij iT P ( ) BBT P ( ) i j
N

i 1

i 1

i 1 j 1

( I N ( P( ) A A P ( ) 2 L P( ) BB T P( ))
Since the graph is connected, zero is a simple eigenvalue of L and all the other eigenvalues are positive. Let
U R N N be such a unitary matrix that U T LU : diag 0,2 ,...,N . Because the right and left
T

L corresponding to the zero eigenvalue are 1 and 1T , respectively, we can choose

1T

Y1 and U T N , with Y1 R N N 1 and Y2 R N 1N . Let

Y2

eigenvectors of

1
U
N

1T ,..., NT U T I n . Then we have

V T ( I N ( P( ) A AT P( ) 2 L P ( ) BB T P ( ))
T

iT P( ) A AT P( ) 2i P( ) BB T P( ) i
N

i 2

By Lemma 3 choosing

2i , i 2,..., N , (13)

(13)
can be continued as :

V iT P( ) A AT P( ) 2 P( ) BB T P ( ) i iT P i 0
N

i 2

i 1

Thus, semi-global consensus is achieved, i.e., such that for a set of original condition

www.ijres.org

xi 0
39 | Page

Semi-global Leaderless Consensus with Input Saturation Constraints via Adaptive Protocols

lim xi (t ) x0 (t ) 0 , i 1, 2,..., N
t

This completes the proof.

IV.

SIMULAION STUDY

Herein the simulation experiments are supplied to clarify the theoretic results on the leaderlessconsensus
developed in Section 3. Consider a multi agents system consisting of eight agents.Assume that the dynamic of
every agent holds the form of (2) with the following parameters:

Obviously the pair

A, B

1 1
0
A
,
B

1
0 1

satisfies Assumption 1. Suppose the topology is considered as in Fig.1. The initial

states of eight agents are randomly assigned. Thus, We chose

obtained by using an exhaustive method. Then

ij = 1 and = 5, 2 = 0.2907 can be

22 2.907 . For the situation of 0.2 , then we

get a set of solutions as

0.3429 0.3086
P 0.2

0.3086 0.5556

K and K as

Thus, the feedback gain

0.0952 0.1715
K 0.3086 0.5556 , K

0.1715 0.3087
1

6
4

Fig. 1: The interaction communication network.

Adaptive weights ( = 0.2)

State Convergence ( = 0.2)

16

10

14

x i1

5
12

5
t

10
mij

-5

10

10

6
4
2

x i2

0
-2

-5

5
t

10

-4

5
t

10

Fig. 2: The states of eight agents.Fig. 3: The adaptive weights mij .

Under the control law (6), the trajectories of the state difference between agents are shown in Fig. 2. It is
obvious that the semi-global leaderless consensus can be achieved.
For the situation of 0.2 , we can get the unique solution

www.ijres.org

40 | Page

Semi-global Leaderless Consensus with Input Saturation Constraints via Adaptive Protocols

0.4883 0.3906
P 0.4

0.3906 0.6250
State Convergence ( = 0.4)

Adaptive weights ( = 0.4)

x i1

0
4

-2
3

5
t

10
mij

-4

5
1

x i2

-5

-10

-1

5
t

10

-2

5
t

10

Fig. 2: The states of eight agents.Fig. 3: The adaptive weights mij .

Thus, the feedback gain

K and K as

0.1526 0.2441
K 0.3906 0.6250 , K

0.2441 0.3906
Under the control law (6), it is obvious that the semi-global leaderless consensus can be achieved. The
simulation results are in Fig. 4. The adaptive weights mij under 0.2 and 0.4 are shown in Fig. 3
and Fig. 5 respectively.

V.

CONCLUSION

In this paper, we consider the semi-global leaderless consensus problem of ANCBC multipleagents system
subject to input saturation restraints both over fixed and switching topology. Theadaptive control law is
proposed by utilizing a low gain feedback strategy. The connectivity of thegraph in fixed networks and the joint
connectivity of the graph in switching networks are the keyrequirements to guarantee the semi-global leaderless
consensus. Finally, the theoretical resultswere verified by simulation study.

VI.

Acknowledgements

This work was partly supported by the State Key Laboratory of Robotics and System (HIT) underGrant
SKLRS-2014-MS-10, the Jiangsu Provincial Key Laboratory of Advanced Robotics Fund Projects under Grant
JAR201401, and the Foundation of Shanghai University of Engineering Science under Grant nhky-2015-06,
14KY0130.

REFERENCES
[1].
[2].
[3].
[4].
[5].
[6].
[7].
[8].
[9].

R. Olfati-Saber, and R.M. Murray, Consensus problems in networks of agents with switchingtopology and time-delays,
IEEETransactions on Automatic Control, vol. 49, no. 9, pp. 1520-1533, 2004.
Jadbabaie, J. Lin, and A.S. Morse, Coordination of groups of mobile autonomous agentsusing nearest neighbor rules, IEEE
Transactions on Automatic Control, vol. 48, no. 6, pp.988-1001, 2003.
L. Wang and F. Xiao, Finite-time consensus problems for networks of dynamic agents,IEEE Transactions on Automatic
Control, vol. 55, no. 4, pp. 950-955, 2010.
W. Ren and R.W. Beard, Consensus seeking in multi-agent systems under dynamicallychanging interaction topologies, IEEE
Transactions on Automatic Control, vol. 50, no. 5,pp. 655-661, 2005.
F. Pasqualetti, D. Borra, and F. Bullo, Consensus networks over finite fields, Automatica,vol. 50, no. 2, pp. 349-358, 2014.
D.J. Pack, P. DeLima, G.J. Toussaint, and G. York, Cooperative control of UAVs for localization of intermittently emitting
mobile targets, IEEE Transactions on Systems, Man, andCybernetics, PartB: Cybernetics, vol. 39, no. 4, pp. 959-970, 2009.
J. Cortes and F. Bullo, Coordination and geometric optimization via distributed dynamicalsystems, SIAM Journal on Control
and Optimization, vol. 44, no. 5, pp. 1543-1547, 2005.
R. Olfati-Saber and P. Jalalkamali, Coupled distributed estimation and control for mobilesensor networks, IEEE Transactions
on Automatic Control, vol. 57, no. 10, pp. 2609-2614,2012.
H. Septanto, B.R. Trilaksono, A.R. Syaichu and R.E. Poetro, Consensus based controllersfor spacecraft attitude alignment:
Simulation results, In International conference on instrumentation, communications, information technology, and bio medical
engineering, pp. 52-57,2011.

www.ijres.org

41 | Page

Semi-global Leaderless Consensus with Input Saturation Constraints via Adaptive Protocols
[10].
[11].
[12].
[13].
[14].
[15].
[16].
[17].
[18].
[19].
[20].
[21].
[22].
[23].
[24].
[25].
[26].
[27].
[28].
[29].

R. Olfati-Saber, Flocking for multi-agent dynamic systems: Algorithms and theory, IEEETransactions on Automatic Control,
vol. 51, no. 3, pp. 401-420, 2006.
K.D. Do, Flocking for multiple elliptical agents with limited communication ranges, IEEETransactions on Robotics, vol. 27,
no. 5, pp.931-942, 2011.
P. Ignaciuk and A. Bartoszewicz, Discrete time sliding mode congestion control in multisource communication networks with
time varying delay, IEEE Transactions on ControlSystems Technology, vol. 19, no. 4, pp. 852-867, 2011.
W. Ren, On consensus algorithms for double-integrator dynamics, IEEE Transactions onAutomatic Control, vol. 53, no. 6, pp.
1503-1509, 2008.
R. Olfati-Saber, J. Fax, and R.M. Murray, Consensus and cooperation in networked multi-agent systems, Proceedings of the
IEEE, vol. 95, no. 1, pp. 215-233, 2007.
W. Ren, R.W. Beard, and E.M. Atkins, Information consensus in multivehicle cooperativecontrol, IEEE Control System
Magazine, vol. 27, no. 2, pp. 71-82, 2007.
S.H. Li, H.B. Dua, and X.Z. Lin, Finite-time consensus algorithm for multi-agent systemswith double-integrator dynamics,
Automatica, vol. 47, no. 8, pp. 1706-1712, 2011.
Abdelkader and T. Abdelhamid, On consensus algorithms for double-integrator dynamicswithout velocity measurements and
with input constraints, Systems & Control Letters, vol.59, no. 12, pp. 812-821, 2010.
P. Wieland, J.S. Kim, and F. Allgower, On topology and dynamics of consensus among linearhigh-order agents, International
Journal of Systems Science, vol. 42, no. 10, pp. 1831-1842,2011.
W. Ni and D.Z. Cheng, Leader-following consensus of multi-agent systems under fixed andswitching topologies, Systems &
Control Letters, vol. 59, no. 3, pp. 209-217, 2010.
J.S. Huang, C.Y. Wen, and W. Wang, Adaptive finite-time consensus control of a group ofuncertain nonlinear mechanical
systems, Automatica, vol. 51, pp. 292-301, 2015.
H.S. Su, G.R. Chen, X.F. Wang, and Z.L. Lin, Adaptive second-order consensus of networkedmobile agents with nonlinear
dynamics, Automatica, vol. 47, no.2, pp. 368-375, 2011.
Z.K. Li, W. Ren, X.D. Liu, and M.Y. Fu, Consensus of multi-agent systems with generallinear and lipschitz nonlinear dynamics
using distributed adaptive protocols, IEEE Transactions on Automatic Control, vol. 58, no.7, pp.1786-1791, 2011.
Y.G. Hong, G.R. Chen, and L. Bushnell, Distributed observers design for leader-followingcontrol of multi-agent networks,
Automatica, vol. 44, no. 3, pp. 846-850, 2008.
Z.Y. Meng, Z.Y. Zhao, and Z.L. Lin, On global leader-following consensus of identical lineardynamic systems subject to
actuator saturation, Systems & Control Letters, vol. 62, no.2,pp. 132-142, 2013.
T. Yang, Z.Y. Meng, D.V. Dimarogonas, and K.H. Johansson, Global consensus for discrete-time multi-agent systems with
input saturation constraints, Automatica, vol. 50, no.2, pp.499-506, 2014.
H.S. Su, Michael Z.Q. Chen, J. Lam, and Z. L. Lin, Semi-global leader-following consensus oflinear multi-agent systems with
input saturation via low gain feedback, IEEE Transactionson Circuits and Systems, vol. 60, no. 7, pp.1881-1889, 2013.
Z.L. Lin, Low gain feedback, Lecture Notes in Control and Information Sciences, London,U.K.: Springer-Verlag, vol. 240,
1998.
H.S. Su, Michael Z.Q. Chen, X.F. Wang, and J. Lam, Semi-global observer-based leader-following consensus with input
saturation, IEEE Transactions on Industrial Electronics, vol.61, no. 6, pp. 2842-2850, 2014.
Q.L. Wang, C.B. Yu, and H.J. Gao, Semiglobal synchronization of multiple generic linearagents with input saturation,
International Journal of Robust and Nonlinear Control, vol.24, no. 18, pp. 3239-3254, 2013.

www.ijres.org

42 | Page