This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING 1

Cooperative Control of Multi-Agent Systems With

Unknown State-Dependent Controlling Effects
Peng Shi, Fellow, IEEE, and Qikun Shen

Abstract—This paper investigates the cooperative control well known that many practical systems are modeled by such
problem of uncertain high-order nonlinear multi-agent systems high-order systems. Hence, the cooperative control problem of
on directed graph with a fixed topology. Each follower is assumed such high-order MASs distributed on communication networks
to have an unknown controlling effect which depends on its own
state. By the Nussbaum-type gain technique and the function ap-
become necessary, which motivates us for this work.
proximation capability of neural networks, a distributed adaptive It is worth noticing that most of the existing results have been
neural networks-based controller is designed for each follower in obtained for the specific systems as in [19], where the control-
the graph such that all followers can asymptotically synchronize ling effect of each follower is equal to one or other constant. In
the leader with tracking errors being semi-globally uniform ulti- fact, the general problem is
mate bounded. Analysis of stability and parameter convergence of
the proposed algorithm are conducted based on algebraic graph
theory and Lyapunov theory. Finally, a example is provided to
validate the theoretical results.
where and are the state and input of agent , and de-
Note to Practitioners—Many practical applications can be mod- notes external disturbance. The first term describes the
eled as uncertain high-order nonlinear multi-agent systems, whose individual dynamics of agent ; means that the control-
node’ controlling effects are state-dependent. In most relevant lit- ling effect of agent is dependent of its own state [22]. In prac-
eratures, however, it is often assumed that each node’ controlling
effects are equal to one. How to cooperative control for the systems tical applications, many systems can be modeled by the above
has become one main focus of control researches. Therefore, in this form. For example, the position of each robot influences its own
paper, an adaptive cooperative control scheme is proposed for such controlling effect in the formation problem with multiple robots
multi-agent systems. Finally, the effectiveness of the control strate- [23]. [23] studied formation control for a multi-agent system
gies is illustrated via simulation study. which was given by mobile robots and the state of each robot
Index Terms—Cooperative control, distributed control, multi- influenced its own controlling effect. The systems whose con-
agent systems. trolling effects depends on the state of the system is also consid-
ered in [24] and [25]. Reference [22] investigated the leader-fol-
lowing consensus problem of multi-agent systems where the
I. INTRODUCTION
leader was static and the controlling effect of each follower de-

C OOPERATIVE control of multi-agent systems (MASs)

has attracted extensive attention from the control com-
munity for the last two decades and obtained fruitful results
pended on its own state. However, the controlling effect was as-
sumed to be known. It is a challenging to handle the leader-fol-
lowing consensus problem of the high-order uncertain nonlinear
[1]–[8], [30]–[36]. In the study of MASs, the leader-following MASs with unknown state-dependent controlling effect, which
tracking problem has sparkled great research interest [15]. How- motivates us for this work.
ever, most of the existing works focused only on linear MASs, In this paper, we investigate the cooperative control problem
and first- or second-order nonlinear MASs [9]–[20]. High-order of unknown high-order nonlinear MASs, and propose a
nonlinear MASs have not been fully studied. Furthermore, it is distributed adaptive neural-networks-based control scheme
such that all followers can asymptotically synchronize the
Manuscript received September 03, 2014; revised December 18, 2014; leader with tracking errors are semi-globally uniform ultimate
accepted January 17, 2015. This paper was recommended for publication bounded (SGUUB). Compared with the existing works in
by Associate Editor D. Sofge and Editor L. Sabattini upon evaluation of the
literature, the contributions from our work are worth to be
reviewers' comments. This work was supported in part by the National Natural
Science Foundation of China under Grant 61174046 and Grant 61473250, in emphasized.
part by the Australian Research Council under Grant DP140102180 and Grant 1) The controlling effect of each follower considered in this
LP140100471, in part by the 111 Project under Grant B12018, and in part by paper is assumed to be the function of its own state. Fur-
the Natural Science Foundation of Jiangsu Higher Education Institution under
Grant 14KJB120013. thermore, not only its value but also its sign is unknown. In
P. Shi is with the College of Automation, Harbin Engineering University, Hei- order to handle the complete controlling effect, the Nuss-
longjiang 150001, China, and with the School of Electrical and Electronic Engi- baum-type gain technique is first extended to the coopera-
neering, University of Adelaide, SA 5005, Australia, and also with the College
of Engineering and Science, Victoria University, VIC 8001, Australia (e-mail: tive control problem of MASs.
peng.shi@Adelaide.edu.au). 2) Compared with the results in [15]–[20], the dynamic leader
Q. Shen is with the College of Information Engineering, Yangzhou Univer- and all followers considered in this paper are high-order
sity, Yangzhou 225127, China (e-mail: qkshen@yzu.edu.cn).
Color versions of one or more of the figures in this paper are available online
and have unknown nonlinear dynamics. In addition, de-
at http://ieeexplore.ieee.org. signing of the proposed cooperative adaptive controllers
Digital Object Identifier 10.1109/TASE.2015.2403261 do not require the assumption that the upper bounds of the

1545-5955 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING

leader node's nonlinear dynamics, optimal approximation Define

errors and external disturbance should known exactly as in
[19].
The rest of this paper is organized as follows. In Section I,
basic graph theory and notations, the problem formulation,
the Nussbaum-type gain and mathematical description of
neural-networks (NNs) are introduced. In Section III, a dis- then the overall graph dynamics is re-written in the following
tributed adaptive neural controller is proposed for each follower compact form:
in the graph. A numerical simulation is presented in Section IV.
These simulation results demonstrate the effectiveness of the
(2)
proposed technique. Finally, Section V draws the conclusion.

II. PRELIMINARIES AND COOPERATIVE Time-varying dynamics of the leader node, labeled 0, is de-
CONTROL FORMULATION scribed as follows:
In this section, basic graph theory and some notations are
(3)
first introduced. Then, the cooperative control problem of multi-
agent system is formulated. In addition, Nussbaum-type gain is
introduced and some valuable lemmas are given. Finally, math- where denotes the state vector of
ematical description of NNs is introduced. leader node 0; is unknown but
bounded.
A. Basic Graph Theory and Notations Let us define the neighborhood synchronization error for fol-
To solve the coordination problems and model the informa- lower as follows:
tion exchange between agents, graph theory is introduced as in
[19]. For convenience, basic graph theorem and some notations
are repeated here.
Let be a weighted graph, is where , , denotes the neighbors
the nonempty set of nodes/agents, is the set of arcs, of follower , is the weight of edge from follower
means there is an edge from node to node . Its to follower , and is the weight of edge from the leader
adjacency matrix , and if node to follower ( ), if there is an edge
; otherwise . Throughout this paper, it is assumed from the leader node to follower .
that and the topology is fixed, i.e., is time-invariant. Define the tracking error for follower is defined as follows:
is a directed graph. Define as the weighted
in-degree of node and as
in-degree matrix. The graph Laplacian matrix is
Let , , then
. Let with appropriate
dimension; then . The set of neighbors of node is
denoted as . If node is a neighbor
of node , then node can get information from node , not where .
necessarily vice versa for directed graph [19]. Define the graph as with
and .
B. Cooperative Tracking Problem
The cooperative tracking control problem considered herein
In this paper, the cooperative control problem of a leader- is as follows: A distributed adaptive neural networks-based con-
following multi-agent system consisting of followers and one troller is designed for each follower in the graph such that all fol-
leader will be investigated. Each follower takes the following lowers can asymptotically synchronize the leader with tracking
dynamics: error being semiglobally uniform ultimate bounded, for all
.
(1) To design distributed controller for each follower, the fol-
lowing assumptions and lemmas are given.
where , denotes Assumption 1: The augmented graph contains a spanning
the measurable state vector of node ; is the control tree with the root node being the leader node 0.
input/protocol; denotes an unknown smooth Assumption 2: There exists an unknown real constant
function; denotes the complete unknown such that .
controlling effect coefficient of each agent, i.e., both its value Assumption 3: For each node , there exists an unknown real
and sign are unknown; and is an external disturbance, constant such that .
which is unknown but bounded. Lemma 1 [19]: Define ,
For notational simplicity, we use to denote . For ex- , ,
ample, is the abbreviation of . then and .
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SHI AND SHEN: COOPERATIVE CONTROL OF MULTI-AGENT SYSTEMS WITH UNKNOWN STATE-DEPENDENT CONTROLLING EFFECTS 3

Lemma 2 [19]: , D. Nussbaum-Type Gain

where is a minimum singular value of matrix . Any continuous function is a Nussbaum-type
function if it has the following properties [28]:
C. Cooperative Synchronization Error Dynamics and Sliding
Mode Error
Define the following notations:

For example, the continuous functions , and

used in [26], [28], are Nussbaum-type func-
tions. The Nussbaum function is considered
Similar to (2), the above synchronization error can be throughout this paper.
re-written in the following compact form: Lemma 4 [28]: Let and be smooth functions de-
fined on with , and be an
even smooth Nussbaum-type function. The following inequality
holds:
where the property is used.
Define the following sliding mode error for the th fol-
lower as follows:
where is a nonzero constant, and represents some suit-
(4) able constants, then and must be
bounded on .
Lemma 5 [28]: Let and be smooth functions de-
where fined on with , and be an
denotes a design parameter, even smooth Nussbaum-type function. The following inequality
. holds:
Let

Lemma 3: Let be defined by (4), and then,

1) if , then ;
2) if , , then , ; where constant is a time-varying parameter which
3) if , , then , takes values in the unknown closed intervals
, , with , and represents some suitable constant, then
where , , denote design parameters, and must be bounded on .
, , .
For simplification, let , E. Mathematical Description of NNs
, then In this paper, since in (1) is unknown, neural networks
(NNs) presented in [28] shall be used to approximate it as

Define overall sliding mode error as

where denotes the optimal approximation
error, ,
, is the number of the NNs,
then
, is the
dimension of , is the width of the receptive field,
and , are the center of the Gaussian
Further, one has function, the ideal weight is defined as

with a constant , denotes an enough large compact

where . set.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING

From the universal approximation results as [28], we know,

NNs can approximate any continuous function to any accuracy (8)
on compact set .
Assumption 4: The ideal weight is bounded and there From [29], we know, for
exists an unknown real constant such that
over a compact set, where .

III. DESIGN OF DISTRIBUTED ADAPTIVE CONTROLLER AND where is a design parameter. So, from (8), one has
STABILITY ANALYSIS
Recalling the overall sliding model error dynamics

Define the following Lyapunov function:

Further, one has

where defined as in Lemma 1.

Differentiating with respect to time , one has

(9)
(5) where .
Since
where , ,
, .
Design the distributed control law for follower as:
one has

(10)

where denotes an identity matrix with appropriate dimension.

So, (9) is thus rewritten as follows:

(6)

where , is a Nussbaum-type function,

is the th element of , is a design
parameter, which satisfies ,
and are design parameters, or collectively (11)

Because
(7)

where ,
, ,
, are the estimates of
, , (12)
, is the estimate of where , the property: and
, . are used, (11) is further developed as
Let , where is the estimate
of , .
In the following, define the notation: .
Substituting the control law (7) into (5), one has

(13)
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SHI AND SHEN: COOPERATIVE CONTROL OF MULTI-AGENT SYSTEMS WITH UNKNOWN STATE-DEPENDENT CONTROLLING EFFECTS 5

Substituting in (7) into (13), one has Define

(22)

Differentiating with respect to time , one has

(14)

where . Since

(23)
(15)
Define the adaptive laws as follows:

(16)

where is a design parameter and (24)

, one has

(25)

(17) where ,
, , are design
Notice that, since , and are bounded, which parameters. The project operator is adopted such that and
are guaranteed by Assumption 4 and the adaptive law (25), are bounded.
is also bounded. Substituting (24) and (25) into (23), one has
Furthermore,from Assumptions 2–4, i.e., , and are
bounded, is bounded, too.
Hence, if and are chosen to be enough large and
small, respectively, then
(26)
(18)
Since
where is a design parameter. Further, one has

(27)

(19)

Define (28)

one has
(20)

where , are design parameters.

Differentiating with respect to time , one has

(21) (29)
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING

From the above analysis, we know, are bounded. which means all signals in the closed-loop system are SGUUB.
Hence, if , and are chosen appropriately, then And since , from ,
one has

(30)

where is a design parameter.

Let , then one has From Lemma 3, one further has, for
(31)

where
.
Since
Since , one has

(32)
Further, from Lemma 2, one has, for ,
where is the th element of , one has

(33)
Further, one has

The above control design procedures and analysis are summa-

rized in the following theorem.
(34) Theorem 1: Consider the distributed system (1) and leader
node (3) under Assumptions 1–4 , the distributed control law
where (7) and adaptive laws (24) and (25), if , , , , ,
. , and are chosen appropriately such that
Applying Lemma 5, we can conclude that, , and
are SGUUB on .
According to Proposition 2 in [27], if the solution of the
closed-loop system is bounded, then .
Let be the upper bound of
, we have,

(35) then all follower nodes in the graph synchronize to the leader
node 0 with the tracking errors , are SGUUB,
Thus, (34) becomes namely, the errors belong to the adjustable set defined as

(36)

Hence, if (36) holds, then, the proposed control input (7) can
ensure that is bounded, namely, the closed-loop system is
semiglobally uniformly ultimately bounded.
From the definition of , one has Proof: From the above analysis, the desired result can be
obtained, and the detailed proof is omitted here.

IV. SIMULATION EXAMPLE

In this example, we consider the consensus control problem
of four inverted pendulums, networked through the graph shown
in Fig. 1, where the numbers on the edge denote the weights
between corresponding two nodes. The dynamics of the leader
Therefore, one has node is described as follows:
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SHI AND SHEN: COOPERATIVE CONTROL OF MULTI-AGENT SYSTEMS WITH UNKNOWN STATE-DEPENDENT CONTROLLING EFFECTS 7

Fig. 1. Communication topology .

Fig. 3. The time profiles of the states of the followers and the leader .

and (25) to systems (1) and (3). As shown in Figs. 2 and 3, the
states of the followers can asymptotically synchronize the leader
with the tracking errors converge to a small neighborhood of the
origin under the controllers (7).

V. CONCLUSION
Fig. 2. The time profiles of the states of the followers and the leader . In this paper, the cooperative control problem of uncertain
nonlinear multi-agent systems on directed graph with a fixed
topology is investigated. It is assumed that, each follower in
The follower nodes are described by two-order nonlinear sys- the graph has a complete unknown state-dependent controlling
tems in the form of (1) with and , for , effect, and the leader gives its commands only to a small por-
defined as: tion of the graph. By the Nussbaum-type gain technique and
the function approximation capability of neural networks, using
the relative state information between each follower node and
its neighbors, a distributed robust adaptive neural networks-
based control scheme is proposed such that the tracking er-
rors are semi-globally uniform ultimate bounded. However, the
topology among nodes considered in this paper is fixed. In prac-
tical applications, the topology may be variable. For example, a
where and are the angular position and speed of the
new node is added to or removed from communication network.
pole of the th pendulum, respectively. Furthermore, and
In addition, transmission nonlinearity and delay may occur in
denote masses of cart and pole. is half-length of the pole
the communication network. In the case, how to handle the co-
of th pendulum. The values of these parameters in this example
operative control problem of nonlinear multi-agent systems is
are given as follows:
important and challenging, which is studied in our further re-
search.

REFERENCES
[1] M. Defoort, T. Floquet, A. Kokosy, and W. Perruquetti, “Sliding-mode
formation control for cooperative autonomous mobile robots,” IEEE
Trans. Ind. Electron., vol. 55, no. 11, pp. 3944–3953, Nov. 2008.
[2] D. Cruz, J. McClintock, B. Perteet, O. Orqueda, Y. Cao, and R.
and the disturbance is random and bounded Fierro, “Decentralized cooperative control-A multivehicle platform
by . for research in networked embedded systems,” IEEE Control Syst.
The initial states are set as: , Mag., vol. 27, no. 3, pp. 58–78, Jun. 2007.
[3] X. Wang, V. Yadav, and S. Balakrishnan, “Cooperative UAV formation
, , and flying with obstacle/collision avoidance,” IEEE Trans. Control Syst.
. Weight vector are taken Technol., vol. 15, no. 4, pp. 672–679, Jul. 2007.
randomly in interval , the sample time is 0.08 s. [4] J. Chen, X. Cao, P. Cheng, Y. Xiao, and Y. Sun, “Distributed collabora-
tive control for industrial automation with wireless sensor and actuator
Figs. 2 and 3 give partial simulation results which are ob- networks,” IEEE Trans. Ind. Electron., vol. 57, no. 12, pp. 4219–4230,
tained by applying the control laws (7) and adaptive laws (24) Dec. 2010.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING

[5] R. Murray, “Recent research in cooperative control of multivehicle [28] S. S. Ge, F. Hong, and T. H. Lee, “Adaptive neural control of non-
systems,” ASME J. Dyn. Syst., Meas., Control, vol. 129, no. 5, pp. linear time-delay systems with unknown virtual control coefficients,”
571–583, 2007. IEEE Trans. Syst., Man, Cybern.-Part B: Cybern., vol. 34, no. 1, pp.
[6] W. Ren and C. Cao, Distributed Coordination of Multi-Agent Net- 499–516, Feb. 2004.
works: Emergent Problems, Models, and Issues. London, U.K.: [29] M. M. Polycarpou and P. A. Ioannou, “A robust adaptive nonlinear
Springer-Verlag, 2011. control design,” Automatica, vol. 31, no. 3, pp. 423–427, 1995.
[7] R. Olfati-Saber, J. Alex Fax, and R. Murray, “Consensus and coopera- [30] M.-H. Hung, C.-F. Chen, H.-C. Huang, H.-C. Yang, and F.-T. Cheng,
tion in networked multi-agent systems,” Proc. IEEE, vol. 95, no. 1, pp. “Modeling and optimization of energy consumption in cooperative
215–233, Jan. 2007. multi-robot systems,” IEEE Trans. Autom. Sci. Eng., vol. 9, no. 2, pp.
[8] W. Ren, R. Beard, and E. Atkins, “Information consensus in multive- 423–428, Apr. 2012.
hicle cooperative control,” IEEE Control Syst. Mag., vol. 27, no. 2, pp. [31] K. Mathiyalagan, J. H. Park, R. Sakthivel, and S. Marshal Anthoni,
71–82, Apr. 2007. “Robust mixed HH and passive filtering for networked Markov jump
[9] Q. K. Shen, B. Jiang, and P. Shi, “Cooperative adaptive fuzzy tracking systems with impulses,” Signal Process., vol. 101, pp. 162–173, 2014.
control for networked unknown nonlinear multi-agent systems with [32] H. Zhao, J. H. Park, Y. Zhang, and H. Shen, “Distributed output
time-varying actuator faults,” IEEE Trans. Fuzzy Syst., vol. 22, no. 3, feedback consensus of discrete-time multi-agent systems,” Neurocom-
pp. 494–504, Jun. 2014. puting, vol. 138, pp. 86–91, 2014.
[10] H. Du, S. Li, and P. Shi, “Robust consensus algorithm for second-order [33] M. H. Nguyen and D. Q. Tran, “A combination trust model for multi-
multi-agent systems with external disturbances,” Int. J. Control, vol. agent systems,” Int. J. Innovative Comput., Inf. Control, vol. 9, no. 6,
85, no. 12, pp. 1913–1928, 2012. pp. 2405–2420, 2013.
[11] S. Chen, P. Shi, W. Zhang, and L. Zhao, “Finite-time consensus on [34] R. Szupiluk, P. Wojewnik, and T. Zabkowski, “Smooth component
strongly convex balls of Riemannian manifolds with switching directed analysis and MSE decomposition for ensemble methods in multi-agent
communication topologies,” J. Math. Anal. Appl., vol. 409, no. 2, pp. environment,” Int. J. Innovative Comput., Inf. Control, vol. 10, no. 4,
663–675, 2014. pp. 1435–1445, 2014.
[12] S. Li, H. Du, and P. Shi, “Cooperative adaptive fuzzy tracking con- [35] A. Sedziwy, “Effective graph representation supporting multi-agent
trol for networked unknown nonlinear multi-agent systems with time- distributed computing,” Int. J. Innovative Comput., Inf. Control, vol.
varying actuator faults,” IEEE Trans. Aerosp. Elect. Syst., Aug. 2013, 10, no. 1, pp. 101–113, 2014.
to be published. [36] Q. K. Shen and P. Shi, “Distributed command filtered backstepping
[13] W. Zhu and D. Cheng, “Leader-following consensus of second-order consensus tracking control of nonlinear multiple-agent systems in
agents with multiple time-varying delays,” Automatica, vol. 46, no. 12, strict-feedback form,” Automatica, 2014.
pp. 1994–1999, 2010.
[14] W. Wang and J. Slotine, “A theoretical study of different leader roles in
networks,” IEEE Trans. Autom. Control, vol. 51, no. 7, pp. 1156–1161,
Jul. 2006.
[15] Y. Hong, G. Chen, and L. Bushnell, “Distributed observers design for
leader-following control of multi-agent networks,” Automatica, vol. Peng Shi (M'95–SM'98–F'15) received the B.Sc. degree in mathematics from
44, no. 3, pp. 846–850, 2008. the Harbin Institute of Technology, Harbin, China, in 1982, the M.E. degree
[16] H. Yu and X. Xia, “Adaptive consensus of multi-agents in net- in systems engineering from Harbin Engineering University, Harbin, China, in
works with jointly connected topologies,” Automatica, vol. 48, pp. 1985, the Ph.D. degree in electrical engineering from the University of New-
1783–1790, 2012. castle, Callaghan, Australia, in 1994, the Ph.D. degree in mathematics from the
[17] W. Ren, K. L. Moore, and Y. Chen, “High-order and model reference University of South Australia, Adelaide, Australia, in 1998, and the Doctor of
consensus algorithms in cooperative control of multivehicle systems,” Science degree from the University of Glamorgan, Pontypridd, U.K.
ASME J. Dyn. Syst., Meas., Control, vol. 129, no. 5, pp. 678–688, 2007. He was a Lecturer with the University of South Australia; a Senior Scientist
[18] H. Zhang, F. L. Lewis, and A. Das, “Optimal design for synchroniza- with the Defence Science and Technology Organization, Australia, and a Pro-
tion of cooperative systems: State feedback, observer and output feed- fessor with the University of Glamorgan. He is now a Professor with the Uni-
back,” IEEE Trans. Autom. Control, vol. 56, no. 8, pp. 1948–1952, versity of Adelaide, and Victoria University, Australia, and a Professor Emer-
Aug. 2011. itus with the University of South Wales, U.K. His main research interests are in
[19] H. Zhang and F. L. Lewis, “Adaptive cooperative tracking control of system and control theory, computational intelligence, and operational research.
higher-order nonlinear systems with unknown dynamics,” Automatica, Dr. Shi is a Fellow of the Institution of Engineering and Technology and a
vol. 48, no. 7, pp. 1432–1439, 2012. Fellow of the Institute of Mathematics and its Applications. He has been an
[20] H. Zhang, F. L. Lewis, and Z. Qu, “Lyapunov, adaptive, and optimal Associate Editor for a number of scientific journals, such as Automatica, the
design techniques for cooperative systems on directed communication IEEE TRANSACTIONS ON AUTOMATIC CONTROL, the IEEE TRANSACTIONS
graphs,” IEEE Trans. Ind. Electron., vol. 59, no. 7, pp. 3026–3041, Jul. ON CYBERNETICS, the IEEE TRANSACTIONS ON FUZZY SYSTEMS, the IEEE
2012. TRANSACTIONS ON CIRCUITS AND SYSTEMS-I, and IEEE ACCESS. He has been
[21] W. Ni and D. Cheng, “Leader-following consensus of multi-agent sys- actively involved in international conferences as a General Chair and Program
tems under fixed and switching topologies,” Syst. Control Lett., vol. 59, Committee/Organizing Committee Member. He is now the Chair of the
no. 3-4, pp. 209–217, 2010. Control, Aerospace and Electronic Systems Chapter, the IEEE South Australia
[22] L. Guo, X. Nian, and H. Pan, “Leader-following consensus of multi- Section, and a College of Expert Member, Australian Research Council.
agent systems with switching topologies and time-delays,” J. Control
Theory Appl., vol. 11, no. 2, pp. 306–310, 2013.
[23] V. T. Haimo, “Finite time controllers,” SIAM J. Control Optim., vol.
24, no. 4, pp. 760–770, 1986. Qikun Shen received the B.Sc. degree in computer science and applications
[24] M. Egerstedt and X. Hu, “Formation constrained multi-agent control,” from the Chinese University of Mining and Technology, Xuzhou, China,
IEEE Trans. Robot. Autom., vol. 17, no. 6, pp. 947–951, Dec. 2001. in 1996, and the M.Sc. degree in computer science and applications from
[25] Z. Qu, Cooperative Control of Dynamical System: Application to Au- Yangzhou University, Yangzhou, China, in 2007. Currently, he is working
tonomous Vehicles. London, U.K.: Springer-Verlag, 2009. towards the Ph.D. degree at the College of Automation Engineering, Nanjing
[26] X. Ye and J. Jiang, “Adaptive nonlinear design without a priori knowl- University of Aeronautics and Astronautics, Nanjing, China.
edge of control directions,” IEEE Trans. Autom. Control, vol. 43, pp. He is currently an Associate Professor with the College of Information Engi-
1617–1621, Nov. 1998. neering, Yangzhou University. His research interests include distributed control,
[27] E. P. Ryan, “A universal adaptive stabilizer for a class of nonlinear consensus control, fault-tolerant control, adaptive control, fuzzy control, neural
systems,” Syst. Control Lett., vol. 16, pp. 209–218, 1991. networks-based control and intelligent control, etc.