Abstract
In this paper, the bearing-only formation control problem of a class of second-order system with unknown disturbance is investigated, where the control law merely depends on the relative bearings between neighboring agents. In order to offset the effect of unknown disturbance on the system, adaptive estimation is introduced. In the design of the control law, the back-stepping design method and the negative gradient method are used. The Barbalatās lemma is used to prove the global stability of the system. The simulation results prove the effectiveness of the proposed formation control algorithm.
1 Introduction
In recent years, formation control of multi-agent systems has attracted great attention due to its wide application in military, scientific research, and daily life. The information commonly used in the existing formation control are categorized into position, relative distance and relative bearing. In particular, the research on formation control based on position or relative distance are very rich (; ; ; ; ; ; ; ; ; ). Relatively speaking, there is a lack of researches on formation control based on relative bearing. However, compared to the formation control approaches that rely on position or relative distance, bearing-only formation control has lower requirements for sensors, so it reduces the production cost and has a wide range of application scenarios. Therefore, the research on bearing-only formation control is of great significance.
have put forward the definition of bearing rigidity for any dimension and proved the conditions of bearing rigidity, which greatly promoted the development of formation control based on relative bearing. Based on the definition, researchers have done a lot of research. For different agent models such as single integrators, double integrators and unicycles, designed a series of bearing-only control laws, which can enable followers to track leaders and reach the specified formation, and proved its stability by use of the standard Lyapunov method. solved the bearing-only formation control problem of the three-dimensional nonholonomic constraints system by transforming the three-dimensional nonholonomic model into Euler-Lagrange-like form and using the back-stepping design method. considered the system with local reference frame, they synchronized the orientations of followers with the leader first, and then adopted bearing-only control law to solve the formation control problem of double-integrators without global reference coordinates. designed a relative position estimator, which uses the relative bearing and the linear velocity information of the leader to estimate the relative position of the leader, and designed a controller with the estimated relative position information to complete the task of tracking the leader. Since collision is devastating to the formation control system, proposed a distributed cooperative consensus control algorithm, and proved that this control method can prevent agents from colliding effectively. Besides, used the maximal clique graph instead of infinitesimal rigidity to express the communication topology between multi-agent with non-holonomic constraints, and designed a bearing-only formation control law to solve the problem of how to achieve global stability without leaders. Compared with these existing research on bearing-only formation control, most of existing research are focused on linear system, while the system studied in this paper is a nonlinear system with unknown time-varying disturbance. Disturbance is inevitable in reality, so it seems of more practical significance to study nonlinear system.
The main contribution of this paper is to propose a novel bearing-only formation control law that can handle second-order system with unknown disturbance. The control algorithm based on the negative gradient method only uses the information of relative bearings between neighboring agents. The adaptive estimation method is introduced to counteract the unknown disturbance, and the back-stepping design method is used in the design process. Then, the global stability of the bearing-only formation system is demonstrated by use of the Barbalatās Lemma.
2 Problem Statement
Consider a formation system comprising of n agents, and the model of each agent is as follows:
In the formula above, stands for the position of agent i, and and represent velocity and control input of agent i, respectively. Here, . is the unknown disturbance, and we assume , is unknown constant.
In this paper, we use an undirected graph to represent the communication relationship of the system. represents the set of agents in the system, and represents the set of edges in the system. The edge indicates that agent i can measure the relative bearing of agent j, and hence agent j is neighbor of i. Let denote the set of neighboring vertices of vertex i, and we also have .
Define the edge vector and bearing vector, respectively:
As shown above, stands for the Euclidean distance between agent i and agent j, is neighbor of agent i. means the bearing of relative to . In this chapter, assuming that all agents can perceive the global coordinate system, there are , . Relative bearing of agents can usually be measured by onboard vision in practice.
In this paper, considering the control of formation without leader, the position information of multiple agents is . An oriented graph is obtained by orienting each edge in the undirected graph G, that is, the direction of any edge in G is calibrated (). Define the edge in the graph G, tagging the order from 1 to m, m represents the total number of edges in graph G. Number the edges in from to , then get the corresponding incidence matrix . The incidence matrix H is a matrix composed of 0, 1, and ā1. The elements on the rows of the incidence matrix are indexed by edges , and the value elements on the columns are indexed by vertices , namely: except for and , all elements in the row k of the incidence matrix H are zero. Among them, when vertex i is the last vertex of edge k, , and when vertex j is the first vertex of edge k, .
Therefore, the incidence matrix is defined as:where denotes the Kronecker product, and is the identity matrix.
Assuming that the edge in the graph corresponds to the edge k in the directed graph, where . The edge and bearing vectors for the kth directed edge can be expressed as: , , where , .
At the same time, assign an expected vector to each edge vector in graph G, which is called . In the same way, define and .
In this paper, we use the infinitesimally bearing rigidity to express the communication topology of multi-agent systems. It means that the shape of the formation can be uniquely determined. put forward the definition of infinitesimally bearing rigidity for any dimension and proved the conditions of infinitesimally bearing rigidity, which is shown in the following lemma.
Lemma 1: (): A framework G in is infinitesimally bearing rigid if and only if rank , where , .
In order to design the globally stable formation control strategy, we give the following assumption and the potential function.
Assumption 1: The initial positions of agents do not coincide, and neighboring agents will not collide during formation.
The potential function between agent i and its neighboring agent j is defined as follows:
Define:
The definition of total potential function of agent i is
The problem of the bearing-only formation control to be solved in this paper is formally described as follows.
Problem 1: Consider multi-agent systems of n agents with unknown disturbance, its communication topology is defined by infinitesimal bearing rigidity. For any initial condition other than coincidence , , design a bearing-only control law to control the relative bearings between agents reaches the desired value, and all agents are moving at the desired velocity , which make the multi-agent systems form the specified formation, that is:
3 Design of Global Stabilizer
Define , , , , then we have
Also, let , , then , so the whole system can be written in the form as:
Then, the control law is designed by use of the back-stepping design method.
Step 1:
Focus on the following subsystems of Eq. 7:
In this subsystem, , and we use the feedback controller to control the relative bearing.
The Lyapunov function candidate is chosen as follows:
Owing to , is symmetric about , we have , , then
The virtual control law is chosen asand it follows that
Step 2:
Then, introducing the error variable
Take the derivative of the above equation, that is
For the above system, the Lyapunov function candidate is chosen as follows:where , is the estimated value of at time t.
Differentiate the function with respect to time, we have
Motivated by , the global stabilized formation control law is designed as follows
Because of , the control law is as follows,
The adaptive law is shown as:
Here , where represents the dth element in , , and the sign function is shown as follows:
Remark 1: . In practice, can generally be measured by onboard vision. At the same time, the rate of change of relative bearing can also be measured by onboard vision with optical flow technology, so can be measured.
Remark 2: The sign function in the control law would cause the chattering phenomenon. In simulation or engineering practice, the saturation function is usually used to replace the sign function to solve the problem of chattering. When , ; when , , where represents the width of the boundary layer. Whether the saturation function or the sign function is used, the results in engineering practice and simulation are unchanged.
4 Stability Analysis
The formation stability is analyzed as follows.
Theorem 1: Under the control law (Eq. 18), the multi-agent systems converges asymptotically to the desired shape and all agents move at the desired velocity , and the closed-loop system is globally asymptotically stable.
Proof: The Lyapunov function candidate is determined by Eq. 15, and note that is smooth and hence regular; its generalized gradient () is a singleton which is equal to its usual gradient everywhere in the state space: (). Along with Eqs. 13ā19, we can obtainthe above inequality is obtained when .
Eq. 20 proves that is non-increasing, and Eq. 15 shows that greater than zero, so it can be inferred that it has a limit . Define , so . Integrating both sides of the equation, we havewhich means that exists and is finite. Now we show that is also uniformly continuous.
Since , are bounded. It is concluded that and are not equal to zero because it is assumed that agents will not collide. So are bounded, which indicates are uniformly continuous with . Since is continuous, it is uniformly continuous with , so it can be deduced that is uniformly continuous with . According to Barbalatās lemma (), we can get , and we can deduce:
In the same wayThat is
In the following, put the above equation into matrix-vector form
Combining Eq. 2 and Eq. 21, we have:
We can get . It means that all the relative bearing vectors meet the expected value , that is, the multi-agent systems meets the expected formation, and the system is globally asymptotically stable.
So, . It can be seen from Eq. 13, , that is to say, .
In summary, the system will converge to the following set:
That is, the relative bearings between agents reach the desired value, and all agents are moving at the same desired velocity; then the multi-agent systems achieve the desired formation shape.
5 Simulations
In the simulation, the dynamic model of agents is expressed by the following equation:
Then, we prove the effectiveness of the control law through 2D and 3D simulation examples.
5.1 Simulation for 2-Dimensional
In simulation, we choose , , , . Set the initial position of each agent to , , , , , the initial velocities of the agents are set to , , , , . is the desired relative bearing vector, where , , , , , , , and the desired velocity is . The simulation results are shown in Figures 1ā3.
FIGURE 1
FIGURE 2
FIGURE 3
As is shown in Figure 1, the formation converges to the desired regular pentagon. Figure 2 shows the bearing errors converge to zero. Figure 3 illustrates that the velocities of agents converge to desired one. The above results prove the effectiveness of the proposed control law in two dimensions.
Increasing the disturbance by a factor of three, and comparing the results of Figures 4ā6 with Figures 1ā3, it can be seen that the magnitude of the disturbance will affect the time required for the system to reach stability. The larger the disturbance, the longer it takes to reach stability.
FIGURE 4
FIGURE 5
FIGURE 6
5.2 Simulation for 3-Dimensional
In simulation, we choose , , , . Setting the initial position of each agent to , , , , , the initial velocities of the agents are set to , , , , . is the desired relative bearing vector, where , , , , , , , and the desired velocity is . The simulation results are shown in Figures 7ā9.
FIGURE 7
FIGURE 8
FIGURE 9
As shown in Figure 7, the formation converges to the desired right square pyramid. Figure 8 shows the bearing errors converge to zero. Figure 9 illustrates that the velocities of agents converge to desired one. The above results prove the effectiveness of the proposed control law in three dimensions.
6 Conclusion
This paper proposes an adaptive formation control algorithm based on negative gradient method to solve the formation problem of a class of second-order system with uncertain disturbances, and the control law only uses the information of relative bearing between neighboring agents. The algorithm introduces adaptive estimation to counteract unknown disturbances. Furthermore, the back-stepping design method is used in the design process. Then, the global stability of the whole formation system is proved by Barbalatās lemma. In the end, the simulation results in 2D and 3D show the effectiveness of the algorithm.
Statements
Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.
Author contributions
SL contributed to the design of control law, simulation and writing of the original draft. QW helped with the methodology, guidance, review and editing. EW and YC helped with review and editing. All authors contributed to the article and approved the submitted version.
Funding
This work was supported in part by the National Natural Science Foundation of China under Grants 61803331, 61873229, 61806175, and 61873346, in part by the Jiangsu Planned Projects for Postdoctoral Research Funds 1601024B.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
References
1
BiggsN. (1974). Algebraic Graph Theory. Cambridge: Cambridge University Press. 10.1017/cbo9780511608704
2
ChenL.MeiJ.LiC.MaG. (2020). Distributed Leader-Follower Affine Formation Maneuver Control for High-Order Multiagent Systems. IEEE Trans. Automat. Contr.65, 4941ā4948. 10.1109/tac.2020.2986684
3
ClarkeF. H. (1990). Optimization and Nonsmooth Analysis. New York: John Wiley & Sons. 10.1137/1.9781611971309
4
GeS. S.HangC. C.LeeT. H.ZhangT. (2002). Stable Adaptive Neural Network Control. New York: Springer US. 10.1007/978-1-4757-6577-9
5
HanZ.GuoK.XieL.LinZ. (2019). Integrated Relative Localization and Leader-Follower Formation Control. IEEE Trans. Automat. Contr.64, 20ā34. 10.1109/TAC.2018.2800790
6
HuZ.YangJ. (2020). Collision Avoidance Cooperative Attack with Multiple Pursuers Based on Bearing-Only Measurements. J. Franklin Inst.357, 437ā456. 10.1016/j.jfranklin.2019.11.061
7
LiX.WenC.ChenC. (2021). Adaptive Formation Control of Networked Robotic Systems with Bearing-Only Measurements. IEEE Trans. Cybern.51, 199ā209. 10.1109/TCYB.2020.2978981
8
LinZ.WangL.HanZ.FuM. (2016). A Graph Laplacian Approach to Coordinate-free Formation Stabilization for Directed Networks. IEEE Trans. Automat. Contr.61, 1269ā1280. 10.1109/TAC.2015.2454711
9
LiuY.HuangP.ZhangF.ZhaoY. (2020). Distributed Formation Control Using Artificial Potentials and Neural Network for Constrained Multiagent Systems. IEEE Trans. Contr. Syst. Technol.28, 697ā704. 10.1109/TCST.2018.2884226
10
LuoX.LiX.LiX.YanJ.GuanX. (2020). Globally Stable Formation Control of Nonholonomic Multiagent Systems with Bearing-Only Measurement. IEEE Syst. J.14, 2901ā2912. 10.1109/JSYST.2019.2935162
11
MehdifarF.BechlioulisC. P.HashemzadehF.BaradaranniaM. (2020). Prescribed Performance Distance-Based Formation Control of Multi-Agent Systems. Automatica119, 109086. 10.1016/j.automatica.2020.109086
12
MehdifarF.HashemzadehF.BaradaranniaM.de QueirozM. (2019). Finite-time Rigidity-Based Formation Maneuvering of Multiagent Systems Using Distributed Finite-Time Velocity Estimators. IEEE Trans. Cybern.49, 4473ā4484. 10.1109/TCYB.2018.2876608
13
RanD.ChenX.MisraA. K.XiaoB. (2017). Relative Position Coordinated Control for Spacecraft Formation Flying with Communication Delays. Acta Astronautica137, 302ā311. 10.1016/j.actaastro.2017.04.011
14
ShevitzD.PadenB. (1993). āLyapunov Stability Theory of Nonsmooth Systems,ā in Proceedings of 32nd IEEE Conference on Decision and Control, San Antonio, TX, December 15, 1993 (Piscataway, NJ: IEEE), 416ā421. 10.1109/CDC.1993.325114
15
SunZ.de MarinaH. G.SeybothG. S.AndersonB. D. O.YuC. (2019). Circular Formation Control of Multiple Unicycle-type Agents with Nonidentical Constant Speeds. IEEE Trans. Contr. Syst. Technol.27, 192ā205. 10.1109/TCST.2017.2763938
16
WangQ.ZhuY.LiJ.HuaQ. (2015). āGlobally Stable Rigid Formation Control for Multi-Robot Systems,ā in 2015 34th Chinese Control Conference (CCC), Hangzhou, China, July 28ā30, 2015 (Piscataway, NJ: IEEE), 7505ā7510. 10.1109/ChiCC.2015.7260829
17
YangZ.ZhuS.ChenC.FengG.GuanX. (2020). Leader-follower Formation Control of Nonholonomic mobile Robots with Bearing-Only Measurements. J. Franklin Inst.357, 1628ā1643. 10.1016/j.jfranklin.2019.11.025
18
YuD.ChenC. L. P. (2020). Automatic Leader-Follower Persistent Formation Generation with Minimum Agent-Movement in Various Switching Topologies. IEEE Trans. Cybern.50, 1569ā1581. 10.1109/TCYB.2018.2865803
19
ZhaoJ.YuX.LiX.WangH. (2021). Bearing-only Formation Tracking Control of Multi-Agent Systems with Local Reference Frames and Constant-Velocity Leaders. IEEE Control. Syst. Lett.5, 1ā6. 10.1109/LCSYS.2020.2999972
20
ZhaoS.LiZ.DingZ. (2019). Bearing-only Formation Tracking Control of Multiagent Systems. IEEE Trans. Automat. Contr.64, 4541ā4554. 10.1109/TAC.2019.2903290
21
ZhaoS.ZelazoD. (2016). Bearing Rigidity and Almost Global Bearing-Only Formation Stabilization. IEEE Trans. Automat. Contr.61, 1255ā1268. 10.1109/TAC.2015.2459191
22
ZouY.WenC.GuanM. (2019). Distributed Adaptive Control for Distanceābased Formation and Flocking Control of Multiāagent Systems. IET Control. Theor. Appl.13, 878ā885. 10.1049/iet-cta.2018.6001
Summary
Keywords
formation control, relative bearing, back-stepping design method, negative gradient method, global stability
Citation
Li S, Wang Q, Wang E and Chen Y (2021) Bearing-Only Adaptive Formation Control Using Back-Stepping Method. Front. Control. Eng. 2:700053. doi: 10.3389/fcteg.2021.700053
Received
25 April 2021
Accepted
11 June 2021
Published
15 July 2021
Volume
2 - 2021
Edited by
Chunyan Wang, Beijing Institute of Technology, China
Reviewed by
Qing Wang, Beihang University, China
Yang Xu, Northwestern Polytechnical University, China
Dandan Wang, Beijing Institute of Technology, China
Updates
Copyright
Ā© 2021 Li, Wang, Wang and Chen.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Qin Wang, qinwang@yzu.edu.cn
This article was submitted to Nonlinear Control, a section of the journal Frontiers in Control Engineering
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.