CN104898663A - Distributed multi-robot containment collision prevention control method - Google Patents

Abstract

A distributed multi-robot containment control method comprises the following steps: (1) determining leader and follower sets in a directed network according to the communication relationship between robots in the network; (2) designing a position control operator of the follower robots when the leaders are static; (3) introducing a potential energy function to ensure that collision between the follower robots in the process of movement is prevented, and forming a containment control rate when the leaders are static; (4) designing a position control operator of the follower robots when the leaders are dynamic; (5) introducing an exponential estimator to a control strategy formulated for the follower robots to estimate the speed of the leader robots, realize the transformation from locally knowable to globally knowable and determine a speed control operator of the follower robots; and (6) integrating the position control operator and a direction control operator of the robots in a weighted manner, introducing the potential energy function to ensure that collision between individuals in the process of movement is prevented, and forming a distributed containment control rate of the dynamic robots.

Description

A kind of distributed multirobot comprises control of collision avoidance method

Technical field

The present invention relates to intelligent robot technology field, relate in particular to a kind of distributed multirobot and comprise control of collision avoidance method.

Background technology

Along with continuous maturation and the development of computer technology, control theory, artificial intelligence theory, sensor technology, the robotics formed by multi-crossed disciplines research also enters a brand-new stage.From the industrial robot of programmable, teaching playback to the robot with certain sensing capability and adaptive faculty, arrive again and be equipped with multiple advanced sensors, there is the intelligent robot of stronger adaptive faculty, the research work of robotics experienced by one from simple to complexity, from function singleness to diverse in function, from industry manufacture field to military surveillance, nuclear industry, Aero-Space, service sector, the numerous areas such as genetic engineering process.Can predict, in the near future, Robotics will be more extensive in the application of every field.And the widespread use of various robot system in the real work requirement new for robotics proposes and new research topic.The research of multi-robot system is exactly propose under these new application demands drive, and becomes an important branch of robotics research gradually along with the development of robotics.

Multi-robot system cooperation control has become the emerging study hotspot of control field, and as the comprehensive advanced subject of control theory circle, its research category relates to the fields such as biology, mathematics, physics, control, computing machine, communication and artificial intelligence, and obtain successful Application at many engineering fields, comprising: robot team formation, Distributed Calculation, space development and detection, wireless sensor network are located and in the engineering reality such as intelligent grid scheduling.Its main task is design con-trol agreement, and realizes overall situation exchange by local exchange, thus reaches desirable coherency state.

When existing traffic signal coordination often concentrates on without pilotage people or single pilotage people, comprise control then more general in actual applications.What is called comprises control, refers to that one group of follower is under the leading of multiple pilotage people, thus arrives and remain on motion in the minimum geometric space (convex closure) surrounded by leader.In actual applications, containing controls jointly to complete in the coordination of tasks such as dangerous substances process, enemy's area searching, fire rescue and cooperative delivery in multiple robot to have a large amount of potential application.For example, one group of dolly with different performance moves to destination from departure place, and now, we only need be furnished with sensor to detect dangerous obstacles on part dolly, and so pilotage people is appointed as by these dollies, and all the other dollies are then follower.By verifying the position of dangerous obstacles, pilotage people can form the moving area of a safety, if follower moves in the safety zone formed by pilotage people always, then this group dolly successfully can arrive destination safely.As a kind of special many navigators situation, the containing of multirobot network controls the very big attention obtaining scholars.

Comprising at multirobot controls in research, and a lot of research work is at present devoted to improve the design of control protocol with the stability improving system, and realizes overall exchange by local exchange, thus reaches desirable coherency state.Such as: Mei J, Ferrari-Trecate (Containment control in mobile networks.IEEE Transactions on Automatic Contro1,2008) have studied it for fixing Undirected networks and comprise control problem, proposition stops-walks strategy, thus orders about one group of simple integral individuality and enter into the convex closure be made up of pilotage people.Ziyang Meng, Ren Wei, (Distributed finite-time containment control for multiple lagrangian systems.Proc of American Control Conf.Baltimore:IEEE Press, 2010) discuss with unknown parameter Lagrange system comprise control problem, the self-adaptation proposed with parameter estimator based on relative position and velocity comprises control protocol.In actual applications, network knows from experience the impact being subject to model uncertainty, network transfer speeds and sensor visual scope, is generally oriented communication between individuality.Yongcan Cao, Daniel Stuat (Distributed containment for multiple autonomous vehicles with double-integrator dynamics:algorithns and experiments.IEEE Trans on Control Systems Technology, 2011) study the containing control problem of second order directed networks, propose the time corresponding control protocol that the person of acting as the leader is static or move.Chengjie Xu; Ying Zheng (Necessary and sufficient conditions for distributed containment control of multi-agent systems without velocity measurement, Control Theory & Applications 2014) propose when only knowing leader's positional information in second order multirobot network, application distribution wave filter is estimated neighbours robot speed, thus realizes comprising control.Comprise in control how reasonably to select leader at large-scale multirobot, and conflict-free problem when avoiding scale to increase between machine individual human, it is the interested problem of control engineering Shi Feichang, this control problem has two notable features, one is there is local message transmission between individuality, two is that individuality independently carries out distributed AC servo system, can find out that the topology of networks formed by individuality plays very important effect, the selection of its leader is that colony realizes synchronous precondition, therefore in order to the sync packet of the multi-robot system solved contains control problem, conflict-free problem when needing the structure of the colony of solution emphatically network topology structure and robot scale to increase between machine individual human.

Summary of the invention

The object of the invention is, in order to solve the conflict-free problem between robot individuality, the invention provides collision problem when a kind of can solve robot group scale known based on local message increases between machine individual human and colony's network topology structure controlled comprise control method.

Realizing technical scheme of the present invention is, the distributed multirobot of a kind of base of the present invention comprises control method, is combined by multirobot network with graph theory, and applies the leader follower set in controllability theoretical stage maximum matching algorithm determination directed networks; After leader's set is determined, when the person of acting as the leader is static, according to the acting force relation between leader robot and follower robot and controling parameters, introduce potential-energy function and realize collision prevention, design leader static time comprise control of collision avoidance rate; When the person of acting as the leader moves, for the situation that local message is known, namely part follower is unknowable to leader's speed, for follower designs the velocity estimation of power estimator realization to leader, and then according to the kinetic characteristic of mobile robot and controling parameters, design multiple robot leader dynamic time comprise control of collision avoidance rate.

Said method comprising the steps of:

(1) for the correspondence between directed networks Zhong Ge robot, application controllability theory determines that leader in network and follower gather;

(2) when the person of acting as the leader is static, by the correspondence between leader and neighbours robot, the position control operator of design follower robot;

(3) on the basis of follower's position control operator, introduce potential-energy function, guarantee follower robot in motion process mutually between can realize collision prevention, formed leader static time comprise control of collision avoidance rate;

(4) when the person of acting as the leader is dynamic, by the correspondence between leader and neighbours robot, the position control operator of design follower robot;

(5) time known for local message, namely there is part follower in network topology to the unknowable situation of leader's speed, power estimator is introduced in the control strategy that follower robot formulates, estimate the speed of leader robot, determine the speeds control operator of follower robot;

(6) finally by the position control operator of robot and the weighting of direction controlling operator integrated, and introduce potential-energy function and guarantee mutual collision prevention between individuality in motion process, the distributed of formative dynamics robot comprises control of collision avoidance rate.

The method that described leader and follower gather has:

1) correspondence between the robot in the method representation network of application drawing opinion, and digraph G (A) is converted to bipartite graph H (A) and represents:

Wherein we are considered as multi-robot system the directed networks G that is made up of N number of node, N × N matrix A={ a _ij| i, j ∈ [1, N] } represent the syntople of nodes, if node i can receive the information of j, there is a _ij> 0.Bipartite graph after conversion wherein represent each row of state matrix A and the node set of each row respectively, Γ={ (x _i, x _j) | a _ij≠ 0} represents limit collection.

2) according to bipartite graph maximum matching algorithm, trying to achieve the matched node in network and non-matching node set, is not the network of 0 for a non-matching nodes, the very non-matching nodes of driving node number.

3) driving node in robot network is considered as leader, other node is follower.

The method comprising control of collision avoidance rate when robot leader is static is as follows:

1) the position control operator of follower robot is determined by correspondence between follower robot and its neighbours robot;

The second-order dynamic model of follower is as follows:

${\stackrel{\·}{x}}_i\left(t\right)=v_i\left(t\right),{\stackrel{\·}{v}}_i\left(t\right)=u_i\left(t\right),i=\mathrm{1,2},...,n---\left(1\right)$

A given directional topology network with n robot, application controllability theory determines that follower's set in network and leader gather, and uses F={v respectively ₁, v ₂..., v _mand L={v _m+1, v _m+2..., v _nrepresent follower's set and leader's set.Its position control operator representation is:

$f_r=-\underset{j\∈\mathrm{FUL}}{\mathrm{\Σ}}{a}_{\mathrm{ij}}\left(t\right)\left[(x_i\left(t\right)-x_j\left(t\right)\right]---\left(2\right)$

Wherein a _ijt () is non-negative weight, represent adjacency matrix A=[a _ij] ∈ R ^{n × N}in the item of (i, j) individual correspondence.

2) for introducing potential-energy function between follower robot, avoid colliding in follower robot kinematics; Potential-energy function is expressed as:

$f_c=-{\▿}_{x_i}{V}_{\mathrm{ij}}(x_i,x_j)---\left(3\right)$

V _ij(x _i, x _j) be defined as follows:

$V_{\mathrm{ij}}(x_i,x_j)=\left\{\begin{array}{cc}{k}_{\mathrm{ij}}\frac{1}{x_{\mathrm{ij}}-\frac{1}{2}{d}^2},& \left|\right|x_i-x_j\left|\right|\≤r,\\ 0,& \left|\right|x_i-x_j\left|\right|>r,\end{array}\right.---\left(4\right)$

Wherein, d refers to the minimum safe distance that should keep in the motion process of all robots, and r is the distance of robot energy perception potential-energy function, k _ij=k _ji> 0, x _ijbe defined as follows:

$x_{\mathrm{ij}}=\left\{\begin{array}{cc}\frac{1}{2}{\left|\right|x_i-x_j\left|\right|}^2,& 0<\left|\right|x_i-x_j\left|\right|\&le;r,\\ 0,& \left|\right|x_i-x_j\left|\right|>r,\end{array}\right.$

3) for follower's Robot Design speed Autonomous test item, by position control operator and potential-energy function weighting integrated, formed the person of acting as the leader static time comprise control of collision avoidance rate.

The method comprising control of collision avoidance rate when multiple robot leader is dynamic is as follows:

1) the position control operator of follower robot is determined by correspondence between follower robot and its neighbours robot, and its expression formula is with (2) formula;

2) to there is part follower in active leader network topology to the unknowable situation of leader's speed, power estimator is introduced in the control strategy that follower robot formulates, estimate the speed of leader robot, determine the speeds control operator of follower robot;

The power estimator proposed is expressed as:

${\stackrel{\&CenterDot;}{\hat{v}}}_i=-\mathrm{\&alpha;sgn}\left[\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}(v_j-v_d)\right]---\left(5\right)$

Wherein α > 0, Sgn is sign function i ∈ F, represent that i-th follower is to the estimation of desired speed.V _drepresent the given speed to leader robot.Its speeds control operator representation is:

$f_s=-\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}\left(t\right)\left[(v_i\left(t\right)-{\hat{v}}_i\left(t\right)\right]---\left(6\right)$

3) finally by the position control operator of robot and the weighting of direction controlling operator integrated, and introduce potential-energy function and guarantee mutual collision prevention between individuality in motion process, the distributed of formative dynamics robot comprises control rate.

The invention has the beneficial effects as follows:

(1) graph theory combines with the control of robot group by the present invention, and visual in image shows the network topology and control action relation that are formed between robot location;

(2) the controllability theory in control theory is applied in multi-robot system by the present invention, bipartite graph maximum matching algorithm in the theoretical and graph theory of application controllability proposes and a kind ofly comprises the method how effectively chosen in control and determine that leader follower gathers, determine the topological structure in network, for the setting of control protocol with established precondition for colony realizes sync packet containing control.

(3) the present invention introduces self-defining potential-energy function when design comprises control rate, when multiple follower can realize the conflict-free problem when robot scale increases between machine individual human in motion process effectively.

(4) the present invention is when known for local message, when the person that do not act as the leader is dynamic namely, part follower is in the unknowable situation of leader's velocity information, power estimator is introduced in the control strategy that follower robot formulates, estimate the speed of leader robot, thus realize by local known to overall known, realize synchronous on velocity reversal of robot group.

(5) comprise control rate when the present invention proposes a kind of new person's of acting as the leader Static and dynamic, and demonstrate that this control rate can realize multirobot effectively by second method of Liapunov and Barbalat theorem etc. comprise control of collision avoidance.

(6) the integrated self-adaptation realized for robot location's relationship change of the weighting of position control operator and velocity reversal Control operators, is ensureing that the sync packet basis that colony's network topology structure is controlled realizing robot group contains.

Accompanying drawing explanation

Fig. 1 is bipartite graph maximum matching algorithm determination driving node schematic diagram of the present invention;

Fig. 1 (a) represents network topology figure, and Fig. 1 (b) represents corresponding bipartite graph, and maximum coupling limit collection is looked in Fig. 1 (c) expression, and Fig. 1 (d) expression is determined to drive joint;

Fig. 2 is network topology structure figure of the present invention;

Fig. 3 comprises control schematic diagram when being leader's static state;

Site error final when Fig. 4 is leader's static state;

Control schematic diagram is comprised when Fig. 5 leader is dynamic;

Fig. 6 be leader dynamic time final site error;

Fig. 7 is the step block diagram of the inventive method.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in detail.

The step of the specific embodiment of the invention as shown in Figure 7.

Adopt bipartite graph maximum matching algorithm determination driving node schematic diagram: generally speaking, the target that complex network controls is: by some Controlling vertex input signals chosen, make whole network can be controlled and reach the state of expection.

In modern control theory, system controllability and measurability is the more general concept of reflection input to the control ability of system state.A given linear time invariant control system:

$\stackrel{\&CenterDot;}{x}=\mathrm{Ax}\left(t\right)+\mathrm{Bu}\left(t\right),x\&Element;R^N,u\&Element;R^M,---\left(8\right)$

Wherein A=(a _ij) _{n × N}with B=(b _ij) _{n × M}be called system matrix and input matrix (M≤N).If for any given initial state x (0)=x ₀with final state x _f, all there is control inputs u and Finite time T and make x (T)=x _f, so just claim system (8) to be controlled.Controllability classical filling wants criterion to be corresponding controllability matrix full rank, namely has

We are considered as system (8) state equation of the directed networks G be made up of N number of node now, wherein N × N matrix A={ a _ij| i, j ∈ [1, N] } represent the syntople of nodes, if node i can receive the information of j, there is a _ij> 0.Vector x (t)=(x ₁(t), x ₂(t) ..., x _n(t)) ^trepresent the state of N number of node at moment t, N × Metzler matrix B={b _ij| i ∈ [1, N], j ∈ [1, M] } represent the annexation of node and external control signal, u (t)=(u ₁(t), u ₂(t) ..., u _m(t)) ^tfor M input node is in the state of t, wherein b _ij=1 represents applying signal u on signal node i _j(t).

Asking for of maximum coupling in a network can solve by the mode of bipartite graph, and digraph G (A) is converted to bipartite graph H (A): $H\left(A\right)=(V_A^{+},V_A^{-},\mathrm{\&Gamma;}),$ Wherein $V_A^{+}=\{x_1^{+},...,x_N^{+}\},V_A^{-}=\{x_1^{-},...,x_N^{-}\},$ Represent each row of state matrix A and the node set of each row respectively, Γ={ (x _i, x _j) | a _ij≠ 0} represents limit collection.According to bipartite graph maximum matching algorithm, can, in the hope of the matched node in network and non-matching node, not be the network of 0 for a non-matching nodes, the very non-matching nodes of driving node number.Control a network G (A), only control signal need be inputted non-matching node and ensure that control signal can reach any matched node.

Fig. 1 describes for a robot network process that application bipartite graph matching algorithm asks for driving node in directed networks.Investigate the directed networks of given topological structure shown in Fig. 1 (a), network is converted into bipartite graph structure shown in Fig. 1 (b), investigate the maximum coupling limit of this bigraph (bipartite graph), its maximum coupling limit collection as shown in Fig. 1 (c) dotted line, limit as seen from the figure: 1 ⁺→ 5 ^-, 2 ⁺→ 1 ^-, 3 ⁺→ 7 ^-, 4 ⁺→ 3 ^-5 ⁺→ 3 ^-, 6 ⁺→ 5 ^-, 7 ⁺→ 5 ^-for maximum coupling limit collection; Node 1,3,5,7 matching section is counted, and 2,4,6 is non-matching nodes, and namely as shown in Fig. 1 (d), hollow node 2,4,6 is driving node, whole network just can be made controlled to the control of these three nodes.

First our the controllability theory shown in application drawing 1 determines the leader follower intersection in network, and known: 2,4,6 nodes are leader's intersection, 1,3,5,7 nodes are follower's intersection.Its topological structure can be expressed as form shown in Fig. 2, and we consider the cooperation control of this network in two-dimensional space, i.e. x _i∈ R ², v _i∈ R ², u _i∈ R ², i=1,2 ..., n.

Synchronous containing phenomenon when leader is static as shown in Figure 3, the control of collision avoidance problem that comprises based on static leader drives m follower robot exactly, in the Convex range making them move into be formed with n-m leader robot, and final speed all trends towards zero.Meanwhile, in the process of whole motion, collision avoidance can be kept.

By recitation of steps above, for each follower robot i ∈ F, design Distributed Control protocols is as follows:

$u_i\left(t\right)=-\mathrm{\&beta;}{v}_i\left(t\right)-{\&dtri;}_{x_i}{V}_{\mathrm{ij}}(x_i,x_j)-\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}\left(t\right)\left[\mathrm{\&gamma;}(x_i\left(t\right)-x_j\left(t\right)\right],---\left(9\right)$

Wherein, 1 > γ > 0, β > 0, a _ijt () is non-negative weight, represent adjacency matrix A=[a _ij] ∈ R ^{n × N}in the item of (i, j) individual correspondence.The leader robot considered is static, and the state of final follower robot also can be static, so v here _it () finally can level off to zero.In addition, V _ij(x _i, x _j) define under as Suo Shi preceding formula (4).

For control law (9), item is the collision avoidance for ensureing colony, another for driving follower position constantly can level off to leader to realize final comprising behavior, and β v _it () then represents the speed dissipation in time of robot.

Fig. 3 is the movement locus of each node in 0-50s.As can be seen from Figure 3 all follower finally can both enter into triangle, and namely all follower finally converge in the static convex closure that leader forms.Robot 3 affects by 2,4,5 robots as shown in Figure 2, and its track approximately moves and to move in 2,4,5 robot convex closures.Robot 5 affects by robot 1,4,6,7, also moves in robot 1,4,6,7 convex closure.Robot 1 affects by robot 2, robot 7 affects by robot 3, and respectively to 2,3 robot motions, and due to the existence of potential-energy function, can realize collision prevention in motion process.Fig. 4 is its site error curve, all converges to as shown in Figure 4 in the convex closure of leader's composition after robot motion 6s.

Synchronous containing phenomenon when leader is dynamic is as shown in Figure 5: dynamic leader contains conflict-free problem, first determines that leader follower gathers with controllability theory, and then driving leader is with constant initial velocity to the motion of controlled dbjective state, meanwhile, for follower arranges suitable control protocol, guarantee that follower can enter in the convex closure formed by leader, and realize collision prevention in motion process.Consider not to be the velocity information that each follower can obtain leader in colony, follower robot can by the information interaction with surrounding neighbours robot, makes speed level off to the speed of leader, enables final motion state reach consistent.Under such circumstances, for the network topology structure of active leader, introduce power estimator in our control strategy for follower's formulation, estimate the speed of leader robot, the power estimator of proposition is as follows:

${\stackrel{\&CenterDot;}{\hat{v}}}_i=-\mathrm{\&alpha;sgn}\left[\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}(v_j-v_d)\right]---\left(5\right)$

Wherein: α > 0; I ∈ F, represent that i-th follower is to the estimation of desired speed.

Because leader robot has unanimously constant speed v _d, the relative coordinate of different leader robot remains unchanged.Adopt the method for velocity estimation to carry out " distributed " information interaction for follower robot, the control law that we provide network dynamic (7) is as follows:

v _i(t)＝v _d(t), i∈L,

$\begin{array}{c}{u}_i\left(t\right)=-\mathrm{\&gamma;sgn}\{{\&dtri;}_{x_i}{V}_{\mathrm{ij}}(x_i,x_j)+\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}\{\mathrm{\&beta;}[x_i\left(t\right)-x_j\left(t\right)]\\ +[v_i\left(t\right)-{\hat{v}}_i\left(t\right)]\left\}\right\},i\&Element;F\end{array}---\left(14\right)$

Observation Fig. 5 is known, and in network topology structure Fig. 2, without contacting directly between robot 7 and leader, when the person of acting as the leader moves, its speed is drawn by power estimator.From Fig. 5 and Fig. 6, robot 1 and robot 7 are close in the convex closure that distance leader forms, and its position is in the front, direction that leader runs, therefore can rapidly converge in the convex closure of leader's composition, as shown in Figure 5, robot 1 is nearer than robot 7 apart from convex closure distance, therefore robot 1 speed of convergence is the fastest.Can only body 3 due to natively in the convex closure of leader's composition, but its direction of motion and convex closure move equidirectional, therefore also can converge to equilibrium position in convex closure faster.Robot 5 is owing to moving in leader in the same way at leader rear therefore restraining the slowest.From Fig. 2 and Fig. 5, robot 5 affects by the attraction of robot 3, and robot 7 affects by the attraction of robot 3, but makes can realize collision prevention between these three robots due to the existence of potential-energy function.All converge to after robot motion 30s as shown in Figure 6 in the convex closure of leader's composition.Analogous diagram 6 gives the signal of the site error in robot kinematics.

The various replacements that concept of the present invention can carry out, change and amendment, these are replaced, change and amendment should not got rid of outside the protection domain of invention.

Claims (4)

1. distributed multirobot comprises a control of collision avoidance method, it is characterized in that, multirobot network combines with graph theory by described method, and applies the leader follower set in controllability theory and bipartite graph maximum matching algorithm determination directed networks; After leader's set is determined, when the person of acting as the leader is static, according to the acting force relation between leader robot and follower robot and controling parameters, introduce potential-energy function and realize collision prevention, design robot leader static time comprise control of collision avoidance rate; When the person of acting as the leader moves, for the situation that local message is known, namely part follower is unknowable to leader's speed, for follower designs the velocity estimation of power estimator realization to leader, and then according to the kinetic characteristic of mobile robot and controling parameters, design multiple robot leader dynamic time comprise control of collision avoidance rate;

Said method comprising the steps of:

(1) for the correspondence between directed networks Zhong Ge robot, application controllability theory determines that leader in network and follower gather;

(2) when the person of acting as the leader is static, by the correspondence between leader and neighbours robot, the position control operator of design follower robot;

(3) on the basis of follower's position control operator, introduce potential-energy function, guarantee follower robot in motion process mutually between can realize collision prevention, formed leader static time comprise control of collision avoidance rate;

(4) when the person of acting as the leader is dynamic, by the correspondence between leader and neighbours robot, the position control operator of design follower robot;

(5) time known for local message, namely there is part follower in network topology to the unknowable situation of leader's speed, power estimator is introduced in the control strategy that follower robot formulates, estimate the speed of leader robot, determine the speeds control operator of follower robot;

(6) finally by the position control operator of robot and the weighting of direction controlling operator integrated, and introduce potential-energy function and guarantee mutual collision prevention between individuality in motion process, the distributed of formative dynamics robot comprises control of collision avoidance rate.

2. the distributed multirobot of one according to claim 1 comprises control of collision avoidance method, it is characterized in that, the method that the described leader of determination and follower gather is:

(1) correspondence in the method representation network of application drawing opinion between robot, and digraph G (A) is converted to bipartite graph H (A) and represents:

Wherein multi-robot system is considered as the directed networks G be made up of N number of node, N × N matrix A={ a _ij| i, j ∈ [1, N] } represent the syntople of nodes, if node i can receive the information of j, there is a _ij> 0; Bipartite graph after conversion wherein represent each row of state matrix A and the node set of each row respectively, Γ={ (x _i, x _j) | a _ij≠ 0} represents limit collection;

(2) according to bipartite graph maximum matching algorithm, try to achieve the matched node in network and non-matching node set, be not the network of 0 for a non-matching nodes, driving node number is non-matching nodes;

(3) driving node in robot network is considered as leader, other node is follower.

3. multirobot sync packet according to claim 1 is containing control of collision avoidance method, it is characterized in that, the method comprising control of collision avoidance rate when described robot leader is static is as follows:

(1) the position control operator of follower robot is determined by correspondence between follower robot and its neighbours robot;

The second-order dynamic model of follower is as follows:

${\stackrel{\&CenterDot;}{x}}_i\left(t\right)=v_i\left(t\right),{\stackrel{\&CenterDot;}{v}}_i\left(t\right)=u_i\left(t\right),$ i＝1,2,...,n

Wherein x is the position quantity of robot, and v is the speed amount of robot, and u is the control inputs of robot.For the directional topology network having arbitrarily n robot, application controllability theory determines that the follower in network gathers and leader's set, uses F={v respectively ₁, v ₂..., v _mand L={v _m+1, v _m+2..., v _nrepresent follower's set and leader's set; Its position control operator representation is:

$f_r=-\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}\left(t\right)\left[(x_i\left(t\right)-x_j\left(t\right)\right]$

Wherein a _ijt () is non-negative weight, represent adjacency matrix A=[a _ij] ∈ R ^{n × N}in the item of (i, j) individual correspondence;

(2) for introducing potential-energy function between follower robot, avoid colliding in follower robot kinematics; Potential-energy function is expressed as:

$f_c=-{\&dtri;}_{x_i}{V}_{\mathrm{ij}}(x_i,x_j)$

V _ij(x _i, x _j) be defined as follows:

$V_{\mathrm{ij}}(x_i,x_j)=\left\{\begin{array}{cc}{k}_{\mathrm{ij}}\frac{1}{x_{\mathrm{ij}}-\frac{1}{2}{d}^2},& \left|\right|x_i-x_j\left|\right|\&le;r,\\ 0,& \left|\right|x_i-x_j\left|\right|>r,\end{array}\right.;$

Wherein, d refers to the minimum safe distance that should keep in the motion process of all robots, and r is the distance of robot energy perception potential-energy function, k _ij=k _ji> 0, x _ijbe defined as follows:

$x_{\mathrm{ij}}=\left\{\begin{array}{cc}\frac{1}{2}{\left|\right|x_i-x_j\left|\right|}^2,& 0<\left|\right|x_i-x_j\left|\right|\&le;r,\\ 0,& \left|\right|x_i-x_j\left|\right|>r,\end{array}\right.;$

(3) for follower's Robot Design speed Autonomous test item, by position control operator and potential-energy function weighting integrated, formed the person of acting as the leader static time comprise control of collision avoidance rate.

4. multirobot sync packet according to claim 1 is containing control of collision avoidance method, and it is characterized in that, the method comprising control of collision avoidance rate when described multirobot leader is dynamic is as follows:

(1) the position control operator of follower robot is determined by correspondence between follower robot and its neighbours robot, and its expression formula is:

$f_r=-\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}\left(t\right)\left[(x_i\left(t\right)-x_j\left(t\right)\right];$

Wherein a _ijt () is non-negative weight, represent adjacency matrix A=[a _ij] ∈ R ^{n × N}in the item of (i, j) individual correspondence;

(2) to there is part follower in active leader network topology to the unknowable situation of leader's speed, power estimator is introduced in the control strategy that follower robot formulates, estimate the speed of leader robot, determine the speeds control operator of follower robot;

The power estimator proposed is expressed as:

${\stackrel{\&CenterDot;}{\hat{v}}}_i=-\mathrm{\&alpha;sgn}\left[\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}(v_j-v_d)\right];$

Wherein α > 0, Sgn is sign function i ∈ F, represent that i-th follower is to the estimation of desired speed; v _drepresent the given speed to leader robot; Its speeds control operator representation is:

$f_s=-\underset{j\&Element;\mathrm{FUL}}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}\left(t\right)\left[(v_i\left(t\right)-{\hat{v}}_i\left(t\right)\right];$

(3) finally by the position control operator of robot and the weighting of direction controlling operator integrated, and introduce potential-energy function and guarantee mutual collision prevention between individuality in motion process, the distributed of formative dynamics robot comprises control rate.