CN110554709A - Distributed bionic multi-agent autonomous cluster control method - Google Patents
Distributed bionic multi-agent autonomous cluster control method Download PDFInfo
- Publication number
- CN110554709A CN110554709A CN201910843110.7A CN201910843110A CN110554709A CN 110554709 A CN110554709 A CN 110554709A CN 201910843110 A CN201910843110 A CN 201910843110A CN 110554709 A CN110554709 A CN 110554709A
- Authority
- CN
- China
- Prior art keywords
- agent
- alpha
- cluster
- agents
- obstacle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000003795 chemical substances by application Substances 0.000 title claims abstract description 269
- 238000000034 method Methods 0.000 title claims abstract description 35
- 239000011664 nicotinic acid Substances 0.000 title claims abstract description 29
- 238000004891 communication Methods 0.000 claims abstract description 36
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 238000004088 simulation Methods 0.000 claims description 27
- 230000001133 acceleration Effects 0.000 claims description 16
- 230000014509 gene expression Effects 0.000 claims description 9
- 230000004888 barrier function Effects 0.000 claims description 8
- 230000003592 biomimetic effect Effects 0.000 claims 4
- 238000004422 calculation algorithm Methods 0.000 description 68
- 230000006870 function Effects 0.000 description 33
- 238000005381 potential energy Methods 0.000 description 26
- 238000005516 engineering process Methods 0.000 description 13
- 239000011159 matrix material Substances 0.000 description 11
- 230000015572 biosynthetic process Effects 0.000 description 9
- 238000004458 analytical method Methods 0.000 description 8
- 238000000926 separation method Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 6
- 241000251468 Actinopterygii Species 0.000 description 5
- 230000006399 behavior Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 238000009499 grossing Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000007654 immersion Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- RZVHIXYEVGDQDX-UHFFFAOYSA-N 9,10-anthraquinone Chemical compound C1=CC=C2C(=O)C3=CC=CC=C3C(=O)C2=C1 RZVHIXYEVGDQDX-UHFFFAOYSA-N 0.000 description 1
- 241000282461 Canis lupus Species 0.000 description 1
- 241000233805 Phoenix Species 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000004791 biological behavior Effects 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000010494 dissociation reaction Methods 0.000 description 1
- 230000005593 dissociations Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012887 quadratic function Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/0206—Control of position or course in two dimensions specially adapted to water vehicles
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0287—Control of position or course in two dimensions specially adapted to land vehicles involving a plurality of land vehicles, e.g. fleet or convoy travelling
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/10—Simultaneous control of position or course in three dimensions
- G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
- G05D1/104—Simultaneous control of position or course in three dimensions specially adapted for aircraft involving a plurality of aircrafts, e.g. formation flying
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a distributed bionic multi-agent autonomous cluster control method which comprises the following steps of S1, system initialization and initial position and speed determination, S2, for the th agent, positioning neighbor agents in the communication range of the agent, S3, calculation according to the neighbor agents, S4, positioning the positions of surrounding obstacles and calculating the positions and the speeds of the obstacle agents, S5, calculation according to the positions and the speeds of the obstacle agents, S6, calculation of the positions and the speeds of target nodes and then , and S7 calculation of new positions and new speeds.
Description
Technical Field
The invention relates to a cluster control method, in particular to a distributed bionic multi-agent autonomous cluster control method.
Background
With the progress and breakthrough of various technologies, a great number of intelligent devices such as mobile phones, computers, unmanned planes, unmanned ships and the like have appeared in recent years, and these systems can be referred to as intelligent agents. Currently, a single drone alone performs a series of challenging tasks, such as exploration, geographic information gathering, entertainment follow-up, and so on. But the processing power and survivability of the individual drones is still too poor. Based on these problems, many drones have received a great deal of attention and research. Not only unmanned aerial vehicles, many unmanned vehicles and the like are concerned about the same and have received wide attention. The existing unmanned aerial vehicle autonomous clustering technology mainly has the following schemes:
1) Centralized autonomous clustering techniques. The centralized or global management type autonomous clustering technology mainly comprises a central management agent and other agents, and is briefly described by a slave agent. The main tasks of the central management agent are to take charge of global information grasping, information management scheduling, task allocation and flight control of all slave agents. In the technology, both the computing capacity and the storage capacity of the central intelligent agent are greatly required, and meanwhile, in an actual combat platform or a common flight mission, if the central intelligent agent is damaged in a fighting manner or collided, the whole cluster completely loses the self-management capacity and the task execution capacity. It can be seen that the centralized clustering technique is not very realistic in practical scenes.
2) In the existing patent, for example, in chinese invention patent CN109917767A, a distributed unmanned aerial vehicle cluster autonomous management system and a control method are proposed, which mainly teach that an unmanned aerial vehicle autonomous cluster system is mainly composed of those several parts, and the working method of each part and the specific operation flow of the whole cluster are not described in principle for the actual unmanned aerial vehicle autonomous cluster technology. Thus, similar to these patents, only the complete framework and concept of unmanned aerial vehicle autonomous clustering system is tried from the system, but no detailed description is given to the actual distributed autonomous clustering technology of the intelligent agent.
3) In the existing patent, for example, in chinese invention patent CN110069076A, an unmanned aerial vehicle cluster air combat method based on bang wolf trapping behavior of biological behavior is proposed, which is mainly a task-oriented intelligent agent cluster technology, mainly applies the cluster technology to cluster air combat, and does not describe the overall cluster autonomous state control when the cluster has no task, and does not describe in detail what basic configuration the cluster maintains the whole. Other similar drone swarm patents have such problems.
the existing unmanned aerial vehicle autonomous clustering technology mainly has the following defects:
(1) With centralized autonomous clustering, the overall cluster is too weak against attacks, and once the central agent is damaged, the overall cluster collapses. Meanwhile, high requirements are put forward on the computing capacity and the storage capacity of the central intelligent agent, and especially, system energy resources formed by most intelligent agents such as unmanned planes and unmanned ships are very important consideration factors, so that the application value of the centralized system in practice is not high.
(2) the existing autonomous clustering technology mainly only provides a solution on a system level, does not carry out detailed technical explanation on the distributed autonomous clustering technology, cannot carry out practical application, and has low application value in practice.
(3) The existing bionic algorithms consider the specific behaviors of specific biological groups, and do not generally consider three basic characteristics when the biological group moves: "bind, separate and align," is less versatile.
Disclosure of Invention
in order to solve the problems in the prior art, the invention provides a distributed bionic multi-agent autonomous cluster control method.
the invention provides a distributed bionic multi-agent autonomous cluster control method, which comprises the following steps:
S1, initializing the system, and determining an initial position and a speed;
S2, positioning the neighbor agents in the communication range of the ith agent;
S3, calculating according to neighbor agentswherein,Is a control quantity uirepresents the acceleration brought by the neighboring alpha agent of the ith alpha agent, wherein the alpha agent is the actually existing agent;
S4, positioning the positions of surrounding obstacles, and calculating the position and the speed of an obstacle intelligent agent;
S5, calculating according to the position and the speed of the obstacle intelligent bodywherein,Is a control quantity uithe part of the system is represented by acceleration brought by a neighbor beta intelligent agent on the k-th barrier surface of the ith alpha intelligent agent, wherein the beta intelligent agent is a defined fictive intelligent agent with a barrier surface and is used for avoiding barriers;
S6, calculating the position and the speed of the target node, and then calculatingwherein,Is a control quantity uiThe part (b) represents the acceleration brought by the ith alpha agent and the target gamma agent, wherein the gamma agent is a defined task destination point and is replaced by a virtual unmanned aerial vehicle;
And S7, calculating a new position and a new speed.
as a further improvement of the present invention, in step S7, a new position and a new velocity are calculated by the following formulas,
qi(tk+1)＝qi(tk)+(tk+1-tk)pi(tk)
pi(tk+1)＝pi(tk)+(tk+1-tk)ui(tk)
Wherein,
qiindicating the location of the ith alpha agent,
piRepresenting the speed of the ith alpha agent,
uiRepresents the acceleration of the ith agent,
tk+1-tkthe time interval representing a discrete time system, i.e. the simulation time interval.
as a further improvement of the present invention, in step S1, N agents are set, and the control quantity of each agent is uiWherein i ∈ [0, N-1 ]]and the number of alpha agents corresponds to N.
as a further improvement of the present invention, the method further includes step S8, where i is equal to i +1, determining whether i is equal to N, if no, returning to step S2, and if yes, k is equal to k +1, and returning to step S2.
As a further improvement of the present invention, the expression of the control amount of each agent is as follows:
Wherein,Is calculated for all neighbour agents within the communication range of the ith agent, this parameter being set according to three criteria of the bio-cluster, u _ i ^ beta is the amount of obstacle control set according to the obstacle scenario, andthe target control quantity is set under the condition that all groups have tasks;
the specific three control quantity expressions are as follows:
wherein,and isQ is a constant parameter for both η 1 or 2 and v α or β or γrindicating the position of the cluster target point, prThe velocity of the indicated target point is,The location of the indicated obstacle agent,The velocity of the agent is indicated, with n (i, j) and n ^ (i, k) inside as follows.
e is a parameter of sigma _ norm;
the multi-agent autonomous cluster control method is distributed, so that each agent only interacts with neighbor agents in the surrounding communication range of the agent, and the control quantity is calculated.
The invention has the beneficial effects that: by the scheme, the motion control of the multi-agent in multiple scenes can be realized, the application value in practice is high, the method is suitable for most biological groups, and the universality is high.
drawings
Fig. 1 is an example of an alpha grid.
FIG. 2 is a drawing of1A norm function image.
fig. 3 is a distribution of eigenvalues of a laplacian matrix.
FIG. 4 is a flow chart of a distributed bionic multi-agent autonomous cluster control method of the invention.
Fig. 5 is a scenario in which an obstacle is encountered.
Fig. 6 is a scenario 1 simulation result.
Fig. 7 is a scene 2 simulation result.
fig. 8 is a scene 3 simulation result.
Fig. 9 is a schematic diagram of a large-scale multi-unmanned system formation performance system.
Detailed Description
The invention is further described with reference to the following description and embodiments in conjunction with the accompanying drawings.
The invention provides a distributed bionic multi-agent autonomous cluster control method, which mainly comprises the following 4 aspects:
Firstly, defining and background explaining related to a clustering algorithm;
Secondly, a cluster control algorithm principle and an application scene;
Thirdly, problems possibly occurring in practice and a solution corresponding to the algorithm;
And fourthly, potential application.
the following is a detailed principle introduction of the distributed bionic multi-agent autonomous cluster control method of the present invention:
Firstly, defining and background explaining related to a clustering algorithm;
1.1 background description
Three basic criteria for biological clustering:
Through a great amount of detailed observation of biological populations such as ant colony, bird colony, fish colony and the like by biologists, the following three basic flight criteria mainly exist when the biological populations fly freely:
to cluster, combine: always trying to fly towards the centroid of the surrounding neighbor multi-agent.
Separation to avoid mutual collision: if the multi-agent is found to be too close to the neighbor, it is far away.
And (3) speed alignment: it is desirable to maintain a speed match with the surrounding multi-agent.
1.2 associated definitions
the variable names and their specific meanings as designed in this application are as follows:
TABLE 1. arrangement of variables in the text and their specific meanings
(1) Graph and network
in mathematical graph theory, a graph is often defined as a set of points v representing the set of all vertices in the network and a set of edges e representing the set of connecting lines between all vertices, i.e., G ═ v, e. In this context, for one figureDefinitions in addition to these two elements, there are two elements, each of which is the adjacency matrix a ═ a of the graphij]Here a isijThe i-th edge and the j-th edge are connected as represented by ≠ 0. There is also a node parameter xiEach vertex is also represented by a parameter x, which may represent node velocity, voltage, stress, etc., and in the present invention, represents the velocity of the multi-agent.
two further basic parameters are associated with the graph G, one is the degree matrix Delta of the graph, which is a diagonal matrix, the sum of the elements on the diagonal and the elements of each row of the adjacent matrix of the graphOne-to-one correspondence is understood to mean the number of edges to which each vertex is connected. Another parameter is the laplacian matrix L ═ L of the graphij]This is an n × n matrix, which is specifically defined as L ═ Δ -a. The laplacian matrix of the graph has an important property that he must have a eigenvalue of λ1The eigen matrix corresponding to this eigenvalue is an all-1 vector 1n＝(1，…，1)T。
(2) Stable grid
based on the first two distance principles of the three principles of the biological cluster movement, it can be seen that the distances between the clusters can not be too far away from each other to form a cluster, and can not be too close to each other to cause the collision of the multi-agent in the cluster. Therefore, the concept of an alpha grid is introduced here, and if the clusters form the alpha grid, the distances between the various agents inside the clusters are very suitable.
The α grid is defined as: a cluster is an alpha mesh if the distances between any agent and all its neighboring nodes within the cluster are d. Mathematically described is:Here Ni(q) represents a neighbor node of the ith node; the neighbor node indicates that the node is at the ith nodeof the communication range of the node. Common alpha meshes are shown in several of fig. 1.
The alpha mesh defines that when the position of each element in the network is the same with the position of the neighbor node, the network is in a stable state, i.e. the position between each element node in the network is stable, and is suitable. While for the transient state the concept of quasi-alpha meshes is defined, quasi-alpha meshes represent that they are not exactly equal to d but close to it with respect to the alpha meshes.
(3) Function of potential energy
After the alpha grid is defined in section (2), it has been determined at what time the system is in a stable state. In this section, the overall stability of the system is determined by using mathematical expressions, and for the clusters, the distance between the clusters can be connected with energy, and the gravitational potential energy is a very obvious expression. Meanwhile, the lowest point of potential energy can be equivalent to the most stable state of distance, and by means of the idea, the distance stability in the network, namely the lowest point of energy, can be defined. This corresponds to the first two distance criteria in the three principles of biological clustering within the context of section (1) of section 1.1.
thus we only need to define a potential energy function and when the lowest points of potential energy correspond to each other at a distance d, we define the potential energy function of the network as:
Wherein ψ (z) ═ z2equivalent to a quadratic function, indicates that the minimum value of the function is z-0, which is consistent with the previous potential definition. As can be seen from equation (1), the potential of the network is the sum of the potentials of each agent.
later, when the gradient of the potential energy function is used for defining the nodes, collision cannot occur, because the potential energy function defined by the formula (1) is in | | qj-qithe gradient of the function around d is the same, so it is not suitable. Therefore, we need the right formulaThe psi (z) in the child is modified. We introduce a smoothing function, like the rolling function in the design of the communication system, as shown in equation (2).
Also with gradients, it is desirable that equation (1) be conductive. But in which there is a process of norm, as shown in figure 2, l1Norm is not differentiable at zero point, so gradient cannot be obtained, and a new definition method similar to norm needs to be introduced, wherein sigma is introduced_Norm, as shown in equation (3). Equation (3) may represent a normal distance function, and equation (4) represents the gradient of the distance function.
The potential energy function of the final network can be obtained by modifying the gradient of the function by using the smooth function and modifying and defining a new norm to enable the potential energy function to be microminiature.
As shown in the formula (5),first psiα(z) the derivative function is adjusted and then the smoothing function is added to equation (6). In equation (6), phi (0) is defined as 0, which means that the distance between the agent and its neighbors is desiredThe derivative of the potential energy function is also 0 when the distance is d, so that the system tends to stabilize according to the gradient being 0 without further adjustment. So that several parameters of equation (6) satisfy this equationWhile paying attention to the inner ra＝||r||σ，dα＝||d||σAll use the same sigma_Norm.
Next, in a formula sense, it can be seen that ρ is introduced in φ (z)hAfter the (z) function, when the neighbor node of an agent approaches to the agent, due to the relationship of the smooth function, the closer the neighbor node is, the larger the potential energy gradient of the system is, which is defined by the first two principles of the biological cluster and accords with the practical scene, and the relative "combination" principle and the "separation" principle are more important because the collision between agents is more important.
(4) Coherence protocol
In 1.2(1), we refer to a graph where each node has a parameter value x, which may be velocity, electric force, temperature, etc. For a distributed network, it is a very important issue for the network to keep the parameter value x of each node the same according to what evolution rule. For example, if x represents the velocity of each node, then each node x is the same, which means that the cluster is going in one direction, which is just a strong proof that the cluster is formed, and also has the same meaning as that represented by the third criterion "alignment" of organism clusters.
The invention provides a first-order parameter evolution strategy, which is divided into a continuous time model and a discrete time model: the continuous-time model isAnd the discrete-time model is xi(k+1)＝xi(k)+∈ui(k) where e is the step parameter. The above equation can be quickly understood from the speed point of view, and the speed differentiation is the accelerationDegree, if x represents the velocity, then u corresponds to the acceleration of each node. We therefore propose a linear consensus protocol:
Here again, the invention is directed to a in a linear consistency protocolija smoothing function is also added to ensure that the criterion of no collision within the safe distance d is more important, and therefore the gradient is set to be larger. Therefore, as shown in formula (9), for aija smoothing function is added.
aij(q)＝ρh(||qj-qi||σ/rα)，aij(q)∈[0，1]，j≠i (9)
second, cluster control algorithm principle and application scenario
2.1 second order System dynamic model
It is stated from the continuous-time model that the motion of the system is expressed by equation (9), that is, the differential of the motion trajectory of the system is the velocity, and the differential of the velocity is the acceleration. From the perspective of a discrete time system, as shown in equation (10). In the real world, each distributed agent is a continuous time system, but for engineering, it is a discrete system using equation (10) by minimizing the time difference ti+1-tiTo simulate a continuous system.
The invention is a cluster algorithm designed based on the second-order dynamic system, and the initial state is not limited and can be any.
2.2 algorithmic description and flow
As can be seen from 2.1, after the initial position and velocity of each agent are determined, as long as the acceleration control amount u at each time can be determined, evolution of group agents can be performed according to the formula (10). The key to the algorithm is how to determine the control quantity u based on three basic criteria of the biological population.
The algorithm provided by the invention is mainly used for the control quantity of each agent, wherein N agents are set, and the control quantity of each agent is uiWherein i ∈ [0, N-1 ]]Is an integer of (1). As shown in formula (11), the invention is an expression of the control quantity of each agent in the distributed bionic multi-agent autonomous cluster control algorithm provided by the invention.
wherein,Is calculated for all neighbor agents within the communication range of the ith agent, the parameter is set according to three criteria of the bio-cluster, u _ i ^ beta is the barrier control quantity set according to the barrier scenario (as will be mentioned in detail in the 2.3 subsequent algorithm principles), andIt is the target control amount set for the case where the group has a task.
The specific three control quantity expressions are as follows:
wherein,and isFor both η 1 or 2 and v α or β or γ are constant parameters. q. q.srindicating the position of the cluster target point, prThe velocity of the indicated target point.The location of the indicated obstacle agent,the speed of the obstacle agent is shown (the calculation method and principle will be described in detail in section 2.3). While n in this_(i, j) and n ^ (i, k) are shown in equation (16).
the algorithm is distributed, so each agent only interacts with neighbor agents in the communication range around the agent, and the control quantity is calculated. As shown in fig. 4, is the overall flow of the algorithm.
The flow of the algorithm written in the pseudo-machine language is as follows:
For k is 0, k → ∞, k plus 1
For i is 0, i → N, i plus 1
1. And calculating the positions of other agents in the communication range of the intelligent agent, and communicating with the intelligent agent to acquire the speed of the intelligent agent.
2. Calculating the result of a neighbour agent
3. The position of the surrounding obstacle is located and the position and velocity of the obstacle agent (mentioned in section 2.3) is calculated.
4. Computing from obstacle agent position and velocity
5. Calculating the position and velocity of the target node, and then calculating
6. A new position and a new velocity are calculated according to equation (11).
2.3 principle of algorithm
As can be seen from the formula (12), the acceleration control amount defined by the present invention is composed of three parts, u _ i ^ α brought by the neighbor agent, u _ i ^ β brought by the obstacle agent, and u _ i ^ r brought by the target node, and each control amount is set according to 3 basic principles possessed by the biological population. These 3 aspects will also be described in detail below.
(1) Brought by a neighbour agent
There are three basic rules for organism clustering, which have been described in section 1.1.
As can be seen from equation (13),Is determined by two items together, the former item corresponds to the first two items of the three criteria of the biological colony: "bind" and "separate", the latter term corresponding to the third criterion: "aligned". The former term contains only the position parameters and not the speed parameters, and the latter term, in contrast, contains only the speed term. Both the "separation" and "association" criteria are set for the distance between the multi-agents, while the "alignment" criteria are described for the velocity of each other.
for convenience of description, will beThe formula of (A) is divided into two terms ofWhere FJ is the first letter of the separation and combination and D denotes the first letter of the alignment. It can be seen that FJ is the gradient of the network potential energy function of formula (5), that is, the potential energy function represented by the term FJ falls in the fastest direction, that is, the controlled variable brought by FJ is to make the potential energy function of the group move to the lowest point of potential energy, and the lowest point of network potential energy is when the distance between the intelligent agent and the intelligent agent in all communication ranges is kept to be d. According to the definition of the stable network in section (2) of the above 1.2, if the distance between the agents in the network is d, the network forms an α mesh, and a stable cluster network is formed. This is why the term FJ is set as the gradient of the potential energy function of the network.
From the above analysis, it can be seen that the setting of FJ is to move the network to the lowest point of potential energy. And D is a linear consensus protocol in section 1.2 (4) of the coherence protocol. From that conclusion and proof, it can be seen that the term D must eventually bring the network to a consensus state.
from this analysis, it can be seen that based on both FJ and Dit is designed according to three basic principles of biological population.
If in free space and without a target site, the biological population evolves according to equation (17). If no other agent is encountered, the movement follows the initial velocity. If other agents are encountered or the agents are described as detecting that other agents enter the communication range of the agents, the agents start to form a cluster with each other, and if the speeds of the agents are almost the same and the distances between the agents are proper, the agents reach the same speed shortly, and a cluster flight is formed. If the speed difference between each other is too large and the distance is not suitable, the adjustment between each other is made as much as possible, and if they are moved away from each other's communication range during the adjustment, they will not form a cluster finally.
The scenario indicated by equation (17) is also the basic scenario 1 in the second section: and (4) a cluster autonomous behavior evolution scene without a destination node. In the next section, an algorithmic simulation analysis of this scenario will be given.
(2) Brought by obstacle intelligent body
When an obstacle is encountered, it is most important for the robot to recognize the obstacle and treat the obstacle, and here, the obstacle-avoiding problem and the tracking and routing problem are not only the problem. The invention provides an obstacle avoidance idea for considering an obstacle as an intelligent body.
The present invention refers to an actually present agent as an alpha agent, and an obstacle as an agent, and refers to an agent present on the surface of the agent as a beta agent. Because there must not be a collision between clusters, if there is an agent on the surface of the obstacle, the alpha agent that encounters the obstacle must not be too close to the obstacle to avoid a collision. Meanwhile, the speed of the beta intelligent body is set to be the projection direction of the alpha intelligent body on the surface of the obstacle, and according to a 1.2 section (4) consistency protocol, the speed of the alpha intelligent body is gradually consistent with that of the beta intelligent body after a while, so that the alpha intelligent body finally close to the obstacle can fly along the surface of the obstacle, and the purpose of avoiding the obstacle is achieved.
As shown in fig. 5(a), when an α agent encounters a planar obstacle, we set the agent on the surface of the obstacle as β agent in the figure, so that in order to avoid collision, he will not be close to β agent, i.e. not close to the obstacle.
The invention provides a calculation method of beta intelligent bodies on the surfaces of two types of obstacles. One is a plane type obstacle and the other is a spherical obstacle, and the calculation principle is triedThe alpha agent is projected directly to the surface of the obstacle. For a vector having unity normal akAnd passes through point ykThe plane type obstacle of (2) can be calculated by the formula (18) and has a position qiVelocity piThe alpha agent of (a) corresponds to the beta agent. For radius RkLocated at the circle center ykThe spherical obstacle of (2) can be calculated according to the formula (19). The obstacle avoidance thought of projecting to the surface of the obstacle to form the beta intelligent body can also be applied to various other types of obstacles.
Wherein,I is an identity matrix
Wherein u ═ Rk/||qi-yk||，ak＝(qi-yk)/||qi-yk||， (19)
Having thus described the obstacle avoidance scheme of the present invention, equation (14) will now be describedThis control quantity is interpreted. It can be found first thatformally andMuch like it, its practical principle also followsAlso, for convenience of description, we also express equation (14) by letters as in (1), i.e.The beta agent used for obstacle avoidance in the first place can also be a new agent type generated by us, so that the beta agent is also put in a cluster for calculation, and naturally, the network potential energy and the consistency protocol of the beta agent are also calculated. Here we also define the network potential between the alpha and beta agents, as shown in equation (20), where rβ＝||r′||σ，dβ＝||d′||σHere d' < d, because the distance for obstacle avoidance should be smaller than the safe distance to the alpha agent, after all the obstacle is a static object. The principle of equation (20) is the same as the previous definition of network potential for the alpha agent, so the previous term FJ in equation (14)βAnd also the gradient of equation (20), it is also desirable to move the potential energy between the alpha and beta agents to the lowest point to reach a steady state, i.e., the alpha lattice state between the alpha and beta agents. However, it can be seen that equation (21) is slightly different from equation (6), that is, equation (21) is d for zβIn time of no phiβ(dβ) The requirement of 0, namely the announcement (14) only acts when the alpha intelligent agent is about to collide with the beta intelligent agent, in practice, the formula (21) can be changed and modified according to the actual situation, and the formula can be adjusted to be the same as the formula (6) and have no influence.
also the second term DβThe words in the heel (1)Similarly, here we combine the consistency principle in (1) to design the consistency protocol between α and β agents, and also, after a suitable time, the protocol will make the speeds of α and β agents tend to be consistent.
From this analysis, it can be seen that based on FJβAnd DβBoth of these termsIt is designed according to three basic principles of biological population.
if there is an obstructed space and there is no target site, the biological population evolves according to equation (22). The scenario represented by equation (22) is particularly suitable for multi-agent search scenarios under unknown destination nodes, and the algorithm is particularly suitable for multi-drone search.
(3) Brought by the target node
Expressed by the formula (15)although it appears that the form is completely different from that of the equations (13) and (14), the principle of expression is the same. Here we define a gamma agent, which is the target node of the cluster, which is a location coordinate qrAnd a velocity coordinate prThe virtual agent (note that the speed here may be zero). Therefore, the distributed alpha agent is regarded as an agent with an infinite communication range, each distributed alpha agent needs to go to the place where the gamma agent is located, and the influence brought by the gamma agent is quantitatively expressed as u _ i ^ gamma.
Similarly, for the target node, the two terms contained in the target node are the same equations corresponding to the potential energy function and the consistency meaning, only one equation is taken during calculation, and the summation is not needed because each alpha intelligent agent only has one target gamma intelligent agent.
For the scene represented by equation (23), this corresponds to the basic scene 2 in the second section: an autonomous cluster mobile control scenario with a task target point; while equation (12) corresponds to the basic scenario 3 in the second section: and the autonomous cluster mobile control scene is under the obstacle space and provided with the task target point. In the next section, an algorithmic simulation analysis of this scenario will be given.
2.4 Algorithm simulation of three basic scenes
In this section, 3 basic scene simulations are presented in the second section, with the algorithmic equations used for the simulations all coming from the combination of different terms of equation (12). For example, scene 2 is the first itemAnd item IIIAre added and summed. Table 2 shows the amount that none of the following simulations changed.
Table 2. global setting of some underlying simulation variables (1) Cluster autonomous behavior evolution scenario without destination node
Such a scenario corresponds to the controlled variable of equation (17) in section 2.3, the simulation parameter settings are shown in table 3, and the simulation results are shown in fig. 6.
TABLE 3 scene 1 simulation setup
In fig. 6, all nodes represent alpha agents, and the arrow above the node represents the speed direction of the alpha agent. And the black lines formed between alpha agents are their respective neighbor nodes within communication range. As can be seen from fig. 6 (a), the speeds of the people are initially relatively large, the people form a network but are not in the most stable alpha grid form, and the speeds of the people are too large, so that the people can fly away from the communication range of each other as long as the people have not come to form a cluster. With such interaction forces, the velocities of each other are constantly decreasing, and then in step-by-step synergy, it can be seen that the graph at (b) has already begun to constitute a relatively stable alpha grid without drastic changes. As can be seen from the last diagrams (c) and (d), the clusters forming the α mesh do not change significantly and drastically over time, forming a stable cluster structure, while the α agents not in communication range are continuously far away.
(2) Autonomous cluster mobility control scenario with mission target points
Such a scenario corresponds to the controlled variable of equation (23) in section 2.3, the simulation parameter settings are shown in table 4, and the simulation results are shown in fig. 7.
TABLE 4 scene 2 simulation setup
In fig. 7, the light-colored arrow nodes represent the α agent, and the arrows above these nodes represent the velocity direction of the α agent. And the black lines formed between alpha agents are their respective neighbor nodes within communication range. The node with the dark arrow represents the target gamma agent and the dark arrow represents its speed direction. From (a) the initial population is randomly distributed with few clusters to (b), and the population makes up clusters to go to the destination node together, it can be seen that the traction force of u _ i ^ gamma drags the whole population to move towards the target gamma agent. From the graphs of (c) and (d), it can be seen that after the gamma agent is reached, the whole population begins to become a stable alpha grid, and the distances are kept relatively appropriate, and the speeds of all agents are consistent. The simulation here just verifies the reasonability of the control algorithm designed based on the potential energy reduction and the consistency protocol.
(3) Autonomous cluster movement control scene under obstacle space and with target point
such a scenario corresponds to the controlled variable of equation (12) in section 2.2, the simulation parameter settings are shown in table 5, and the simulation results are shown in fig. 8.
TABLE 5 scene 3 simulation setup
In fig. 8, all nodes in (a) represent alpha agents, and the arrows above the nodes represent the speed direction of the alpha agents. And the black lines formed between alpha agents are their respective neighbor nodes within communication range. A large gray circular pattern like that appearing at the beginning of graph (b) represents the obstacle represented by matrix M, the black nodes on the surface of the obstacle are the obstacle beta agents, and the arrows on the black nodes represent the speed direction of the beta agents.
The formation of a relatively stable alpha grid towards the destination node from the top left hand side of the chaotic placement of the multi-agent to the top right is contrary to the results of scenario 2 in (2). By the middle left, the cluster encountered an obstacle, i.e., the beta agent indicating the surface of the obstacle, at which time the simulation results are the same as the theoretical analysis of section (2) in section 4.2.3. Set in view of simulation after an alpha agent encounters a beta agentSo that beta agent will be guaranteed preferentiallyThere is no collision with the alpha agent and since the beta agent is a projection of the alpha agent onto the surface of the obstacle, the alpha agent will soon maintain a velocity with the beta agent (from the consistency analysis in section (2) of 2.3 above), so the alpha agent will be in close proximity to the obstacle surface and fly parallel to the obstacle surface tangent. Then, after the cluster passes all the obstacles, the cluster continues to move towards the target gamma agent in the form of an alpha grid, and the lower right graph shows that the cluster has finally reached the position of the gamma agent and continues to move forward following the virtual gamma agent.
The simulation result is in contrast with the analysis in the previous section 2.3, and the rationality and the feasibility of obstacle avoidance by designing the beta intelligent body on the surface of the obstacle are proved.
Third, problems possibly occurring in practice and corresponding solutions of the algorithm
(1) What do? if only a small number of alpha agents know gamma agent information
In practice, it may not be that all alpha agents know the location and specific information of the target gamma agent, and if only some of the agents know how the entire cluster should behave
The solution presented by the invention herein is to let alpha agents that do not know gamma agent information regard the centroid of alpha agents that know gamma agent information as a gamma agent and then let these alpha agents that do not know information move towards this centroid.
(2) How to do? if nobody knows the navigator information
In practice, it is also possible that no alpha agent knows the information of the gamma agent, at which time the whole cluster should behave? how
The solution is as follows: all alpha agents are operated in the manner of scenario (1) section 2.4, and then if any agent finds information of a target gamma agent in the communication range of the agent, the agent follows the target gamma agent cluster to move.
(3) What if during exercise, communication of the sudden partial alpha agent is interrupted
if during the actual movement, some alpha agents suddenly interrupt communication, that is, they cannot receive the information about how to do the target gamma agent
The solution is as follows: if the information of the target gamma agent is not known, they move according to the control quantity described in the formula (22), which means that the agent has no task and becomes a free target-searching agent.
(4) Influence of coefficient setting on actual results
andThe ratio of coefficients of (a) may have some effect on the alignment status of the clusters. Therefore, in the practical application scene, the coefficient can be adjustedAndTo adjust the weight of the distance rules and alignment rules, e.g. if the weight is to be increasedIf the adjustment is larger, the force resulting from the rule of decreasing potential energy of the system is greater than the force resulting from the alignment rule, i.e. the system is less likely to reach an aligned state to operate. And vice versa if it is toIf the force is reduced, the system is aligned with the rule that the force is larger, and finally the system is more likely to be in an aligned state to move.
Fourth, potential applications and corresponding system principles
based on the theory, the functional application with rich content and diversified forms can be further realized.
1. Target tracking system of large-scale multi-unmanned aerial vehicle (multi-unmanned vehicle or multi-unmanned ship)
From the 2.4-section scene (2) autonomous cluster movement control scene with task target points, the invention provides a dynamic target gamma intelligent idea, which is very suitable for a large-scale target tracking system. First, whether in an actual combat setting or in a simple target tracking system for civilian use, it is a very important issue for mobile type target tracking, where an entire tracking task may end up failing once a single agent loses a target or is hit. The control algorithm can provide effective support if a multi-agent system, such as a multi-unmanned aerial vehicle system, is adopted for the cluster target tracking task. If there is an obstacle in the target tracking system, it is sufficient to adjust the control algorithm to the control amount of equation (12).
The control algorithm provided by the invention is distributed, and the distribution is particularly important for multiple unmanned aerial vehicles and other unmanned systems with limited resources. Meanwhile, the algorithm is simple to implement, and each unmanned system only needs to interact with other unmanned systems and target systems to be tracked within the surrounding communication range of the unmanned system, so that the method has the characteristics of strong robustness, strong attack resistance and less resource consumption.
2. Bionic motion performance of large-scale multi-unmanned aerial vehicle (multi-unmanned vehicle or multi-unmanned ship)
The basis of the formation algorithm for the formation performance application may be based on the control algorithm provided by the present invention. As shown in fig. 9, a flow chart of a distributed formation control is shown. Namely, when no formation change instruction appears, the whole formation performance system can move according to the distributed cluster control algorithm provided by the invention. The FJ of the control quantity in the control algorithm of the invention can be set if the cluster function is not needed, i.e. the multi-unmanned system only needs to keep at the current formation position for waiting for the next formation change instructionγ＝0，FJ＝0，FJβControl when being equal to 0Only part of the coherency protocol remains in the volume, then the entire enqueue position can be made immobile and then wait for the next enqueue instruction at the same time.
3. Cluster motion control algorithm for intelligent game
The movement of large-scale troops in modern large-scale games such as red police and empire era is quite simple and not very intelligent, and more qualified players have deep sense. How to keep the motion of large-scale intelligent agents in the game to be real and enable players to have more immersion and better game experience is important, and the multi-agent control algorithm is particularly important.
According to the multi-agent control algorithm provided by the invention, each role in a large-scale army in a game can be changed into the alpha agent in the invention, a target to be tracked in the game can be set as the gamma agent in the invention, and an obstacle avoidance method in the game is set as the beta agent in the invention, so that the whole motion can be the same as that of a real biological group, the motion is close to the real world, the game experience is stronger, and a player has better game experience.
4. Basic control algorithm of multi-agent cluster search (emergency communication) system
For the distributed control algorithm provided by the invention, part of the content is relatively similar in a cluster search and emergency communication system, namely, the control part can adopt the control algorithm provided by the invention after the system finishes a task.
For the cluster searching task, the whole system can work like the system described by the formula (22), on one hand, the model of the large-scale cluster evolution free cluster is shown, and on the other hand, the cluster obstacle avoidance capability is provided. The whole algorithm engineer is distributed, only the multiple intelligent agents need to be released, and extra expenses do not need to be added. Like the scene simulation in section 4.2.4 (1) for the system described by equation (22), the final overall cluster starts a free cluster search all around, which is especially suitable for large-scale multi-agent requirements, such as multi-drones, unmanned ships, etc.
For emergency communication systems, the energy overhead of the overall communication is at a maximum, at which time it is optimal not to require additional global control overhead if multiple agents (e.g., multiple drones) have been deployed over the nodes requiring communication via remote control or adaptive algorithms. In this time, each unmanned aerial vehicle only needs to be adjusted to the control algorithm provided by the invention, distributed control overhead is realized, each unmanned aerial vehicle has obstacle avoidance capability and clustering capability, and extra monitoring and global control overhead is not needed in the process of personnel. The distributed clustering algorithm provided by the invention plays a particularly important role in resource saving at this time.
5. control system for bionic fish (bird) swarm
On the spring festival of 2019, the Shenzhen dividend shows the performance of a large bionic phoenix brought by the flyers from the Harbin industrial university (Shenzhen) team, and is really resource-consuming. If the control algorithm proposed by the invention is deployed on the large bionic birds, large-scale personnel are not needed to deploy the large bionic birds. For the change of hardware, only laser radar or other positioning equipment is needed to be added, and only the positions of neighbor nodes in the surrounding communication range can be determined, so that free cluster control can be realized.
The method is the same for large-scale fish school bionic performance, and the bionic motion of the self-organization of the fish school can be realized only by adding precisely-positioned (laser radar, binocular camera and the like) to the bionic fish. The bionic autonomous clustering algorithm provided by the invention is particularly suitable for the application scene.
the invention provides a distributed bionic multi-agent autonomous cluster control method, which aims to design a bionic multi-agent cluster control algorithm, and realizes three basic characteristics of biological group motion from the angle of the algorithm, namely 'combination, separation and alignment'. Based on the algorithm, the motion control of the multi-agent in a plurality of scenes can be realized. The invention gives detailed simulation results of three scenes in detail, and the first one can realize cluster motion of multiple intelligent agents in free space. Second, motion control of the clusters under the underlying tasking assignment can be achieved. Third, motion control in the case of an obstructed space can be achieved. The multi-agent can be unmanned aerial vehicle, underwater robot, etc. In recent years, control algorithms for drones, smart cars and many individual agents have advanced very much, but how to emulate each other between agents, control algorithms between agents have been a blank area. The traditional control idea is that a master control machine is added in the whole situation, and a control instruction is issued to each intelligent agent by the whole intelligent agent, so that the control idea does not conform to the actual cluster organism. Or the existing distributed control algorithms do not consider three basic characteristics of the biological population, but only aim-driven algorithm design, for example, for the reason, one idea of the invention is to provide a bionic distributed intelligent control algorithm, namely, each intelligent body only interacts with the intelligent bodies in the communication range of the intelligent body. Meanwhile, the invention also fits the practical situation, and provides some improved schemes of the algorithm aiming at the problems possibly occurring in practice.
from the results of previous biologists' studies on biological populations, biological populations have largely contained three basic states, namely "association, separation, and alignment," respectively. Binding indicates that biological populations are found too far apart from each other, and are close to each other; separation means that the biological populations are found too close to each other and, in order to avoid collisions, will be separated from each other; alignment means that the organisms meet each other, and the parameters are kept as the same as possible, thereby forming a whole. The invention mainly provides a control algorithm meeting three criteria on the basis.
The algorithm provided by the invention is a second-order continuous time system model, and because the algorithm is a distributed system, each agent calculates the acceleration of the agent according to the algorithm and then deduces the speed change and the position change. The acting force mainly applied when the self acceleration is calculated is all from the intelligent bodies in the communication range of the self. The intelligent body acceleration only comprises two items, wherein the first item is mainly the acceleration calculated according to the distance and corresponds to two criteria of 'combination' and 'separation', and the second item is mainly calculated according to the speed of the intelligent body and corresponds to the criterion of 'alignment'.
The distributed bionic multi-agent autonomous cluster control method provided by the invention is not only suitable for unmanned aerial vehicles, but also can be applied to unmanned vehicles, unmanned ships, unmanned underwater vehicles and the like. Secondly, the algorithm proposed by the inventor is a distributed autonomous clustering technology, which is different from the traditional centralized control. Meanwhile, the algorithm is also an autonomous clustering technology formed based on three basic principles of organism clustering. Finally, the algorithm is not only applied to the following 3 scenarios: 1. a cluster autonomous behavior evolution scene without a destination node; 2. an autonomous cluster mobile control scenario with a task target point; 3. an autonomous cluster movement control scenario under an obstructed space and with a target point. And the evolution of the fusion between these three scenarios, are within the contemplation of the present invention. Meanwhile, the invention also considers how the situations that the communication capability is limited in practice, faults occur, the destination node is unknown and the like should be handled.
the invention provides a distributed bionic multi-agent autonomous cluster control method which has the following characteristics:
(1) The distributed bionic multi-agent autonomous cluster control algorithm not only has the functions realized by the algorithm, but also has the following aspects in the algorithm:
a) A steady-state thought defined by the potential energy function and a defined alpha grid stable structure concept.
b) The relationship of potential energy function corresponding to "association" and "dissociation" in the biological colony is used.
c) The proposed consistency protocol and the corresponding relation of the consistency protocol and the organism cluster 'alignment' rule.
d) And a beta intelligent body on the surface of the obstacle is introduced to realize the autonomous cluster obstacle avoidance function.
e) and a virtual gamma intelligent body is introduced to realize the definition and the setting of a target task, so that the tracking of a target and the self-adaptive large-scale bionic cluster can be realized.
(2) For the problems that the algorithm may have in practice and the solution proposed by the invention:
a) A problem that may arise in practice and a corresponding problem (1) in the solution of the present algorithm, if only a part of the agents knows the target y agent, and the corresponding solution.
b) Problems that may arise in practice and in the corresponding solution of the present algorithm (2), if no agent knows the target gamma agent, and the corresponding solution
c) problems that may occur in practice and problems (3) in the solutions corresponding to the present algorithm, if the overall system communication is interrupted during the implementation of the cluster, and the corresponding solutions
d) problems which can occur in practice and problems (4) in the solution corresponding to the algorithm are the solution idea of adjusting the coefficients to adapt to different scene requirements.
(3) For the scenario of algorithm adaptation provided by the invention and the corresponding solution provided:
a) Corresponding formulas and principles for the three basic scenarios in section 2.4 of the present invention, and other possible application scenarios principles set forth therein.
b) for the hardware implementation, the present invention in the fourth section mentions several application scenarios, and the corresponding system principles and solutions required for the hardware implementation
compared with the prior art, the distributed bionic multi-agent autonomous cluster control method provided by the invention has the following beneficial effects:
(1) The distributed cluster control algorithm has the advantages of low resource consumption and low control overhead.
(2) The autonomous clustering algorithm does not need human intervention, only needs to set a program, and can be freely realized in the real world by the multi-agent, namely, the autonomous clustering scheme.
(3) And corresponding control quantity formulas are provided for various different application scenes, so that the system is convenient to implement.
(4) The practicability is high by considering a plurality of factors existing in reality.
(5) The algorithm of the system comes from three basic rules of organism clusters summarized by biologists, is close to reality and high in reality, and if the system is applied to the field of games, players have strong experience and the game immersion sense is good.
(6) The method starts from a real scene based on biological clusters, is not an algorithm developed by taking tasks as guidance, has communication performance, and is suitable for various different task scenes.
the foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (5)
1. A distributed bionic multi-agent autonomous cluster control method is characterized by comprising the following steps:
S1, initializing the system, and determining an initial position and a speed;
S2, positioning the neighbor agents in the communication range of the ith agent;
S3, calculating according to neighbor agentsWherein,Is a control quantity uiRepresents the acceleration brought by the neighboring alpha agent of the ith alpha agent, wherein the alpha agent is the actually existing agent;
S4, positioning the positions of surrounding obstacles, and calculating the position and the speed of an obstacle intelligent agent;
s5, intelligent posture according to obstaclePosition and velocity calculationWherein,Is a control quantity uiThe part of the system is represented by acceleration brought by a neighbor beta intelligent agent on the k-th barrier surface of the ith alpha intelligent agent, wherein the beta intelligent agent is a defined fictive intelligent agent with a barrier surface and is used for avoiding barriers;
S6, calculating the position and the speed of the target node, and then calculatingWherein,Is a control quantity uithe part (b) represents the acceleration brought by the ith alpha agent and the target gamma agent, wherein the gamma agent is a defined task destination point and is replaced by a virtual unmanned aerial vehicle;
and S7, calculating a new position and a new speed.
2. A distributed biomimetic multi-agent autonomous cluster control method as recited in claim 1, wherein: in step S7, a new position and a new velocity are calculated by the following formulas,
qi(tk+1)＝qi(tk)+(tk+1-tk)pi(tk)
pi(tk+1)＝pi(tk)+(tk+1-tk)ui(tk)
Wherein,
qiIndicating the location of the ith alpha agent,
piRepresenting the speed of the ith alpha agent,
uiRepresents the acceleration of the ith agent,
tk+1-tkThe time interval representing a discrete time system, i.e. the simulation time interval.
3. A distributed biomimetic multi-agent autonomous cluster control method as recited in claim 2, wherein: in step S1, N agents are set, and the control amount of each agent is uiWherein i ∈ [0, N-1 ]]And the number of alpha agents corresponds to N.
4. a distributed biomimetic multi-agent autonomous cluster control method as recited in claim 3, wherein: further comprising step S8, i equals i +1, determines whether i is equal to N, if no, returns to step S2, if yes, k equals k +1, returns to step S2.
5. the distributed biomimetic multi-agent autonomous cluster control method of claim 4, wherein:
The expression of the control amount of each agent is as follows:
Wherein,Is calculated for all neighbour agents within the communication range of the ith agent, this parameter being set according to three criteria of the bio-cluster, u _ i ^ beta is the amount of obstacle control set according to the obstacle scenario, andThe target control quantity is set under the condition that all groups have tasks;
The specific three control quantity expressions are as follows:
wherein,and isFor theη1 or 2 and v α or β or γ are constant parameters, qrIndicating the position of the cluster target point, prThe velocity of the indicated target point is,The location of the indicated obstacle agent,The velocity of the agent is indicated, with n (i, j) and n ^ (i, k) inside as follows.
E is a parameter of sigma _ norm;
The multi-agent autonomous cluster control method is distributed, so that each agent only interacts with neighbor agents in the surrounding communication range of the agent, and the control quantity is calculated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910843110.7A CN110554709A (en) | 2019-09-06 | 2019-09-06 | Distributed bionic multi-agent autonomous cluster control method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910843110.7A CN110554709A (en) | 2019-09-06 | 2019-09-06 | Distributed bionic multi-agent autonomous cluster control method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110554709A true CN110554709A (en) | 2019-12-10 |
Family
ID=68739335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910843110.7A Pending CN110554709A (en) | 2019-09-06 | 2019-09-06 | Distributed bionic multi-agent autonomous cluster control method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110554709A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110940985A (en) * | 2019-12-13 | 2020-03-31 | 哈尔滨工程大学 | Multi-UUV tracking and trapping system and method |
| CN110989639A (en) * | 2019-12-31 | 2020-04-10 | 中国电子科技集团公司信息科学研究院 | Underwater vehicle formation control method based on stress matrix |
| CN111123978A (en) * | 2019-12-26 | 2020-05-08 | 一飞智控(天津)科技有限公司 | Position-based cluster unmanned aerial vehicle automatic numbering processing system and method, and unmanned aerial vehicle |
| CN112783132A (en) * | 2021-01-11 | 2021-05-11 | 中国船舶重工集团公司第七二三研究所 | Cognitive cooperation energizing intelligent body unit in unmanned cluster |
| CN112882491A (en) * | 2021-01-14 | 2021-06-01 | 北京航空航天大学 | Multi-agent distributed control method capable of autonomously traversing two-dimensional linear pipeline |
| CN112925345A (en) * | 2021-01-26 | 2021-06-08 | 北京航空航天大学 | Cluster enclosure control method of unmanned combat aircraft imitating wolf hunting behaviors |
| CN113052447A (en) * | 2021-03-15 | 2021-06-29 | 广东博智林机器人有限公司 | Negotiation work order distribution method, device, storage medium and electronic equipment |
| CN113050681A (en) * | 2021-03-11 | 2021-06-29 | 广东工业大学 | Singular group system consistency analysis and control method |
| CN113325852A (en) * | 2021-06-10 | 2021-08-31 | 浙江大学 | Leader follower mode-based control method for formation change of multiple intelligent agents in advancing process |
| CN113391556A (en) * | 2021-08-12 | 2021-09-14 | 中国科学院自动化研究所 | Group distributed control method and device based on role distribution |
| CN113645317A (en) * | 2021-10-15 | 2021-11-12 | 中国科学院自动化研究所 | Loose cluster control method, device, equipment, medium and product |
| CN114429721A (en) * | 2022-01-19 | 2022-05-03 | 北京航空航天大学 | Aircraft safety interval regulation and control method for autonomous operation airway |
| CN114510048A (en) * | 2022-01-30 | 2022-05-17 | 汕头市快畅机器人科技有限公司 | Group enclosure robot |
| CN117762151A (en) * | 2024-02-22 | 2024-03-26 | 北京航空航天大学 | Distributed shape control method for intelligent agent clusters without reference numerals and intelligent agent |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107340784A (en) * | 2017-08-21 | 2017-11-10 | 中国人民解放军军械工程学院 | Unmanned plane cluster control method |
| CN108549407A (en) * | 2018-05-23 | 2018-09-18 | 哈尔滨工业大学（威海） | A kind of control algolithm of multiple no-manned plane collaboration formation avoidance |
| CN108549403A (en) * | 2018-03-30 | 2018-09-18 | 北京润科通用技术有限公司 | A kind of collaborative obstacle avoidance method and device |
| US20190176333A1 (en) * | 2017-12-13 | 2019-06-13 | Disney Enterprises, Inc. | Robot navigation in context of obstacle traffic including movement of groups |
| CN110320929A (en) * | 2019-06-04 | 2019-10-11 | 杭州电子科技大学 | A kind of Autonomous UAV multiple targets follow barrier-avoiding method |
-
2019
- 2019-09-06 CN CN201910843110.7A patent/CN110554709A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107340784A (en) * | 2017-08-21 | 2017-11-10 | 中国人民解放军军械工程学院 | Unmanned plane cluster control method |
| US20190176333A1 (en) * | 2017-12-13 | 2019-06-13 | Disney Enterprises, Inc. | Robot navigation in context of obstacle traffic including movement of groups |
| CN108549403A (en) * | 2018-03-30 | 2018-09-18 | 北京润科通用技术有限公司 | A kind of collaborative obstacle avoidance method and device |
| CN108549407A (en) * | 2018-05-23 | 2018-09-18 | 哈尔滨工业大学（威海） | A kind of control algolithm of multiple no-manned plane collaboration formation avoidance |
| CN110320929A (en) * | 2019-06-04 | 2019-10-11 | 杭州电子科技大学 | A kind of Autonomous UAV multiple targets follow barrier-avoiding method |
Non-Patent Citations (1)
| Title |
|---|
| REZA OLFATI-SABER: "Flocking for Multi-Agent Dynamic Systems: Algorithms and Theory", 《IEEE TRANSACTIONS ON AUTOMATIC CONTROL》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110940985A (en) * | 2019-12-13 | 2020-03-31 | 哈尔滨工程大学 | Multi-UUV tracking and trapping system and method |
| CN111123978A (en) * | 2019-12-26 | 2020-05-08 | 一飞智控(天津)科技有限公司 | Position-based cluster unmanned aerial vehicle automatic numbering processing system and method, and unmanned aerial vehicle |
| CN111123978B (en) * | 2019-12-26 | 2023-08-22 | 一飞智控(天津)科技有限公司 | Automatic numbering processing system and method for clustered unmanned aerial vehicle based on position and unmanned aerial vehicle |
| CN110989639A (en) * | 2019-12-31 | 2020-04-10 | 中国电子科技集团公司信息科学研究院 | Underwater vehicle formation control method based on stress matrix |
| CN110989639B (en) * | 2019-12-31 | 2023-08-11 | 中国电子科技集团公司信息科学研究院 | Underwater vehicle formation control method based on stress matrix |
| CN112783132A (en) * | 2021-01-11 | 2021-05-11 | 中国船舶重工集团公司第七二三研究所 | Cognitive cooperation energizing intelligent body unit in unmanned cluster |
| CN112882491A (en) * | 2021-01-14 | 2021-06-01 | 北京航空航天大学 | Multi-agent distributed control method capable of autonomously traversing two-dimensional linear pipeline |
| CN112925345A (en) * | 2021-01-26 | 2021-06-08 | 北京航空航天大学 | Cluster enclosure control method of unmanned combat aircraft imitating wolf hunting behaviors |
| CN112925345B (en) * | 2021-01-26 | 2022-01-28 | 北京航空航天大学 | Cluster enclosure control method of unmanned combat aircraft imitating wolf hunting behaviors |
| CN113050681A (en) * | 2021-03-11 | 2021-06-29 | 广东工业大学 | Singular group system consistency analysis and control method |
| CN113052447A (en) * | 2021-03-15 | 2021-06-29 | 广东博智林机器人有限公司 | Negotiation work order distribution method, device, storage medium and electronic equipment |
| CN113052447B (en) * | 2021-03-15 | 2023-10-10 | 广东博智林机器人有限公司 | Method and device for dispatching negotiable worksheet, storage medium and electronic equipment |
| CN113325852A (en) * | 2021-06-10 | 2021-08-31 | 浙江大学 | Leader follower mode-based control method for formation change of multiple intelligent agents in advancing process |
| CN113391556B (en) * | 2021-08-12 | 2021-12-07 | 中国科学院自动化研究所 | Group distributed control method and device based on role distribution |
| CN113391556A (en) * | 2021-08-12 | 2021-09-14 | 中国科学院自动化研究所 | Group distributed control method and device based on role distribution |
| CN113645317A (en) * | 2021-10-15 | 2021-11-12 | 中国科学院自动化研究所 | Loose cluster control method, device, equipment, medium and product |
| CN113645317B (en) * | 2021-10-15 | 2022-01-18 | 中国科学院自动化研究所 | Loose cluster control method, device, equipment, medium and product |
| CN114429721B (en) * | 2022-01-19 | 2022-12-09 | 北京航空航天大学 | Aircraft safety interval regulation and control method for autonomous operation airway |
| CN114429721A (en) * | 2022-01-19 | 2022-05-03 | 北京航空航天大学 | Aircraft safety interval regulation and control method for autonomous operation airway |
| CN114510048A (en) * | 2022-01-30 | 2022-05-17 | 汕头市快畅机器人科技有限公司 | Group enclosure robot |
| CN117762151A (en) * | 2024-02-22 | 2024-03-26 | 北京航空航天大学 | Distributed shape control method for intelligent agent clusters without reference numerals and intelligent agent |
| CN117762151B (en) * | 2024-02-22 | 2024-04-26 | 北京航空航天大学 | Distributed shape control method for intelligent agent clusters without reference numerals and intelligent agent |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110554709A (en) | Distributed bionic multi-agent autonomous cluster control method | |
| Altan | Performance of metaheuristic optimization algorithms based on swarm intelligence in attitude and altitude control of unmanned aerial vehicle for path following | |
| CN111552314B (en) | Self-adaptive formation tracking control method for multiple unmanned aerial vehicles | |
| CN104407619B (en) | Multiple no-manned plane under uncertain environment reaches multiple goal approachs simultaneously | |
| Zhu et al. | Task assignment and path planning of a multi-AUV system based on a Glasius bio-inspired self-organising map algorithm | |
| CN112180967B (en) | Multi-unmanned aerial vehicle cooperative countermeasure decision-making method based on evaluation-execution architecture | |
| CN107367944A (en) | A kind of cluster control method towards multi-agent system | |
| CN113900445A (en) | Unmanned aerial vehicle cooperative control training method and system based on multi-agent reinforcement learning | |
| Chang et al. | Coactive design of explainable agent-based task planning and deep reinforcement learning for human-UAVs teamwork | |
| Wang et al. | Improving maneuver strategy in air combat by alternate freeze games with a deep reinforcement learning algorithm | |
| CN111443728B (en) | Chaos wolf optimization-based unmanned aerial vehicle formation control method | |
| CN110597059B (en) | Large-leaved dogwood group type intelligent group dynamic network topology construction method facing unmanned system | |
| CN110347181B (en) | Energy consumption-based distributed formation control method for unmanned aerial vehicles | |
| CN113759935B (en) | Intelligent group formation mobile control method based on fuzzy logic | |
| CN111077909B (en) | Novel unmanned aerial vehicle self-group self-consistent optimization control method based on visual information | |
| Elmi et al. | Multi-objective grasshopper optimization algorithm for robot path planning in static environments | |
| CN114492805A (en) | Air combat maneuver decision design method based on fuzzy reasoning | |
| CN113434966B (en) | Elastic force field-based bee colony collision avoidance control method | |
| Chen et al. | Deep reinforcement learning based strategy for quadrotor UAV pursuer and evader problem | |
| CN114527784A (en) | Unmanned aerial vehicle cluster overall obstacle avoidance method based on behavior control method | |
| CN113985915A (en) | High-order group system multi-leader clustering formation tracking control method | |
| CN114518772A (en) | Unmanned aerial vehicle swarm self-organization method in rejection environment | |
| CN110162084B (en) | Formation control method of flying missile cluster system based on consistency theory | |
| CN112000070A (en) | Heterogeneous cluster cooperative motion planning method | |
| CN111007848A (en) | Multi-agent cooperative operation control method based on bounded space |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191210 |
|
| RJ01 | Rejection of invention patent application after publication |