A cluster control method based on Henneberg constraints only in azimuth

Abstract

The present invention discloses a cluster control method based on the Henneberg constraint mode under azimuth only. First, identities are assigned to all agents in the cluster in the field of unmanned aerial vehicles and unmanned vehicles, and the leader, the first follower and the remaining followers are determined. The leader and the first follower move autonomously according to the task requirements. Each agent determines all its neighboring agents that are located. The agent obtains positioning information based on the relative azimuth of the neighboring agents, obtains the error to the target position in its own coordinate system, inputs the target error into the motion control law, and completes the formation following behavior. The method of the present invention greatly reduces the demand for detection capabilities, simplifies hardware equipment, reduces costs, and has good stability and robustness. The formation can be expanded on a large scale without being restricted by communications and sensors. The newly added follower agents do not generate additional information requirements for the central node, and are suitable for large-scale cluster area coverage under azimuth measurement only.

Description

Technical Field

[0001] The invention belongs to the field of cluster intelligent control, and in particular relates to a cluster control method based on a Henneberg constraint mode under azimuth only. Background Art

[0002] Unmanned aerial vehicles or unmanned vehicles, which have certain autonomous mission capabilities, have a higher advantage over humans when performing high-risk tasks. Such tasks usually involve scanning and covering a certain area, and often have time-limited requirements. Therefore, multi-agent operations can better meet the above requirements. Compared with single-agent operations, the agent platform in the multi-agent system can only obtain the target information of other agents within a certain range through the sensors and communication equipment carried on it. This information includes: azimuth measurement information relative to its own coordinate system, and the identity information of the other party. The meaning of these unmanned agent platforms performing regional coverage is that during their movement, they use sensors to perform a continuous and complete sweep operation on all areas within the mission area. Such operations include but are not limited to: searching for survivors in unknown wild environments after disasters, dangerous exploration of dangerous areas that are not suitable for people to enter, and plant protection operations in farmland and mountainous areas. The multi-agent system can complete a larger area coverage in the same time by using its parallel characteristics, which has certain advantages in task efficiency and system robustness. The difference between cluster coverage and single-machine coverage is that when the coverage range of each agent in the cluster covers the target area, the cluster is considered to have completed coverage of the area, so it has a higher task efficiency than single-machine coverage. However, when multiple agents work in the same area at the same time, in order to avoid route conflicts and repeated operations, they need to communicate and coordinate with each other. The leader-follower framework is one of the methods in the field of cluster control, which is suitable for tasks that require multiple agents to form a formation.

[0003] In practical applications, especially when obstacle avoidance or mission planning is required in complex environments, the formation usually needs to be scaled or rotated as a whole. Traditional positioning methods include positioning based on global position measurement, positioning based on displacement measurement, and positioning based on distance measurement. These three methods cannot meet the above task requirements well and require a large amount of computing power for a single agent. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a cluster member that only needs to determine the relative positions of other members within its detection range and the boundaries of the area to be covered. By utilizing the recursive neighboring agent structure, each agent only needs to obtain the relative azimuth measurement information of a limited number of neighboring agents, thereby realizing the motion control of the cluster according to the constraints and ultimately completely covering the area of ​​unknown shape in advance. The steps are simple and the system has good robustness against member failure under the corresponding task allocation algorithm.

[0005] The technical solution adopted by the present invention is: a cluster control method based on the Henneberg constraint method under azimuth only, and the specific steps are as follows:

[0006] S1. Assign identities to all agents in the cluster of drones and unmanned vehicles, determine the leader, the first follower, and the remaining followers. The sensors and communication equipment carried by each member can determine the relative azimuth measurement information and identity information of other members within its detection range;

[0007] S2, the leader and the first follower move autonomously according to task requirements;

[0008] S3, each agent determines all its neighboring agents;

[0009] S4, the agent obtains positioning information based on the relative azimuth of the neighboring agent and obtains the error to the target position in its own coordinate system;

[0010] S5. The agent inputs the motion control law according to the target error and completes the formation following behavior.

[0011] Furthermore, in step S3, the process of determining three agents with different identities is as follows:

[0012] The remaining followers 1 take the leader and the first follower as neighboring agents; the remaining followers 2 take the first follower and follower 1 as neighboring agents; the remaining followers i take the remaining followers (i-1) and the remaining followers (i-2) as neighboring agents;

[0013] Each agent can obtain the identity information and relative azimuth measurement information of other agents within the detection range of its own sensors;

[0014] The starting positions of the cluster members at the beginning of the task are close to each other at a reasonable distance and are located in the task area to be covered. The definition of a reasonable distance is that an agent can detect its neighboring agents and form a communication topology relationship.

[0015] Furthermore, in step S4, the leader and the first follower follow their task instructions to move. The remaining followers first determine their neighboring agents. For the second follower, its neighboring agents are the leader and the first follower; when n>2, for the nth follower, its neighboring agents are determined to be the (n-1)th follower and the (n-2)th follower;

[0016] Agent n can obtain the angle between its own forward direction and the global coordinate system as θ asn , the two neighboring agents are in the two-dimensional space coordinate system C of the nth follower n The relative angle θ snsn-1 and θ snsn-2, set the agent distance value ρ specified by the system s Then, the following formula is used to calculate the agent n in the (n-2)th follower two-dimensional space coordinate system C fn-2 Next current position n =(x n ,y n ) T , where T represents the matrix transpose, (x n ,y n ) represents the two-dimensional coordinate value at time n:

[0017]

[0018] At this time, according to the original system settings of the nth follower, its target position s in the same coordinate system is obtained t , according to the calculated s n , calculate the error value relative to the target position at this time

[0019] Further, in step S5, The design control rate is:

[0020]

[0021] Among them, ω fi is the direction the agent is pointing to, u fi is the speed target of the agent at this time. Under the constraints of the two, the follower agent converges to the expected position in the formation; K ai is the amplitude parameter, which, combined with the physical meaning of the system, is the maximum movement speed allowed by the follower; K fi It is a frequency parameter. By properly adjusting its parameter value, the control effect of the system can be adjusted.

[0022] The area to be covered is an irregular connected space surrounded by the outer boundary contour of the cluster task area. The process of the formation cluster task meets the following conditions:

[0023] A: The shape of the area to be covered is unknown to all members of the cluster before the task;

[0024] B: The narrowest part of the area to be covered allows the formation to pass through in the smallest state without causing collision;

[0025] C: The first follower determines the relative angle direction of its own target position relative to the leader in the positive direction of its movement according to the needs during the task. This position determines the overall movement direction of the formation shape;

[0026] D: The first follower determines the relative distance of its own target position relative to the leader in the positive direction of its movement according to the needs during the task. This position determines the overall size of the formation shape;

[0027] E: During the normal progress of the task, the distance between each agent and its neighboring agents is subject to an upper limit constraint, which ensures that the agent and its neighboring agents can maintain detection and communication, thereby ensuring the overall integrity of the formation;

[0028] F: The distance between any two agents is subject to a lower limit constraint, so that at every moment, the two agents maintain a safe distance, that is, at any moment, any two agents in the formation will not collide with each other;

[0029] G: The distance between any agent and the boundary is subject to a lower limit constraint, so that at every moment, the two maintain a safe distance, that is, at any moment, no agent in the formation will collide with the boundary.

[0030] Beneficial effects of the present invention: The method of the present invention assigns identities to all agents in the cluster in the field of drones and unmanned vehicles, determines the leader, the first follower and the remaining followers, the leader and the first follower move autonomously according to the task requirements, each agent determines all its own positioned neighboring agents, the agent obtains positioning information based on the relative azimuth of the neighboring agent, obtains the error to the target position in its own coordinate system, inputs the target error into the motion control law, and completes the formation following behavior. The method of the present invention greatly reduces the demand for detection capabilities, simplifies hardware equipment, reduces costs, and has good stability and robustness. The formation can be expanded on a large scale without being restricted by communications and sensors. The newly added follower agents do not generate additional information requirements for the central node, and are suitable for large-scale cluster area coverage under azimuth measurement only. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The present invention is a flow chart of a cluster control method based on the Henneberg constraint mode under azimuth only.

[0032] Figure 2 The present invention is a schematic diagram of a wake-up task process based on azimuth only in an embodiment of the present invention.

[0033] Figure 3 It is a schematic diagram of expanding the size of a formation through Henenberg constraints in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0035] like Figure 1As shown, the cluster control method based on the Henneberg constraint mode under azimuth only has the following specific steps:

[0036] S1. Assign identities to all agents in the cluster of drones and unmanned vehicles, determine the leader, the first follower, and the remaining followers. The sensors and communication equipment carried by each member can determine the relative azimuth measurement information and identity information of other members within its detection range;

[0037] S2, the leader and the first follower move autonomously according to task requirements;

[0038] S3, each agent determines all its neighboring agents;

[0039] S4, the agent obtains positioning information based on the relative azimuth of the neighboring agent and obtains the error to the target position in its own coordinate system;

[0040] S5. The agent inputs the motion control law according to the target error and completes the formation following behavior.

[0041] In this embodiment, the roles of the cluster members include a leader, a first follower and other followers, a total of three types, and the sensors and communication devices carried by each member can determine the relative azimuth information and identities of other members within its detection range.

[0042] In this embodiment, in step S3, the process of determining three agents with different identities is as follows:

[0043] The remaining followers 1 take the leader and the first follower as neighboring agents; the remaining followers 2 take the first follower and follower 1 as neighboring agents; the remaining followers i take the remaining followers (i-1) and the remaining followers (i-2) as neighboring agents;

[0044] Each agent can obtain the identity information and relative azimuth measurement information of other agents within the detection range of its own sensors.

[0045] The starting positions of the cluster members at the beginning of the task are close to each other at a reasonable distance and are located in the task area to be covered. The definition of a reasonable distance is that an agent can detect its neighboring agents and form a communication topology relationship.

[0046] After the identity of each agent is determined in step S3, in step S4, the leader and the first follower follow their task instructions to move. The remaining followers first determine their neighboring agents. For the second follower, its neighboring agents are the leader and the first follower; when n>2, for the nth follower, its neighboring agents are determined to be the (n-1)th follower and the (n-2)th follower.

[0047] Agent n can obtain the angle between its own forward direction and the global coordinate system as θ asn , the two neighboring agents are in the two-dimensional space coordinate system C of the nth follower n The relative angle θ snsn-1 and θ snsn-2 , set the agent distance value ρ specified by the system s Then, the following formula is used to calculate the agent n in the (n-2)th follower two-dimensional space coordinate system C fn-2 Next current position n =(x n ,y n ) T , where T represents the matrix transpose, (x n ,y n ) represents the two-dimensional coordinate value at time n:

[0048]

[0049] At this time, according to the original system settings of the nth follower, its target position s in the same coordinate system is obtained t , according to the calculated s n , calculate the error value relative to the target position at this time

[0050] In this embodiment, in step S5, The design control rate is:

[0051]

[0052] Among them, ω fi is the direction the agent is pointing to, u fi is the speed target of the agent at this time. Under the constraints of the two, the follower agent converges to the expected position in the formation; K ai is the amplitude parameter, which, combined with the physical meaning of the system, is the maximum movement speed allowed by the follower; K fi It is a frequency parameter. By properly adjusting its parameter value, the control effect of the system can be adjusted.

[0053] The area to be covered is an irregular connected space surrounded by the outer boundary contour of the cluster task area. The process of the formation cluster task meets the following conditions:

[0054] A: The shape of the area to be covered is unknown to all members of the cluster before the task;

[0055] B: The narrowest part of the area to be covered allows the formation to pass through in the smallest state without causing collision;

[0056] C: The first follower determines the relative angle direction of its own target position relative to the leader in the positive direction of its movement according to the needs during the task. This position determines the overall movement direction of the formation shape;

[0057] D: The first follower determines the relative distance of its own target position relative to the leader in the positive direction of its movement according to the needs during the task. This position determines the overall size of the formation shape;

[0058] E: During the normal progress of the task, the distance between each agent and its neighboring agents is subject to an upper limit constraint, which ensures that the agent and its neighboring agents can maintain detection and communication, thereby ensuring the overall integrity of the formation;

[0059] F: The distance between any two agents is subject to a lower limit constraint, so that at every moment, the two agents maintain a safe distance, that is, at any moment, any two agents in the formation will not collide with each other;

[0060] G: The distance between any agent and the boundary is subject to a lower limit constraint, so that at every moment, the two maintain a safe distance, that is, at any moment, no agent in the formation will collide with the boundary.

[0061] The present invention proposes a method for controlling and expanding a cluster by relying only on azimuth measurement and using Henneberg constraints for a multi-agent cluster system, so that the azimuth positioning method alone optimizes the calculation scale of the cluster algorithm when scaling and rotating the formation, and reduces the demand for hardware.

[0062] Figure 2 The figure shows the implementation process of a typical cluster cooperative control method based on Henneberg constraints under azimuth measurement only. In the figure, the black solid triangle represents the leader, the black solid square represents the first follower, and the black solid prototype represents the remaining followers. In the experiment, the intelligent cluster is set to a pentagonal formation. Figure 2 The positions of the overall formation of the team at three moments in the mission and the trajectories of the movements of each agent during the march are shown. The formation at the second moment shows the application of controlling the scale of the formation under azimuth measurement only. When there are obstacles in the covered area or the area is restricted by terrain, the formation can be scaled to pass through. The position of the formation at the third moment shows the application of controlling the rotation of the formation under azimuth measurement only. When the target area presents an irregular path or sweeps back and forth across the area, the formation needs to complete this movement as a whole.

[0063] Figure 3The process of using Henneberg constraints to expand the number of multi-agent formations is illustrated. When the number of remaining followers in the formation is greater than or equal to 1, the newly added nth follower tracks and locates the (n-1)th follower and (n-2)th follower as neighboring agents. The advantage of this positioning method is that the formation as a whole is expanded in a cascade manner without relying on a central node. At the same time, when an agent in the formation fails, the subsequent agents can fill in the gap nearby, improving the robustness of the formation.

Claims

1. A cluster control method based on the Henneberg constraint method under azimuth only, the specific steps are as follows: S1. Assign identities to all agents in the cluster of drones and unmanned vehicles, determine the leader, the first follower, and the remaining followers. The sensors and communication equipment carried by each member can determine the relative azimuth measurement information and identity information of other members within its detection range; S2, the leader and the first follower move autonomously according to task requirements; S3, each agent obtains the relative azimuth information of its neighboring agents; S4, the agent obtains positioning information based on the relative azimuth of the neighboring agent and obtains the error to the target position in its own coordinate system; S5, the agent inputs the motion control law according to the target error and completes the formation following behavior; In step S3: The process of determining the three different identities of the agent is as follows: The remaining followers 1 take the leader and the first follower as neighboring agents; the remaining followers 2 take the first follower and follower 1 as neighboring agents; the remaining followers i take the remaining followers (i-1) and the remaining followers (i-2) as neighboring agents; Each agent can obtain the identity information and relative azimuth measurement information of other agents within the detection range of its own sensors. The starting positions of cluster members at the beginning of the task are close to each other at a reasonable distance and are located in the task area to be covered; In step S4: Agent n obtains its own forward azimuth angle as θ asn , the azimuth angle of the nth follower’s neighboring agent relative to itself is θ snsn-1 and θ snsn-2 , set the agent distance value ρ specified by the system s , according to the original system settings of the nth follower, obtain its target position s in the same coordinate system t , according to the calculated s n , calculate the error value relative to the target position at this time The position s of agent n relative to the (n-2)th follower is calculated by the following formula: n =(x n ,y n ) T ,in, T represents the matrix transpose, (x n ,y n ) represents the two-dimensional coordinate value at time n:

2. The cluster control method based on the Henneberg constraint method under azimuth only according to claim 1, characterized in that: In step S5: use The designed control law is: Among them, ω fi is the direction the agent is pointing to, u fi is the speed target of the agent at this time. Under the constraints of the two, the follower agent converges to the expected position in the formation; K ai is the amplitude parameter, K fi is the frequency parameter; The area to be covered is an irregular connected space surrounded by the outer boundary contour of the cluster task area. The process of the formation cluster task meets the following conditions: A: The shape of the area to be covered is unknown to all members of the cluster before the task; B: The narrowest part of the area to be covered allows the formation to pass through in the smallest state without causing collision; C: The first follower determines the relative angle direction of its own target position relative to the leader in the positive direction of its movement according to the needs during the task. This position determines the overall movement direction of the formation shape; D: The first follower determines the relative distance of its own target position relative to the leader in the positive direction of its movement according to the needs during the task. This position determines the overall size of the formation shape; E: During the normal progress of the task, the distance between each agent and its neighboring agents is subject to an upper limit constraint, which ensures that the agent and its neighboring agents can maintain detection and communication, thereby ensuring the overall integrity of the formation; F: The distance between any two agents is subject to a lower limit constraint, so that at every moment, the two agents maintain a safe distance, that is, at any moment, any two agents in the formation will not collide with each other; G: The distance between any agent and the boundary is subject to a lower limit constraint, so that at every moment, the two maintain a safe distance, that is, at any moment, no agent in the formation will collide with the boundary.

Images