Rapid cooperative task allocation method adaptive to group dynamic change

Abstract

The invention discloses a rapid cooperative task allocation method adaptive to group dynamic change, which adopts a decision-making mode of distributed simultaneous calculation of a dynamic decision-making center and each flight platform node, recommends optimal allocation to the decision-making center through distributed nodes, and issues the optimal allocation mode to all nodes by the decision-making center. Optimal allocation result coordination is realized, frequent allocation information interaction is avoided, and the global allocation consistency convergence speed is increased; according to the method, by adopting random selection of flight platforms and task target matching calculation, the global optimization ability is improved, distributed simultaneous calculation of all flight platform nodes is achieved, the matching optimization efficiency is improved, the optimization iteration frequency of all the nodes is limited, it is guaranteed that an approximate optimal decision result is rapidly obtained, and rapid cooperative task allocation under cluster dynamic change is achieved.

Description

Technical Field

[0001] The invention relates to a cluster collaborative task allocation method, in particular to a fast collaborative task allocation method adapting to dynamic changes of a group, and belongs to the technical field of cluster collaborative task decision making. Background Art

[0002] With the maturity of unmanned equipment technology and the rapid development of artificial intelligence technology, the coordinated execution of tasks by intelligent unmanned swarms has gradually become the norm. The coordinated task allocation of multi-agents against multiple targets is the key to exerting the swarm's collaborative capabilities. Especially in air confrontation tasks, the scale and networking modes of the aircraft groups of the two sides are diverse, and the number of clusters, relative spatial positions, and inter-group communication status change dynamically during the confrontation. It is necessary to quickly solve the optimal task allocation strategy online according to real-time situation changes to improve the efficiency of group coordination.

[0003] At present, the collaborative task allocation method is divided into centralized allocation and distributed allocation according to the collaborative information processing mode. In the highly dynamic environment of air confrontation, the distributed allocation method has attracted much attention due to its strong system robustness and the advantages of autonomous decision-making of each node. The patent of Beijing Institute of Technology (CN111489049A) proposes a multi-agent distributed task allocation method, which adopts the idea of ​​auction algorithm to allocate tasks under the condition of partial communication topology, and solves the problem of multi-agent distributed task allocation under the condition of partial communication topology. The patent of the 28th Institute of China Electronics Technology Group Corporation (CN117892933A) proposes a multi-UAV real-time distributed task allocation and redistribution method to solve the problems of unreasonable task allocation and redistribution and low operating efficiency, and realize the real-time generation of multi-UAV multi-task pre-allocation and redistribution schemes. The patent of National University of Defense Technology (CN118377608A) proposes a distributed task allocation method and system based on auction algorithm, which makes full use of the computing resources of each agent in the multi-agent system to improve the efficiency and speed of task allocation, and is suitable for large-scale, heterogeneous and dynamic agent systems. However, the existing distributed task allocation methods are mainly improved based on auction algorithms, which have the following problems:

[0004] (1) Fully distributed information interaction requires a high quality of inter-group communication. Task publishers and bidders exchange information in turns to determine the final allocation plan, which requires a relatively stable communication connection and a high communication bandwidth between groups.

[0005] (2) Fully distributed information interaction is difficult to adapt to the highly dynamic environment of the group. In a highly dynamic environment with changes in the number of individuals in the cluster, the number of targets, and the relative positions of the two opposing parties, multiple rounds of interaction with fully distributed allocations take a long time, making it difficult to quickly achieve global consistent allocations, and the group coordination speed is insufficient. Summary of the invention

[0006] The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art, to provide a fast collaborative task allocation method that adapts to the dynamic changes of the group, and to solve the problem that the distributed allocation method requires frequent information interaction; the technical problem further solved by the present invention is the problem that the distributed allocation caused by the dynamic changes in the number of clusters and the communication network environment is difficult to quickly reach global consistency.

[0007] To achieve the above objectives, the present invention provides a fast collaborative task allocation method that adapts to group dynamic changes, and the implementation steps are as follows:

[0008] Step 1: Set the variables to store the flight platform node state set {Sm} and the task set {Tn}, and initialize the set to an empty set;

[0009] Step 2: Obtain the node status locally and receive other node statuses from the cluster communication network, and add the node statuses participating in the task allocation to the node status set {Sm}. The single node status at least includes the flight platform number, position, speed, task execution status, and timestamp information;

[0010] Step 3: Maintain the task decision center node according to the data in the node status set {Sm} for optimal task allocation and release;

[0011] Step 4: Receive all task statuses from the local node and other nodes, and put the tasks that need to be assigned into the cluster task set {Tn}. The information of a single task includes at least the number, location, type, and timestamp of the task target.

[0012] Step 5: Calculate the optimal task allocation based on the node state set {Sm} and the cluster task set {Tn};

[0013] Step 6: If it is the first calculation, use the calculated optimal task allocation to update the historical optimal allocation; otherwise, determine whether the calculated optimal task allocation is better than the historical optimal allocation. If yes, go to step 7; if no, go to step 9;

[0014] Step 7: Update the historical optimal allocation with the calculated optimal task allocation;

[0015] Step 8: Mark the allocation status as having local updates;

[0016] Step 9: Mark the allocation status as no local update;

[0017] Step 10: Read data from the network port to determine whether a new optimal allocation is received; if yes, go to step 11; if no, go to step 14;

[0018] Step 11: Determine whether the received new optimal allocation is better than the local historical optimal allocation; if yes, go to step 12; if no, go to step 14;

[0019] Step 12: Update the local historical optimal allocation with the received new optimal allocation;

[0020] Step 13: Mark the allocation status as having external update;

[0021] Step 14: Determine whether this node is a task decision center; if yes, go to step 15; if no, go to step 18;

[0022] Step 15: Determine whether the allocation status is updated, that is, determine whether there is a local update or an external update; if yes, go to step 16; if no, go to step 17;

[0023] Step 16: Publish the historical optimal task allocation to all nodes;

[0024] Step 17: Return to step 2;

[0025] Step 18: Determine whether the allocation status has been locally updated; if yes, go to step 19; if no, go to step 17;

[0026] Step 19: Publish the historical optimal task allocation to the decision center node and return to step 2.

[0027] Furthermore, the task decision center node is maintained in step 3, and the implementation steps are as follows:

[0028] Step 3.1: Determine whether the node number is the smallest among all nodes according to the data in the node status set {Sm}; if yes, go to step 3.2; if no, go to step 3.8;

[0029] Step 3.2: This node calculates the maintenance task decision center number;

[0030] Step 3.3: Calculate the cluster center coordinates (X, Y) to get the average center position of all nodes. The calculation method is as follows:

[0031]

[0032] Step 3.4: Calculate the relative distance between the average center of all nodes and all nodes, and select the node with the closest distance as the decision center;

[0033] Step 3.5: Determine whether to update the decision center number, compare the newly calculated decision center with the historically stored decision center, if they are consistent, then it is yes, go to step 3.6; if they are inconsistent, then it is no, and the calculation ends;

[0034] Step 3.6: Publish the task decision center number to all nodes;

[0035] Step 3.7: Update the local task decision center number and end the calculation;

[0036] Step 3.8: Determine whether a new decision center number has been received. If yes, go to step 3.9; if no, end the calculation;

[0037] Step 3.9: Update the local task decision center number and end the calculation.

[0038] Furthermore, the optimal task allocation calculation in step 5 is implemented as follows:

[0039] Step 5.1: Determine whether there are tasks to be assigned. If the task set {Tn} is not empty, there are tasks to be assigned, and go to step 5.2; otherwise, end the calculation;

[0040] Step 5.2: Set the optimization times k to 0 and the optimization limit times N, where N is determined by the product of the total number of flight platform nodes Nm and the total number of tasks Nt, and the maximum value of N should be limited according to the requirements of rapid decision-making, so as to reduce the computing time in the case of a large number of clusters and tasks. When the task allocation computing time is required to be within tens of milliseconds, the maximum value of N is set to within 100. When the task allocation computing time is required to be larger, the maximum value of N can be appropriately increased;

[0041] Step 5.3: Determine whether the optimization times k is less than the optimization limit times N; if yes, go to step 5.4; otherwise, end the calculation;

[0042] Step 5.4: Assign the node state set {Sm} to the new node state set {Smn};

[0043] Step 5.5: Randomly select a node from the node state set {Smn} for task pairing, remove the selected node from {Smn} and update {Smn};

[0044] Step 5.6: Set the index number i of the task in the task set to 0;

[0045] Step 5.7: Determine whether the index number i of the task reaches the total number of tasks; if yes, go to step 5.11; otherwise, go to step 5.8;

[0046] Step 5.8: Select the task with task index number i from the task set {Tn} and match it with the node selected in step 5.5;

[0047] Step 5.9: Determine whether the selected node matches the task; if yes, go to step 5.10; otherwise, go to step 5.16; the criterion for determining whether the match is successful includes at least that the flight platform represented by the node can reach the task location, and the flight platform's maneuverability is greater than the target maneuverability when arriving;

[0048] Step 5.10: Select the allocation relationship storage value allocation result set {A};

[0049] Step 5.11: Determine whether the node state set {Smn} is an empty set; if yes, go to step 5.12; otherwise, go to step 5.4;

[0050] Step 5.12: Calculate the cost value under the current distribution relationship;

[0051] Step 5.13: Determine whether the calculated cost value is better than the historical optimal allocation cost value; if yes, go to step 5.14; otherwise, go to step 5.15;

[0052] Step 5.14: Update the historical optimal allocation with the current allocation relationship and update the cost value;

[0053] Step 5.15: Add 1 to the value of optimization times k and go to step 5.3;

[0054] Step 5.16: Add 1 to the value of the task index number i and go to step 5.7. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 The figure is a flowchart of task allocation in one embodiment of the present invention.

[0056] Figure 2 This is a flowchart of the task decision center point maintenance in task allocation step 3 in one embodiment of the present invention.

[0057] Figure 3 This is a task allocation calculation flow chart of step 5 of task allocation in one embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following is combined with Figures 1 to 3The present invention is further described in detail with specific implementation methods. According to the following description, the advantages and features of the present invention will be clearer. It should be noted that the drawings are in a very simplified form and use non-precise proportions, which are only used to conveniently and clearly assist in explaining the purpose of the implementation methods of the present invention. In order to make the purposes, features and advantages of the present invention more obvious and easy to understand, please refer to the drawings. It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention, so they have no technical substantive significance. Any structural modification, change in proportional relationship or adjustment of size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope of the technical content disclosed by the present invention.

[0059] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, unmanned cluster collaborative task allocation, article or field equipment including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, unmanned cluster collaborative task allocation, article or field equipment. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, unmanned cluster collaborative control system, article or field equipment including the elements.

[0060] See also Figure 1 As shown, this embodiment provides a fast collaborative task allocation method that adapts to dynamic changes in a group, and the implementation steps are as follows:

[0061] Step 1: Set the variables to store the flight platform node state set {Sm} and the task set {Tn}, and initialize the set to an empty set;

[0062] Step 2: Obtain the node status locally and receive other node statuses from the cluster communication network, and add the node statuses participating in the task allocation to the node status set {Sm}. The single node status at least includes the flight platform number, position, speed, task execution status, and timestamp information;

[0063] Step 3: Maintain the task decision center node according to the data in the node status set {Sm} for optimal task allocation and release;

[0064] Step 4: Receive all task statuses from the local node and other nodes, and put the tasks that need to be assigned into the cluster task set {Tn}. The information of a single task includes at least the number, location, type, and timestamp of the task target.

[0065] Step 5: Calculate the optimal task allocation based on the node state set {Sm} and the cluster task set {Tn};

[0066] Step 6: If it is the first calculation, use the calculated optimal task allocation to update the historical optimal allocation; otherwise, determine whether the calculated optimal task allocation is better than the historical optimal allocation. If yes, go to step 7; if no, go to step 9;

[0067] Step 7: Update the historical optimal allocation with the calculated optimal task allocation;

[0068] Step 8: Mark the allocation status as having local updates;

[0069] Step 9: Mark the allocation status as no local update;

[0070] Step 10: Read data from the network port to determine whether a new optimal allocation is received; if yes, go to step 11; if no, go to step 14;

[0071] Step 11: Determine whether the received new optimal allocation is better than the local historical optimal allocation; if yes, go to step 12; if no, go to step 14;

[0072] Step 12: Update the local historical optimal allocation with the received new optimal allocation;

[0073] Step 13: Mark the allocation status as having external update;

[0074] Step 14: Determine whether this node is a task decision center; if yes, go to step 15; if no, go to step 18;

[0075] Step 15: Determine whether the allocation status is updated, that is, determine whether there is a local update or an external update; if yes, go to step 16; if no, go to step 17;

[0076] Step 16: Publish the historical optimal task allocation to all nodes;

[0077] Step 17: Return to step 2;

[0078] Step 18: Determine whether the allocation status has been locally updated; if yes, go to step 19; if no, go to step 17;

[0079] Step 19: Publish the historical optimal task allocation to the decision center node and return to step 2.

[0080] For further information, see Figure 2 As shown, this embodiment provides a maintenance task decision center node in the task allocation step 3, and the implementation steps are as follows:

[0081] Step 3.1: Determine whether the node number is the smallest among all nodes according to the data in the node status set {Sm}; if yes, go to step 3.2; if no, go to step 3.8;

[0082] Step 3.2: This node calculates the maintenance task decision center number;

[0083] Step 3.3: Calculate the cluster center coordinates (X, Y) to get the average center position of all nodes. The calculation method is as follows:

[0084]

[0085] Step 3.4: Calculate the relative distance between the average center of all nodes and all nodes, and select the node with the closest distance as the decision center;

[0086] Step 3.5: Determine whether to update the decision center number, compare the newly calculated decision center with the historically stored decision center, if they are consistent, then it is yes, go to step 3.6; if they are inconsistent, then it is no, and the calculation ends;

[0087] Step 3.6: Publish the task decision center number to all nodes;

[0088] Step 3.7: Update the local task decision center number and end the calculation;

[0089] Step 3.8: Determine whether a new decision center number has been received. If yes, go to step 3.9; if no, end the calculation;

[0090] Step 3.9: Update the local task decision center number and end the calculation.

[0091] For further information, see Figure 3 As shown, this embodiment provides the optimal task allocation calculation in the task allocation step 5, and the implementation steps are as follows:

[0092] Step 5.1: Determine whether there are tasks to be assigned. If the task set {Tn} is not empty, there are tasks to be assigned, and go to step 5.2; otherwise, end the calculation;

[0093] Step 5.2: Set the optimization times k to 0 and the optimization limit times N, where N is determined by the product of the total number of flight platform nodes Nm and the total number of tasks Nt, and the maximum value of N should be limited according to the requirements of rapid decision-making, so as to reduce the computing time in the case of a large number of clusters and tasks. When the task allocation computing time is required to be within tens of milliseconds, the maximum value of N is set to within 100. When the task allocation computing time is required to be larger, the maximum value of N can be appropriately increased;

[0094] Step 5.3: Determine whether the optimization times k is less than the optimization limit times N; if yes, go to step 5.4; otherwise, end the calculation;

[0095] Step 5.4: Assign the node state set {Sm} to the new node state set {Smn};

[0096] Step 5.5: Randomly select a node from the node state set {Smn} for task pairing, remove the selected node from {Smn} and update {Smn};

[0097] Step 5.6: Set the index number i of the task in the task set to 0;

[0098] Step 5.7: Determine whether the index number i of the task reaches the total number of tasks; if yes, go to step 5.11; otherwise, go to step 5.8;

[0099] Step 5.8: Select the task with task index number i from the task set {Tn} and match it with the node selected in step 5.5;

[0100] Step 5.9: Determine whether the selected node matches the task; if yes, go to step 5.10; otherwise, go to step 5.16; the criterion for determining whether the match is successful includes at least that the flight platform represented by the node can reach the task location, and the flight platform's maneuverability is greater than the target maneuverability when arriving;

[0101] Step 5.10: Select the allocation relationship storage value allocation result set {A};

[0102] Step 5.11: Determine whether the node state set {Smn} is an empty set; if yes, go to step 5.12; otherwise, go to step 5.4;

[0103] Step 5.12: Calculate the cost value under the current distribution relationship;

[0104] Step 5.13: Determine whether the calculated cost value is better than the historical optimal allocation cost value; if yes, go to step 5.14; otherwise, go to step 5.15;

[0105] Step 5.14: Update the historical optimal allocation with the current allocation relationship and update the cost value;

[0106] Step 5.15: Add 1 to the value of optimization times k and go to step 5.3;

[0107] Step 5.16: Add 1 to the value of the task index number i and go to step 5.7.

[0108] Compared with the prior art, the present invention has at least the following beneficial effects:

[0109] (1) The present invention adopts distributed simultaneous calculation of optimal allocation by each flight platform node, the node recommends the optimal allocation to the decision center, and the decision center publishes the optimal allocation method to all nodes, so that each node sends the calculation result only when the result calculated is better than the global optimal allocation, thus avoiding frequent allocation information interaction, reducing communication requirements, and increasing the global allocation consistency convergence speed;

[0110] (2) The present invention adopts a dynamic maintenance decision center method to realize that the cluster has a unified and dynamically adjustable task decision center. The decision center can be re-determined after the flight platform is lost, avoiding the failure of the cluster system after the fixed center is lost, and improving the robustness of the system;

[0111] (3) The present invention adopts random selection of flight platforms and matching calculations with mission targets at the flight platform nodes to increase the global optimization capability. At the same time, since the flight platform nodes perform distributed and simultaneous calculations, the matching optimization efficiency is greatly increased compared to single platform calculations.

[0112] (4) The present invention utilizes the advantages of distributed simultaneous computing of flight platform nodes and limits the number of optimization iterations of each node, thereby ensuring that the approximate optimal decision result can be quickly obtained first, and then the global optimal result is gradually updated through continuous optimization calculation and recommendation of each node.

[0113] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A fast collaborative task allocation method that adapts to group dynamic changes, characterized in that: The steps include: Step 1: Set the variables to store the flight platform node state set {Sm} and the task set {Tn}, and initialize the set to an empty set; Step 2: Obtain the node status locally and receive other node statuses from the cluster communication network, and add the node statuses participating in task allocation to the node status set {Sm}; Step 3: Maintain the task decision center node according to the data in the node status set {Sm} for optimal task allocation and release; Step 4: Receive all task statuses from the local node and other nodes, and put the tasks that need to be assigned into the cluster task set {Tn}. The information of a single task includes at least the number, location, type, and timestamp of the task target. Step 5: Calculate the optimal task allocation based on the node state set {Sm} and the cluster task set {Tn}; Step 6: If it is the first calculation, the calculated optimal task allocation is used to update the historical optimal allocation; otherwise, determine whether the calculated optimal task allocation is better than the historical optimal allocation; If yes, go to step 7; if no, go to step 9; Step 7: Update the historical optimal allocation with the calculated optimal task allocation; Step 8: Mark the allocation status as having local updates; Step 9: Mark the allocation status as no local update; Step 10: Read data from the network port to determine whether a new optimal allocation has been received; If yes, go to step 11; if no, go to step 14; Step 11: Determine whether the received new optimal allocation is better than the local historical optimal allocation; If yes, go to step 12; if no, go to step 14; Step 12: Update the local historical optimal allocation with the received new optimal allocation; Step 13: Mark the allocation status as having external update; Step 14: Determine whether this node is a task decision center; If yes, go to step 15; if no, go to step 18; Step 15: Determine whether the allocation status is updated, that is, determine whether there is a local update or an external update; If yes, go to step 16; if no, go to step 17; Step 16: Publish the historical optimal task allocation to all nodes; Step 17: Return to step 2; Step 18: Determine whether the allocation status has been locally updated; If yes, go to step 19; if no, go to step 17; Step 19: Publish the historical optimal task allocation to the decision center node and return to step 2.

2. A fast collaborative task allocation method that adapts to group dynamic changes according to claim 1, characterized in that: In step 2, the state of a single node includes at least the number, position, speed, mission execution status, and timestamp information of the flight platform.

3. A fast collaborative task allocation method that adapts to group dynamic changes according to claim 2, characterized in that: In step 3, the task decision center node is maintained, and the implementation steps are as follows: Step 3.1: Determine whether the node number is the smallest among all nodes according to the data in the node status set {Sm}; If yes, go to step 3.2; if no, go to step 3.8; Step 3.2: This node calculates the maintenance task decision center number; Step 3.3: Calculate the cluster center coordinates (X, Y) to get the average center position of all nodes. The calculation method is as follows: Cluster center coordinates Node i coordinate (X, Y) / total number of nodes Nm; Step 3.4: Calculate the relative distance between the average center of all nodes and all nodes, and select the node with the closest distance as the decision center; Step 3.5: Determine whether to update the decision center number, and compare the newly calculated decision center with the historically stored decision center; If they are consistent, then it is yes and go to step 3.6; if they are inconsistent, then it is no and the calculation ends; Step 3.6: Publish the task decision center number to all nodes; Step 3.7: Update the local task decision center number and end the calculation; Step 3.8: Determine whether a new decision center number has been received; If yes, go to step 3.9; if no, end the calculation; Step 3.9: Update the local task decision center number and end the calculation.

4. A fast collaborative task allocation method that adapts to group dynamic changes according to claim 3, characterized in that: The optimal task allocation calculation in step 5 is implemented as follows: Step 5.1: Determine whether there are tasks to be assigned. If the task set {Tn} is not empty, there are tasks to be assigned, and go to step 5.2; otherwise, end the calculation; Step 5.2: Set the optimization times k to 0 and the optimization limit times N, where N is determined by the product of the total number of flight platform nodes Nm and the total number of tasks Nt, and the maximum value of N should be limited according to the requirements of rapid decision-making, so as to reduce the computing time in the case of a large number of clusters and tasks. When the task allocation computing time is required to be within tens of milliseconds, the maximum value of N is set to within 100. When the task allocation computing time is required to be larger, the maximum value of N can be appropriately increased; Step 5.3: Determine whether the optimization times k is less than the optimization limit times N; if yes, go to step 5.4; otherwise, end the calculation; Step 5.4: Assign the node state set {Sm} to the new node state set {Smn}; Step 5.5: Randomly select a node from the node state set {Smn} for task pairing, remove the selected node from {Smn} and update {Smn}; Step 5.6: Set the index number i of the task in the task set to 0; Step 5.7: Determine whether the index number i of the task reaches the total number of tasks; If yes, go to step 5.11; otherwise, go to step 5.8; Step 5.8: Select the task with task index number i from the task set {Tn} and match it with the node selected in step 5.5; Step 5.9: Determine whether the selected node matches the task; if yes, go to step 5.10; otherwise, go to step 5.16; the criterion for determining whether the match is successful includes at least that the flight platform represented by the node can reach the task location, and the flight platform's maneuverability is greater than the target maneuverability when arriving; Step 5.10: Select the allocation relationship storage value allocation result set {A}; Step 5.11: Determine whether the node state set {Smn} is an empty set; if yes, go to step 5.12; otherwise, go to step 5.4; Step 5.12: Calculate the cost value under the current distribution relationship; Step 5.13: Determine whether the calculated cost value is better than the historical optimal allocation cost value; if yes, go to step 5.14; otherwise, go to step 5.15; Step 5.14: Update the historical optimal allocation with the current allocation relationship and update the cost value; Step 5.15: Add 1 to the value of optimization times k and go to step 5.3; Step 5.16: Add 1 to the value of the task index number i and go to step 5.7.

Images