A collaborative decision-making method for vehicle delivery point and target grouping for task planning in vehicle delivery mode

Abstract

A method for collaborative decision-making on vehicle delivery points and target groups for task planning in a vehicle delivery mode belongs to the field of unmanned equipment support technology. Step 1: Sort all targets; Step 2: Select a feasible delivery point set to determine whether all target points can be covered: If so, execute Step 5, otherwise execute Step 3; Step 3: Record target points that are not covered and select delivery points for all recorded target points; Step 4: Fix the selected delivery point and remove the recorded target point from the total target point set, update the target point set, sort, and return to Step 2; Step 5: Input the selected delivery point to obtain a target point group set centered on the delivery point; Step 6: Merge the isolated target point into other groups to obtain target point classification and corresponding delivery point. The present invention can improve the efficiency and safety of task execution, reduce the risk of being discovered and intercepted by the enemy, and reduce the vehicle's navigation distance and execution time.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned equipment support, and relates to a vehicle delivery point and target grouping collaborative decision-making method for task planning in a vehicle delivery mode. Background Art

[0002] Generally speaking, targets on a battlefield are widely distributed, and different targets reside in diverse environments. Given the limited range of drones, how to safely and quickly deploy them to the battlefield has become a widely sought-after issue in modern warfare. One feasible approach is to utilize high-range, high-capacity vehicles to deploy drones. The core of this issue lies in how to select feasible delivery points across a vast battlefield for drone delivery missions. To rapidly plan vehicle delivery routes and drone mission execution plans, a viable approach is to use clustering algorithms to partition targets and select delivery points. Currently, mainstream clustering algorithms can be divided into three categories: partitioning methods, represented by K-means and KNN; hierarchical methods, represented by AGNES and DIANA; and density-based methods, represented by DBSCAN and DPC. Existing improvements span all major aspects of algorithm design, including algorithm input, process, output, and conceptual modifications. However, most improvements focus on improving the data itself and the algorithm's solution speed, with little attention paid to the constraints within the clustering process. For example, the Chinese invention patent "Large-Scale UAV Mission Planning Method and System" (CN119045505A) uses a spectral clustering algorithm to divide multiple initial mission sequences into multiple groups when faced with large-scale UAV missions. However, this algorithm fails to consider the actual constraints that may arise during mission execution and is therefore unsuitable for actual combat operations. For example, the Chinese invention patent "UAV and Manned Aircraft Joint Search and Rescue Path Planning Method under Airspace Layering" (CN119759042A) proposes a rule-based airspace layering model to provide an initial scope for joint path planning between manned and UAVs. This invention considers environmental partitioning in rescue scenarios and ensures rescue efficiency while satisfying heterogeneous aircraft constraints. However, due to the heterogeneous constraints in different scenarios, it cannot be transferred to battlefield applications. For example, the Chinese invention patent "Millimeter Wave Anti-blocking Multi-UAV Deployment Method Based on Building Geometry Analysis" (CN114039652A) proposes a BT-K-means multi-UAV clustering algorithm. Based on the minimum path loss principle, multiple iterations are performed to determine user clustering and initial UAV positions. Similarly, while this invention fully considers the practical constraints of urban areas, it is still not suitable for battlefield environments. To meet actual combat needs, the selection of delivery points must take into account practical constraints such as drone range, no-fly zones, and radar, to ensure that the vehicle can safely and efficiently perform delivery combat missions. Summary of the Invention

[0003] To address the aforementioned technical issues, the present invention proposes a collaborative decision-making method for vehicle delivery points and target grouping for mission planning in a vehicle delivery mode. This method is primarily applicable to mission planning and delivery point selection for drone swarms in battlefield environments. This method comprehensively considers multiple factors, including enemy radar threats, no-fly zone restrictions, and drone range. By inputting target location information and calculating local density and relative distance for preliminary screening, it then proposes a multi-constraint fusion strategy for selecting delivery points, ensuring that they are located outside radar coverage and avoid no-fly zones. Furthermore, the problem of uneven target density is addressed through the principle of regional equivalence. For target points within radar coverage and in no-fly zones, the map grid is traversed, and the grid with the highest value is selected as the new delivery point. Isolated target points are then merged into other groups to reduce the number of delivery points, minimize vehicle travel distance, and reduce potential hazards. This method employs a logic-based clustering method, effectively addressing the practical limitations of traditional clustering algorithms and improving the efficiency and safety of drone swarm mission execution in complex battlefield environments.

[0004] In order to achieve the above purpose, the specific technical solution adopted by the present invention is:

[0005] A method for collaborative decision-making between vehicle delivery points and target groups for task planning in a vehicle delivery mode, the method comprising the following steps:

[0006] Step 1: Input the target point location information, use statistical formulas to solve and obtain the local density and relative distance of each point, and sort all targets in descending order according to the product of local density and relative distance;

[0007] Step 2: Based on the sorted target points, a constraint fusion strategy that considers multiple constraints is proposed to select a set of feasible delivery points. Based on the selected delivery points, it is determined whether the delivery points can cover all target points. If so, step 5 is executed; otherwise, step 3 is executed.

[0008] Step 3: Record the target points not covered in step 2, use the grid map to traverse the map grid, and select the delivery points for all recorded target points according to the grid value;

[0009] Step 4: Fix the delivery point selected in step 3 and remove the target point recorded in step 3 from the total target point set to update the target point set. Sort the updated target point set in descending order based on the product of local density and relative distance, and return to step 2.

[0010] Step 5: Input the selected delivery point and group the targets using the minimum distance principle to obtain a target point group set with the delivery point as the center.

[0011] Step 6: Input the target point group set, fuse the groups that only include one target point, fuse the isolated target points into other groups, and finally obtain the target point classification scheme and the corresponding delivery points.

[0012] Furthermore, the step 1 is specifically as follows:

[0013] Input the target point location information, first calculate the local density of each point, where the local density calculation uses the Gaussian kernel formula:

[0014] (1)

[0015] In formula (1), and Indicates two different target points. Indicates the target point The local density of Indicates the target point and The Euclidean distance of The preset neighborhood cutoff distance is The value of ,in Is a parameter that adjusts the range, and its value is less than , The range of the drone.

[0016] After completing the local density calculation of all target points, the relative distance calculation is carried out. Compared with The distance between the nearest target point with higher local density. The local density is the highest among all target points, so Target point The local density of , then the relative distance expression formula is:

[0017] (2)

[0018] Otherwise the target point The relative distance is expressed as:

[0019] (3)

[0020] After obtaining the local density and relative distance of each target point, the product of the local density and relative distance of each target point is calculated, and this is used as an indicator to sort the target points from large to small.

[0021] Furthermore, the step 2 is specifically as follows:

[0022] Step 2.1: Use the constraint fusion strategy to select delivery points;

[0023] After the target points are sorted, the delivery point selection process is carried out. Based on the sorted target points, this paper proposes a constraint fusion strategy that considers multiple constraints to select a feasible delivery point set. According to the sorted order, the target points are input one by one for retrieval. If the current target point meets the constraint fusion strategy, it will be selected as the delivery point and stored in the delivery point set. . After completing the retrieval of one target point, the next target point will be searched. When the selected delivery point can cover all target points or all target points have been searched once, the retrieval process stops. In the constraint fusion strategy, the present invention takes into account radar constraints, no-fly zone constraints, screening constraints and drone range constraints, and gives corresponding solutions and delivery point screening measures around the above three constraints. First, the screening conditions for the three constraints are introduced. When the target point meets the three screening conditions, it will be selected as the delivery point. The details are as follows:

[0024] 1) Build filter conditions around radar constraints;

[0025] In this strategy, the target point will be selected as the delivery point only when it is outside the radar coverage. For the sake of convenience, in the constraint fusion strategy, the target point is represented as , and its position is represented by ,in Indicates a point The horizontal axis, Indicates a point The radar is represented by , then its position is expressed as , Respectively represent the horizontal and vertical coordinates of the radar. Then if the target point To be selected as a delivery point, the following conditions must be met:

[0026] (4)

[0027] in, For radar The detection coverage of the target point is expressed in formula (4). The distance between any radar and the delivery point must be greater than the radar's detection coverage, otherwise it cannot be selected as a delivery point.

[0028] 2) Building screening conditions around no-fly zone constraints;

[0029] In addition, the selection of delivery points needs to take into account the no-fly zones in the battlefield. The delivery points cannot be located within the no-fly zones. The constraints are as follows:

[0030] (5)

[0031] in, Indicates the target point location; Indicates the the coverage of the no-fly zones; Indicates the number of no-fly zones; Used to describe relationships that do not belong mathematically.

[0032] 3) Constructing screening conditions around screening constraints and drone range constraints;

[0033] This setting is used to address the limitations of the algorithm in the case of uneven target point density and the UAV range constraints. The specific form is as follows:

[0034] For the convenience of description, assume that several delivery points have been selected, and set the set of current delivery points as .in Represents the first delivery point selected in the set, and sets Delivery point is the center of the circle, The set of all target points within the circle coverage area of ​​radius . In the The delivery points are , and set Centered is the center of the circle, Is the set of target points covered by a circle with a radius of The latest cluster center added in is , the current search point is , then:

[0035] (6)

[0036] Formula (6) represents the target point of the current retrieval Cannot be located in an existing delivery points, otherwise it cannot be selected as a new delivery point. Representing a collection When the retrieved target point satisfies the three constraints of formula (4) to formula (6), it will be added to the delivery point set. .

[0037] Step 2.2: Determine whether there are any remaining target points that cannot be covered by the current delivery point;

[0038] When step 2.1 is completed, the delivery point set is obtained , then calculate the number of target points covered by the delivery point and make a judgment, let the target point set be , the elements in the set are all target points, then:

[0039] (7)

[0040] in, Indicates the set of target points that are not covered by the current delivery point. For the convenience of description, it is named as the no-home point set. The target points that are not covered are no-home points. If the number of elements in is 0, execute step 5; if the number of elements in is not 0, execute step 3.

[0041] Furthermore, the step 3 is specifically as follows:

[0042] In this step, the corresponding delivery point selection will be carried out for the no-belonging points obtained in step 2. From step 2, the no-belonging point set is obtained Since this algorithm will be repeated for multiple rounds, each time step 3 is executed, In order to record the no-belonging points that appear during the entire algorithm iteration process, the global no-belonging point set is set as , Initially an empty set. The coverage radius control parameter for selecting delivery points for non-attribution points is used to control the coverage range of the selected delivery points in this step. The size limit is , perform the following steps:

[0043] Step 3.1, generate a 2D battlefield environment map and grid the battlefield. Extract all the no-belonging points and record the radius of each no-belonging point. The grid of the circular domain intersecting with , and record the non-attributable points in this round: ;

[0044] Step 3.2, calculate the value of the recorded grid, the value function is: ,in, Indicates that the grid is the center. The number of unattached points covered in the circular domain with a radius of Indicates radar threat weight; Indicates the radar threat to the grid, in the form of:

[0045] (8)

[0046] in, A parameter that represents the acceptable distance between the delivery point and the radar. When the grid is within the unacceptable range, its radar threat is infinite. Indicates radar With the current grid the distance between them; Represents an infinite number; Represents a radar ensemble.

[0047] Step 3.3, after completing the value calculation of all grids, select the grid with the highest value For the new delivery point, store the grid into the collection , and update the set of no-belonging points :

[0048] (9)

[0049] in, Indicates that it is in the maximum value grid There is no belonging point in the circular domain; represents the maximum value grid; Indicates the maximum range of the drone; Indicates the coverage radius control parameter for selecting delivery points for non-home points;

[0050] Finally, repeat steps 3-2 to 3-3 until There is no longer any no-home point in the system, and finally the no-home point delivery point set is obtained. .

[0051] The step 4 is specifically as follows:

[0052] Save the delivery point selected in step 3 Since the no-home points cannot find the delivery points through the constraint fusion strategy, and the corresponding delivery points are found through step 3, and because the target points will affect each other's local density and relative distance, it is necessary to discard the no-home points, update the local density and relative distance of the remaining target points, and re-sort them. First, the no-home points recorded in step 3 are From the total target point collection Remove and update the target point set. Then recalculate the set according to the formula in step 1 The local density and relative distance of each target point in the target area are calculated, and the target points are sorted in descending order according to the product of the local density and the relative distance. Finally, , return to step 2.

[0053] The step 5 is specifically as follows:

[0054] This step is established after all target points have corresponding delivery points, and there is a final delivery point set. , each element in the set represents a delivery point, Indicates the number of delivery points. Then the target grouping process is carried out, that is, the task of the target point is performed by the drone delivered by which delivery point. This step adopts the minimum distance principle to group the target point, that is, the target point is divided into the group of the closest delivery point, and finally the group set is obtained. , where each element in the set represents a group. For example, For Grouping of destination points as delivery points.

[0055] The step 6 is specifically as follows:

[0056] After completing the target point grouping in step 5, some isolated points may exist (a group containing only one target point). This step aims to integrate these isolated points into other groups by changing the delivery points of other groups. This reduces the number of delivery points, shortens the vehicle's navigation distance, and reduces potential dangers to the vehicle. The specific process is as follows:

[0057] First, the set Retrieve Elements with a modulus length of 1 are added to the set , conveniently named as isolated point set. After the screening is completed, select The elements in are judged, and the currently selected point is set to , is the set of isolated points Middle elements, do the following:

[0058] Step 5.1, select isolated points , for isolated points is the center and the radius is Search the circular domain and remove all the The target points are classified according to the group they belong to, expressed as:

[0059]

[0060] Among them, all the points covered in the circular domain belong to Different clusters, Indicates that it is within the circular domain and belongs to A collection of grouped points.

[0061] Step 5.2, select The target point contained in each element in , if it meets the following two conditions: (1) If the point is used as the new delivery point of the group to which it belongs, it can Fusion into this group (2) does not violate the radar and no-fly zone constraints in the constraint fusion strategy. Then the target points that meet the two conditions are recorded into a new set middle.

[0062] Step 5.3, when After all elements are retrieved, if If the number of elements is greater than one, select Isolated point The nearest target point will Merge into the group to which the target point belongs, and update the delivery point and group information of the group.

[0063] The above process is performed until the isolated point set Complete the search at each point in the delivery point and finally get a new set of delivery points And the corresponding grouping set .

[0064] The beneficial effects of the present invention are:

[0065] (1) This paper comprehensively considers actual combat constraints such as enemy radar threats, no-fly zone restrictions, and drone range, and proposes a constraint fusion strategy to ensure that the selected delivery point avoids high-risk areas and meets the drone's mission execution conditions, significantly improving the safety of vehicle delivery and the feasibility of mission planning.

[0066] (2) By dynamically calculating the local density and relative distance of target points and combining a multi-round screening mechanism, the present invention can efficiently cover all target points in the battlefield, avoiding the coverage blind spots caused by traditional clustering algorithms ignoring actual constraints, and ensuring that the task is executed without omission.

[0067] (3) The present invention effectively reduces the number of redundant delivery points by introducing the strategy of isolated point fusion, thereby reducing the vehicle's navigation distance and mission execution cost, while improving the utilization rate of drones and achieving efficient resource allocation.

[0068] (4) The method proposed in this invention is not only applicable to UAV swarm mission planning in complex battlefield environments, but can also be extended to other fields that require multi-constraint collaborative decision-making, such as logistics distribution, disaster relief, etc., and has strong versatility and scalability.

[0069] In summary, the present invention can improve the efficiency and safety of drone swarms in mission execution in complex battlefield environments, reduce the risk of drone swarms being discovered and intercepted by the enemy, reduce the vehicle's navigation distance and mission execution time, and provide an efficient, safe and reliable solution for drone swarm mission planning in battlefield environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 Flowchart of the invention.

[0071] Figure 2 A schematic diagram of the battlefield.

[0072] Figure 3 A simplified diagram of the algorithm.

[0073] Figure 4 Schematic diagram of no belonging points after the first round of constraint fusion strategy execution.

[0074] Figure 5 Schematic diagram for selecting the delivery points for the first round of no-belonging points.

[0075] Figure 6 Schematic diagram of the grouping of target points after the second round of constraint fusion strategy execution.

[0076] Figure 7 Schematic diagram of no-attribution points after the second round of constraint fusion strategy execution.

[0077] Figure 8 Schematic diagram for selecting the delivery points for the second round of no-belonging points.

[0078] Figure 9 This is a schematic diagram of the target point grouping and isolated points after the third round of constraint fusion strategy execution.

[0079] Figure 10 Schematic diagram of grouping after isolated points are fused.

[0080] Figure 11 This is the final effect diagram of the algorithm. DETAILED DESCRIPTION

[0081] The present invention is further described below with reference to specific embodiments.

[0082] Figure 2 The initial schematic diagram shows the battlefield, where the battlefield area is 435 km × 435 km, with 200 homogeneous targets, 14 obstacles of various sizes and shapes, and 10 radars, showing the distribution of targets, obstacles, radars, and vehicle starting points. Obstacles are depicted in gray in the figure, target points are marked as dots on the figure, and vehicle starting points are diamond points. The radar is a red triangle, the area within the solid circle represents the absolute infeasible region of the vehicle, and the area within the red dotted line represents the coverage radius of the radar. This embodiment provides a collaborative decision-making method for vehicle delivery points and target groups for task planning in vehicle delivery mode, comprising the following steps:

[0083] The step 1 is specifically as follows:

[0084] Input the target point location information, first calculate the local density of each point, where the local density calculation uses the Gaussian kernel formula:

[0085] (1)

[0086] In formula (1), and Indicates two different target points. Indicates the target point The local density of Indicates the target point and The Euclidean distance of The preset neighborhood cutoff distance is The value of ,in Is a parameter that adjusts the range, and its value is less than , The range of the drone.

[0087] After completing the local density calculation of all target points, the relative distance calculation is carried out. Compared with The distance between the nearest target point with higher local density. The local density is the highest among all target points, so Target point The local density of , then the relative distance expression formula is:

[0088] (2)

[0089] Otherwise the target point The relative distance is expressed as:

[0090] (3)

[0091] After obtaining the local density and relative distance of each target point, the product of the local density and relative distance of each target point is calculated, and this is used as an indicator to sort the target points from large to small.

[0092] The step 2 is specifically as follows:

[0093] Step 2.1: Use the constraint fusion strategy to select delivery points;

[0094] After the target points are sorted, the delivery point selection process is carried out. Based on the sorted target points, this paper proposes a constraint fusion strategy that considers multiple constraints to select a feasible delivery point set. According to the sorted order, the target points are input one by one for retrieval. If the current target point meets the constraint fusion strategy, it will be selected as the delivery point and stored in the delivery point set. . After completing the retrieval of one target point, the next target point will be searched. When the selected delivery point can cover all target points or all target points have been searched once, the retrieval process stops. In the constraint fusion strategy, the present invention takes into account radar constraints, no-fly zone constraints, screening constraints and drone range constraints, and provides corresponding solutions and delivery point screening measures around the above three constraints. First, the screening conditions for the three constraints are introduced. When the target point meets the three screening conditions, it will be selected as the delivery point.

[0095] (1) Constructing screening conditions around radar constraints;

[0096] In this strategy, the target point will be selected as the delivery point only when it is outside the radar coverage. For the sake of convenience, in the constraint fusion strategy, the target point is represented as , and its position is represented by ,in Indicates a point The horizontal axis, Indicates a point The radar is represented by , then its position is expressed as , Respectively represent the horizontal and vertical coordinates of the radar. Then if the target point To be selected as a delivery point, the following conditions must be met:

[0097] (4)

[0098] in, For radar The detection coverage of the target point is expressed in formula (4). The distance between any radar and the delivery point must be greater than the radar's detection coverage, otherwise it cannot be selected as a delivery point.

[0099] (2) Constructing screening conditions around no-fly zone constraints;

[0100] In addition, the selection of delivery points needs to take into account the no-fly zones in the battlefield. The delivery points cannot be located within the no-fly zones. The constraints are as follows:

[0101] (5)

[0102] in, Indicates the target point location; Indicates the the coverage of the no-fly zones; Indicates the number of no-fly zones; Used to describe relationships that do not belong mathematically.

[0103] (3) Constructing screening conditions around screening constraints and drone range constraints;

[0104] This setting is used to address the limitations of the algorithm in the case of uneven target point density and the UAV range constraints. The specific form is as follows:

[0105] For the convenience of description, assume that several delivery points have been selected, and set the set of current delivery points as .in Represents the first delivery point selected in the set, and sets Delivery point is the center of the circle, The set of all target points within the circle coverage area of ​​radius . In the The delivery points are , and set Centered is the center of the circle, Is the set of target points covered by a circle with a radius of The latest cluster center added in is , the current search point is , then:

[0106] (6)

[0107] Formula (6) represents the target point of the current retrieval Cannot be located in an existing delivery points, otherwise it cannot be selected as a new delivery point. Representing a collection When the retrieved target point satisfies the three constraints of formula (4) to formula (6), it will be added to the delivery point set. .

[0108] Complete initial screening;

[0109] Step 2.2: Determine whether there are any remaining target points that cannot be covered by the current delivery point;

[0110] When step 2-1 is completed, the delivery point set is obtained , then calculate the number of target points covered by the delivery point and make a judgment, let the target point set be , the elements in the set are all target points, then:

[0111] (7)

[0112] in, Indicates the set of target points that are not covered by the current delivery point. For the convenience of description, it is named as the no-home point set. The target points that are not covered are no-home points. If the number of elements in is 0, execute step 5; if the number of elements in is not 0, execute step 3.

[0113] The step 3 is specifically as follows:

[0114] In this step, the corresponding delivery point selection will be carried out for the no-belonging points obtained in step 2. From step 2, the no-belonging point set is obtained Since this algorithm will be repeated for multiple rounds, each time step 3 is executed, In order to record the no-belonging points that appear during the entire algorithm iteration process, the global no-belonging point set is set as , Initially an empty set. The coverage radius control parameter for selecting delivery points for non-attribution points is used to control the coverage range of the selected delivery points in this step. The size limit is , perform the following steps:

[0115] Step 3.1, generate a 2D battlefield environment map and grid the battlefield. Extract all the no-belonging points and record the radius of each no-belonging point. The grid of the circular domain intersecting with , and record the non-attributable points in this round: ;

[0116] Step 3.2, calculate the value of the recorded grid, the value function is: ,in, Indicates that the grid is the center. The number of unattached points covered in the circular domain with a radius of Indicates radar threat weight; Indicates the radar threat to the grid, in the form of:

[0117] (8)

[0118] in, A parameter that represents the acceptable distance between the delivery point and the radar. When the grid is within the unacceptable range, its radar threat is infinite. Indicates radar With the current grid the distance between them; Represents an infinite number; Represents a radar ensemble.

[0119] Step 3.3, after completing the value calculation of all grids, select the grid with the highest value For the new delivery point, store the grid into the collection , and update the set of no-belonging points :

[0120] (9)

[0121] in, Indicates that it is in the maximum value grid There is no belonging point in the circular domain; represents the maximum value grid; Indicates the maximum range of the drone; Indicates the coverage radius control parameter for selecting delivery points for non-home points;

[0122] Finally, repeat steps 3-2 to 3-3 until There is no longer any no-home point in the system, and finally the no-home point delivery point set is obtained. .

[0123] The step 4 is specifically as follows:

[0124] Save the delivery point selected in step 3 Since the no-home points cannot find the delivery points through the constraint fusion strategy, and the corresponding delivery points are found through step 3, and because the target points will affect each other's local density and relative distance, it is necessary to discard the no-home points, update the local density and relative distance of the remaining target points, and re-sort them. First, the no-home points recorded in step 3 are From the total target point collection Remove and update the target point set. Then recalculate the set according to the formula in step 1 The local density and relative distance of each target point in the target area are calculated, and the target points are sorted in descending order according to the product of the local density and the relative distance. Finally, , return to step 2.

[0125] The step 5 is specifically as follows:

[0126] This step is established after all target points have corresponding delivery points, and there is a final delivery point set. , each element in the set represents a delivery point, Indicates the number of delivery points. Then the target grouping process is carried out, that is, the task of the target point is performed by the drone delivered by which delivery point. This step adopts the minimum distance principle to group the target point, that is, the target point is divided into the group of the closest delivery point, and finally the group set is obtained. , where each element in the set represents a group. For example, For Grouping of destination points as delivery points.

[0127] The step 6 is specifically as follows:

[0128] After completing the target point grouping in step 5, some isolated points may exist (a group containing only one target point). This step aims to integrate these isolated points into other groups by changing the delivery points of other groups. This reduces the number of delivery points, shortens the vehicle's navigation distance, and reduces potential dangers to the vehicle. The specific process is as follows:

[0129] First, the set Retrieve Elements with a modulus length of 1 are added to the set , conveniently named as isolated point set. After the screening is completed, select The elements in are judged, and the currently selected point is set to , is the set of isolated points Middle elements, do the following:

[0130] Step 5.1, select isolated points , for isolated points is the center and the radius is Search the circular domain and remove all the The target points are classified according to the group they belong to, expressed as:

[0131]

[0132] Among them, all the points covered in the circular domain belong to Different clusters, Indicates that it is within the circular domain and belongs to A collection of grouped points.

[0133] Step 5.2, select The target point contained in each element in , if it meets the following two conditions: (1) If the point is used as the new delivery point of the group to which it belongs, it can Fusion into this group (2) does not violate the radar and no-fly zone constraints in the constraint fusion strategy. Then the target points that meet the two conditions are recorded into a new set middle.

[0134] Step 5.3, when After all elements are retrieved, if If the number of elements is greater than one, select Isolated point The nearest target point will Merge into the group to which the target point belongs, and update the delivery point and group information of the group.

[0135] The above process is performed until the isolated point set Complete the search at each point in the delivery point and finally get a new set of delivery points And the corresponding grouping set .

[0136] In order to more intuitively see the implementation process of the algorithm proposed in this invention, an example is shown as follows: The battlefield environment is simplified as follows Figure 3 As shown in the figure, firstly, the constraint fusion strategy is used to select the preparatory delivery point for the target point, and then the existence of the unattached point is checked. Figure 4 In the figure, we mark the non-attributable points with light gray; then we perform grid sampling on the points that cannot be delivered, and find the delivery points by considering the radar threat and the number of target points covered, that is, Figure 5 The dots connected to the gray dots are shown in the figure. Based on the delivery points selected in the previous step, the constraint fusion strategy is executed to select the strong constraint delivery points outside the radar coverage range, and the target points are allocated according to the allocation rules. Figure 6 In the figure, black crosses represent the selected delivery points, and dashed lines connect each target point to its corresponding delivery point. Furthermore, we observe that the generation of delivery points from grid sampling causes changes in the decision value of the target point. This causes the delivery point selected by the constraint fusion strategy in this round to differ from the prepared delivery point in step 4. This, in turn, causes some of the initially selected points within the radar coverage area to become non-clustered points.

[0137] In order to make it more obvious to everyone, we Figure 6 、 7 In the process, boxes are used to locate the points where changes occur. When such points appear, grid sampling is performed again to find the corresponding delivery points. The solution is as follows: Figure 8 Similarly, based on all the selected delivery points, the constraint fusion strategy is executed and the target points are allocated according to the allocation rules. At this time, all target points have corresponding delivery points that can be responsible for them. The results are as follows: Figure 9 Considering that there are target points in a relatively isolated position, a group containing only one target is formed. In order to reduce the cost of task execution, such points are fused. The comparison results are shown in Figure 9 、 10 The final cluster assignment results are shown in the following figure: Figure 11 given.

[0138] The above-described embodiments merely express the implementation methods of the present invention, but should not be understood as limiting the scope of the present invention. It should be pointed out that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, which all fall within the scope of protection of the present invention.

Claims

1. A collaborative decision-making method for vehicle delivery points and target groups for task planning in vehicle delivery mode, characterized in that: The method for collaborative decision-making of vehicle delivery points and target groups includes the following steps: Step 1: Input the target point location information, use statistical formulas to solve and obtain the local density and relative distance of each point, and sort all targets in descending order according to the product of local density and relative distance; Step 2: Based on the sorted target points, a constraint fusion strategy that considers multiple constraints is proposed to select a set of feasible delivery points. Based on the selected delivery points, it is determined whether the delivery points can cover all target points. If so, step 5 is executed; otherwise, step 3 is executed. Step 3: Record the target points not covered in step 2, use the grid map to traverse the map grid, and select the delivery points for all recorded target points according to the grid value; Step 4: Fix the delivery point selected in step 3, remove the target point recorded in step 3 from the total target point set, and update the target point set; sort the updated target point set in descending order according to the product of local density and relative distance, and return to step 2; Step 5: Input the selected delivery point and group the targets using the minimum distance principle to obtain a target point group set with the delivery point as the center. Step 6: Input the target point group set, fuse the groups that only include one target point, fuse the isolated target points into other groups, and finally obtain the target point classification scheme and the corresponding delivery points.

2. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 1, characterized in that: The step 1 is specifically as follows: Input the target point location information, first calculate the local density of each point, where the local density calculation uses the Gaussian kernel formula: （1）， In formula (1), and Indicates two different target points. Indicates the target point The local density of Indicates the target point and The Euclidean distance of is the preset neighborhood cutoff distance; in order to fit the range limit of the drone, the cutoff distance is The value of ,in Is a parameter that adjusts the range, and its value is less than , The range of the drone; After completing the local density calculation of all target points, the relative distance calculation is carried out; the relative distance refers to the target point Compared with The distance between the target points with higher local density and the closest one; if the target point The local density is the highest among all target points, so Target point The relative distance is expressed as follows: （2）， Otherwise the target point The relative distance is expressed as: （3）， After obtaining the local density and relative distance of each target point, the product of the two is calculated for each target point, and this is used as an indicator to sort the target points from large to small.

3. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 2, characterized in that: The step 2 is specifically as follows: Step 2.1: Use the constraint fusion strategy to select delivery points; After completing the sorting of the target points, the delivery point selection process is carried out. Based on the sorted target points, a constraint fusion strategy that considers multiple constraints is proposed to select a feasible delivery point set. According to the sorted order, the target points are input one by one for retrieval. If the current target point meets the constraint fusion strategy, it will be selected as the delivery point and stored in the delivery point set. After completing the search for one target point, the search for the next target point will be carried out; when the selected delivery point can cover all target points or all target points have been searched once, the search process stops; in the constraint fusion strategy, radar constraints, no-fly zone constraints, screening constraints and drone range constraints are considered, and solutions and delivery point screening measures are given for radar constraints, no-fly zone constraints, screening constraints and drone range constraints respectively; when the target point meets the three screening conditions, it will be selected as the delivery point Step 2.2: Determine whether there are any remaining target points that cannot be covered by the current delivery point; When step 2.1 is completed, the delivery point set is obtained , then calculate the number of target points covered by the delivery point and make a judgment, let the target point set be , the elements in the set are all target points, then: （7）， in, Indicates the set of target points that are not covered by the current delivery point. For the convenience of description, it is named as the no-home point set. The target points that are not covered are no-home points. If the set If the number of elements in is 0, execute step 5; if the number of elements in is not 0, execute step 3.

4. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 3, characterized in that: In step 2.1, the screening conditions for the three constraints are as follows: 1) Build filter conditions around radar constraints; The target point will be selected as the delivery point only when it is outside the radar coverage area; in the constraint fusion strategy, the target point is represented as , and its position is represented by ,in Indicates a point The horizontal axis, Indicates a point The vertical coordinate of the radar is , then its position is expressed as , Respectively represent the horizontal and vertical coordinates of the radar; then if the target point To be selected as a delivery point, the following conditions must be met: （4）， in, For radar The detection coverage of the target point is expressed in formula (4). The distance to any radar must be greater than the radar's detection coverage, otherwise it cannot be selected as a delivery point; 2) Building screening conditions around no-fly zone constraints; The selection of delivery points must take into account the no-fly zones in the battlefield. Delivery points cannot be located within the no-fly zones. The constraints are as follows: （5）， in, Indicates the target point location; Indicates the the coverage of the no-fly zones; Indicates the number of no-fly zones; Used to describe relationships that do not belong mathematically; 3) Constructing screening conditions around screening constraints and drone range constraints; Assuming that several delivery points have been selected, let the set of current delivery points be ;in Represents the first delivery point selected in the set, and sets Delivery point is the center of the circle, The set of all target points within the circle coverage area of ​​radius In the The delivery points are , and set Centered is the center of the circle, Is the set of target points covered by a circle with a radius of The latest cluster center added in is , the current search point is , then: （6）， Formula (6) represents the target point of the current retrieval Cannot be located in an existing delivery points, otherwise it cannot be selected as a new delivery point. Representing a collection The number of delivery points; When the retrieved target point satisfies the three constraints of formula (4) to formula (6), it is added to the delivery point set. .

5. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 3, characterized in that: The step 3 is specifically as follows: The corresponding delivery point selection will be carried out for the no-belonging points obtained in step 2; from step 2, the no-belonging point set is obtained ; Since this algorithm will perform multiple rounds of iterations, each time step 3 is executed, In order to record the no-belonging points that appear during the entire algorithm iteration process, the global no-belonging point set is set as , Initially, it is an empty set; The coverage radius control parameter for selecting delivery points for non-attribution points is used to control the coverage range of the selected delivery points in this step. The size limit is .

6. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in a vehicle delivery mode according to claim 5, characterized in that: The steps of step 3 are as follows: Step 3.1, generate a 2D battlefield environment map and grid the battlefield; extract all the no-belonging points and record the radius of each no-belonging point. The grid of the circular domain intersecting with , and record the non-attributable points in this round: ; Step 3.2, calculate the value of the recorded grid, the value function is: ,in, Indicates that the grid is the center. The number of unattached points covered in the circular domain with a radius of Indicates radar threat weight; Indicates the radar threat to the grid, in the form of: （8）， in, A parameter that represents the acceptable distance between the delivery point and the radar. When the grid is within the unacceptable range, the radar threat is infinite. Indicates radar With the current grid the distance between them; Represents an infinite number; represents a radar ensemble; Step 3.3, after completing the value calculation of all grids, select the grid with the highest value For the new delivery point, store the grid into the collection , and update the set of no-belonging points : （9）， in, Indicates that it is in the maximum value grid There is no belonging point in the circular domain; represents the maximum value grid; Indicates the range of the drone; Indicates the coverage radius control parameter for selecting delivery points for non-home points; Finally, repeat steps 3-2 to 3-3 until There is no longer any no-home point in the system, and finally the no-home point delivery point set is obtained. .

7. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 5, characterized in that: The step 4 is specifically as follows: Save the delivery point selected in step 3 Since the no-home point cannot find the delivery point through the constraint fusion strategy, and the corresponding delivery point is found through step 3, and because the target points will affect each other's local density and relative distance, it is necessary to abandon the no-home point, update the local density and relative distance of the remaining target points, and re-sort them; first, the no-home point recorded in step 3 From the total target point collection Remove and update the target point set; then recalculate the set according to the formula in step 1 The local density and relative distance of each target point in the target area are calculated, and the target points are sorted in descending order according to the product of the local density and the relative distance. Finally, , return to step 2.

8. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 7, characterized in that: The step 5 is specifically as follows: The step 5 is established after all target points have corresponding delivery points, and there is a final delivery point set. , each element in the set represents a delivery point, Indicates the number of delivery points; Then the target grouping process is carried out, that is, the task of the target point is performed by the drone delivered by which delivery point; the grouping is performed using the minimum distance principle, that is, the target point is divided into the group of the delivery point closest to it, and finally the group set is obtained. , where each element in the set represents a group, For Grouping of destination points as delivery points.

9. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 8, characterized in that: The step 6 is specifically as follows: First, the set Retrieve Elements with a modulus length of 1 are added to the set , for the convenience of description, it is named as the isolated point set; After the screening is completed, select The elements in are judged, and the currently selected point is set to , is the set of isolated points Middle elements, do the following: Step 5.1, select isolated points , for isolated points is the center and the radius is Search the circular domain and remove all the The target points are classified according to the group they belong to, expressed as: ， Among them, all the points covered in the circular domain belong to Different clusters, Indicates that it is within the circular domain and belongs to A collection of grouped points; Step 5.2, select The target points contained in each element in , if they meet the two conditions, will be recorded into a new set middle; Step 5.3, when After all elements are retrieved, if If the number of elements is greater than one, select Isolated point The nearest target point will Merge into the group to which the target point belongs, and update the delivery point and group information of the group; Execute the above process until the isolated point set Complete the search at each point in the delivery point and finally get a new set of delivery points And the corresponding grouping set .

10. The method for collaborative decision-making of vehicle delivery points and target groups for task planning in vehicle delivery mode according to claim 9, characterized in that: The two conditions of step 5.2 are: (1) If the point is used as a new delivery point for the group to which it belongs, it can Fusion into this group (2) does not violate the radar and no-fly zone constraints in the constraint fusion strategy.

Images