Multi-unmanned aerial vehicle cooperative target searching method and system

Abstract

The invention discloses a multi-unmanned aerial vehicle cooperative target searching method and system, and the method comprises the steps: carrying out the discrete coordinate rasterization of a searching region, setting the initial parameters of an unmanned aerial vehicle, and taking the position of a target entering the region as the starting point position of the unmanned aerial vehicle; all the unmanned aerial vehicles search at the starting point positions at the same time, and region search is carried out by using the optimized region walk algorithm; obtaining the grid score of the grid where the information is located according to the correlation degree of the search information and the target; according to the grid score, obtaining a weight coefficient of unmanned aerial vehicle search flight path planning, and according to a linear function of the weight coefficient in each direction, carrying out unmanned aerial vehicle flight path planning; and if more than two features with the motion direction of the target are found in the search process, predicting the position of the target, and surrounding the target. The method has the advantages of being easy to implement and low in complexity, and target searching tasks can be efficiently and rapidly completed.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) path planning, and in particular to a multi-UAV collaborative target search method and system. Background Art

[0002] In recent years, drones have been widely used in many fields such as agricultural seeding, power inspection, disaster relief, etc. due to their light weight, relatively low manufacturing cost, simple maintenance, and direct take-off and landing. However, when performing the above tasks, a single drone faces many difficulties, such as low survival rate, long execution time, etc., and lacks high execution efficiency and completion. The collaborative execution of multiple drones has emerged and has become a hot research topic today, and has been widely used in emergency rescue.

[0003] General search path planning problems can be divided into two categories: (1) when searching for multiple targets, quickly capture the target; (2) search as much as possible in areas without prior information, pursuing a high coverage rate. When searching with a drone, the traditional default is that all unit areas have the same priority, which will result in the drone being unable to quickly prioritize the search, which will waste time and is not suitable for emergency search and rescue. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for collaborative target search of multiple UAVs, a search method for trajectory planning using the idea of ​​local optimization, which realizes multi-UAV search path planning, and uses the search time as the path criterion to improve the timeliness of the search.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A multi-UAV collaborative target search method comprises the following steps:

[0007] S1. Perform discrete coordinate rasterization on the search area, set the initial parameters of the UAV, and take the position where the target P enters the area as the starting position of the UAV.

[0008] S2. All drones search at the starting point at the same time, using the optimized regional wandering algorithm to search the area, and synchronize the obtained search information to the search map of all drones within the communication range; the search information is the feature with the movement direction of the target.

[0009] S3. According to the correlation degree between the search information and the target, the grid score of the grid where the information is located is obtained and synchronized to the search map of all drones within the communication range.

[0010] S4. Obtain the weight coefficient of the UAV search track planning according to the grid score, and perform the UAV track planning according to the linear function of the weight coefficient in each direction.

[0011] S5. If more than two pieces of search information are found during the search process, the target position is predicted and the target is captured. Otherwise, the process returns to step S4.

[0012] S6. When the target appears within the visual detection range of the multiple UAVs, it indicates that the target has been successfully captured, and the search mission ends; otherwise, steps S4-S5 are repeated.

[0013] Furthermore, in step S1, the search area is set to M and the size of the search area is set to L. x ×L y , the planned number of grids is N×A, then the grid size l x , l y The calculation formula is as follows:

[0014]

[0015] gxy={(x,y)(x,y)∈M}

[0016] Where N is the number of columns of the grid, A is the number of rows of the grid, x is the horizontal distance, y is the vertical distance, g is xy A raster representing the search area.

[0017] The initial parameters of the UAV include the lateral detection angle γ uc , the maximum detection distance R uc , drone search speed, drone maximum roundup speed and drone minimum roundup speed.

[0020] Furthermore, in step S2, performing the area search includes the following sub-steps:

[0021] S201. The first turning interval of the hth UAV is:

[0022]

[0023] in, Indicates the first turning angle range of the drone. m represents the total number of drones.

[0024] S202, the UAV randomly generates a steering value within its first steering angle range to turn, moves according to a random step length and searches, and after the movement is completed, marks the coordinate point where the current UAV is located as searched and adds it to the search graph.

[0025] S203, randomly generate an angle θ in the interval [0,360°], the drone rotates θ and moves a random step to the next coordinate point, and updates the search graph.

[0026] S204, repeat steps S202-S203 until a feature with the target's moving direction is found in the search area.

[0027] Furthermore, in step S3, obtaining the grid score includes the following:

[0028] The different probabilities of marking the features with the target's moving direction found by the drone during the search process reflect the probability that the target is located in the corresponding grid, and the grid score corresponding to the grid is obtained. The specific formula is:

[0029]

[0030] Among them, (i,j) represents the center coordinates of the grid discovered by the current drone; S(i,j) represents the grid score of (i,j); P(i,j) represents the weight of the feature in the grid; Covered(i,j) indicates whether the grid has been searched, 0 means it has been searched, and 1 means it has not been searched; (x,y) represents the coordinates of the current drone.

[0031] Furthermore, in step S4, the trajectory planning of the UAV includes the following contents:

[0032] The grid scores of the n grids where the searched features are located are normalized to obtain the weight coefficients of the direction vectors from all drones to the center of each grid. The weight coefficient of the direction vector of the rth grid is K r .

[0033] After discovering the feature, the steering angle function D of the hth UAV h for:

[0034]

[0035] in, represents the direction vector randomly generated by the drone, represents the direction vector from the h-th UAV to the center of the r-th grid, N represents the number of columns of the grid, and A represents the number of rows of the grid.

[0036] After all drones have completed searching a grid, the searched area will be automatically updated to the search map of all drones within the communication radius. Drones will allocate search tasks based on their own positions and grid scores. Drones will prioritize searching grids with the largest grid scores.

[0037] Furthermore, in step S5, the target capture includes the following contents:

[0038] A linear fit is performed on the position of the feature with the target's moving direction. If the fitted straight line passes through the center point of a grid in the direction reflected by the feature, the grid is identified as the location of the target and the target is captured in the grid.

[0039] Based on the size of each grid score, the distance negotiation method is used to allocate the capture area. The flight speed of each UAV is adjusted according to the time consistency constraint so that each UAV can reach its paired capture point at the same time, forming a capture formation and conducting the capture.

[0040] Furthermore, the present invention also proposes a multi-UAV cooperative target search system, comprising:

[0041] The search initial setting module is used to rasterize the search area into discrete coordinates, set the initial parameters of the drone, and take the position where the target P enters the area as the starting position of the drone.

[0042] The regional search module is used for all drones to search at the starting point simultaneously. It uses the optimized regional wandering algorithm to perform regional search and synchronizes the obtained search information to the search map of all drones within the communication range; the search information is the feature with the movement direction of the target.

[0043] The grid score acquisition module is used to obtain the grid score of the grid where the search information is located according to the degree of correlation between the search information and the target, and synchronize it to the search map of all drones within the communication range.

[0044] The trajectory planning module is used to obtain the weight coefficient of the UAV search trajectory planning according to the grid score, and to plan the UAV's trajectory according to the linear function of the weight coefficient in each direction.

[0045] The target prediction module is used to predict the target position and capture the target if more than two search information are found during the search process. Otherwise, it returns to the track planning module.

[0046] The target capture module is used to terminate the search mission when the target appears within the visual detection range of multiple UAVs, indicating that the target has been successfully captured; otherwise, the trajectory planning module and target prediction module are repeated.

[0047] Furthermore, the present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the multi-UAV collaborative target search method described above when executing the computer program.

[0048] Furthermore, the present invention also proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is run by a processor, the multi-UAV collaborative target search method is executed.

[0049] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0050] The method proposed in the present invention can flexibly and timely search key areas according to the findings in the search, so that the search and rescue time is greatly shortened, not only the persons to be rescued can be found as soon as possible, but also the operation cost is saved. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a visual detection area map of the unmanned aerial vehicle of the present invention.

[0052] Figure 2 It is an overall implementation flow chart of the present invention.

[0053] Figure 3 It is a rasterized schematic diagram of the present invention.

[0054] Figure 4 Schematic diagram of environment initialization according to an embodiment of the present invention.

[0055] Figure 5 Schematic diagram of prediction results of an embodiment of the present invention.

[0056] Figure 6 This is a center distribution map of sub-areas of the drone search path according to an embodiment of the present invention.

[0057] Figure 7 1 is a graph showing the path length results of multiple UAVs obtained using different methods according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0059] like Figure 1 As shown in Figure 2, when the target Ρ is at t 0 Enter a finite two-dimensional space S at time t 0 -t 1 The target P moves in this area. The motion state of the target P is unknown, and the features reflecting the target's trajectory will be randomly dropped when passing through the path, and the amount of information contained in the features is different. m drones are used to conduct collaborative target search for the target P. Since the two-dimensional area is considered in this embodiment, this mission scenario ignores the parameters such as the installation tilt angle and longitudinal detection angle of the visual sensor in the actual drone, and only considers the lateral detection angle γ ucAnd the maximum detection distance R uc , l uc is the maximum longitudinal detection distance of the UAV, the hth UAV u h The visual detection area is centered on the current position of the drone and is divided into two parts: uc is the radius, γ uc is the sector area of ​​the central angle of the circle, x is the horizontal distance, and y is the vertical distance.

[0060] To achieve the above object, the present invention proposes a multi-UAV collaborative target search method, such as Figure 2 As shown, the specific steps are as follows:

[0061] S1. Parameter initialization: The search area is rasterized into discrete coordinates, and the position where the target P enters the area is used as the starting position of the UAV. The specific contents are as follows: Figure 3 As shown:

[0062] M represents the search area, the size of which is L x ×L y , the planned number of grids is N×A, then the grid size l x , l y The calculation formula is as follows:

[0063]

[0064] gxy={(x,y)|(x,y)∈M}

[0065] Where N is the number of columns of the grid, A is the number of rows of the grid, x is the horizontal distance, y is the vertical distance, g is xy Represents any grid in the search area.

[0066] At this time xy It only carries coordinate information, which is the basic information of each grid, and more information will be given according to the specific situation.

[0067] Set the initial parameters of the drone, including the lateral detection angle γ uc Maximum detection distance R uc Drone search speed,

[0068] ,, the maximum capture speed of the drone and the minimum capture speed of the drone.

[0069] S2, Regional Search: All drones search at the starting point at the same time. To achieve a larger coverage rate, an optimized regional wandering algorithm is used to search the area, and the search information obtained is synchronized to the search map of all drones within the communication range. The specific contents are:

[0070] S201. The first turning interval of the hth UAV is:

[0071]

[0072] in, Indicates the first turning angle range of the drone. m represents the total number of drones.

[0073] S202, the UAVs randomly generate steering values ​​within their respective first steering angle ranges to perform steering, move according to random step lengths and perform searching, and after the movement is completed, mark the coordinate point where the current UAV is located as searched and add it to the search area.

[0074] S203, randomly generate an angle θ in the interval [0,360°], the drone rotates θ and moves a random step to the next coordinate point, and updates the search graph.

[0075] S204, repeating steps S202-S203 until a feature with the target's moving direction is found in the search area, and then stopping. The feature is the search information.

[0076] S3, local optimization: assign grid scores to the grid where the search information is located according to the degree of correlation between the search information and the target, and synchronize it to the search map of all drones within the communication range. The specific contents are:

[0077] During the search process of the drone in the area, some features with the target's movement direction may be found. Different features are marked with different probabilities, which reflects the degree of association between the feature and the target. The degree of association between the feature and the target reflects the probability that the target is located in the grid, and the grid score corresponding to the grid is obtained. Since the drone should give priority to searching the grid with a high probability of target movement and a close distance to itself, the grid score is positively correlated with the probability of finding the feature in the grid and negatively correlated with the distance. The specific formula is:

[0078]

[0079] Among them, (i,j) represents the center coordinates of the grid discovered by the current drone; S(i,j) represents the grid score of (i,j); P(i,j) represents the weight of the feature in the grid; Covered(i,j) indicates whether the grid has been searched, 0 means it has been searched, and 1 means it has not been searched; (x,y) represents the coordinates of the current drone.

[0080] S4, track planning: Obtain the weight coefficient of the drone search track planning based on the grid score, and plan the drone's track based on the linear function of the weight coefficient in each direction to ensure that the drone prioritizes searching grids with short distances and large weights. The specific contents are:

[0081] The grid scores of the n grids where the searched features are located are normalized to obtain the weight coefficients of the direction vectors from all drones to the center of each grid. The weight coefficient of the direction vector of the rth grid is K r .

[0082] After discovering the feature, the steering angle function D of the hth UAV h for:

[0083]

[0084] in, represents the direction vector randomly generated by the drone, Represents the direction vector from the h-th UAV to the center of the r-th grid.

[0085] After each drone completes searching a grid, it automatically updates the searched area to the search map of all drones within the communication radius. The weight of the searched grid in the entire search area will be reduced, thereby reducing the probability of repeated searches. Each drone allocates search tasks according to its own position and grid score. After the task is allocated, the drone will first search for grids with a greater probability of the target appearing and closer to its own position (i.e., grids with the largest grid scores), ensuring that grids with a greater probability around the feature are searched as much as possible.

[0086] S5, judging whether the effective information is sufficient: if two or more features with the target's moving direction are successfully found during the search process, the target position is predicted and the target is captured, otherwise, the search is repeated in step S4. The specific contents are:

[0087] A linear fit is performed on the position of the feature with the target's moving direction. If the fitted straight line passes through the center point of a grid in the direction reflected by the feature, the grid is identified as the location of the target and the target is captured in the grid.

[0088] In terms of target capture, the distance negotiation method is used to allocate the capture area based on the size of each grid score. The flight speed of each drone is adjusted according to the time consistency constraint (that is, drones with longer distances have faster speeds, while drones with shorter distances have slower speeds) so that each drone can reach its paired capture point at the same time, forming a capture formation for capture.

[0089] S6. Determine whether the target is successfully captured: When the target appears within the visual detection range of the multiple UAVs, it indicates that the target is successfully captured, and the mission ends; otherwise, repeat steps S4-S5.

[0090] The simulation tool used in this embodiment is Matlab 2021b. The simulation area is a two-dimensional plane, and the task area is a 160×160 square plane, which is discretized and gridded. The coordinate point where the target P enters the area is (40, 0). The target P takes this as the starting point at time t 0 -t 1 Randomly move within the target search area and drop features containing location information during the movement. Set the simulation parameters as shown in Table 1:

[0091] Table 1 Simulation parameters

[0092] parameter Numeric <![CDATA[γ uc / °]]> 120 <![CDATA[R uc / M]]> 12 UAV search speed v / (m / s) 5 UAV maximum capture speed v / (m / s) 6 Minimum capture speed of drone v / (m / s) 1

[0093] like Figure 4 As shown, the randomly generated path of the target in the search area is shown, as well as the features and their distribution locations that are randomly dropped during the journey.

[0094] like Figure 5 As shown in the figure, the paths taken by each drone are displayed. It can be seen that after drone 1 discovers the feature with target path information, the weight of each drone's search direction toward the direction of the feature increases, and the track direction of each drone deviates toward that direction. When a feature with a larger weight is discovered again, a straight line fitting is performed on the positions of the two clues, and drones 1, 2, and 4, which are closer, encircle and capture the target on the target's trajectory.

[0095] like Figure 6 As shown, in the method proposed by the present invention, under the conditions of multi-UAV collaborative search and shared search map, the search path areas of each UAV have less overlap and higher efficiency. During the search process, the UAV will prioritize and search in areas with higher possibilities as quickly as possible based on the discovered features. This method is more suitable for emergency rescue.

[0096] In order to compare and analyze the method proposed in the present invention with the traversal method in a more scientific and reasonable manner and verify their advantages and disadvantages, this embodiment sets 5 search areas of different sizes, from large to small, 2.83 km 2 、2.3km 2 、1.87km 2 、1.54km 2 and 0.93km 2 , the number of drone racks allocated to each area is 35, 30, 25, 20 and 15 respectively. The rest of the parameter settings are consistent with Table 1. Through the starting point determination method mentioned in the present invention, the two methods use the same starting point to start the search. The line graphs of the two methods in terms of path degree L are shown in Figure 7The line graph directly and clearly reflects that in the five search areas with different sizes, the method proposed by the present invention can significantly shorten the path length, thereby effectively reducing the search time, verifying its high efficiency and superiority.

[0097] The present invention proposes a simple and easy-to-implement multi-UAV collaborative target search method for emergency rescue and target search. First, a random walk is performed in the mission area, and then based on the existing search information, the track planning is performed using the idea of ​​local optimization. By considering the degree of correlation between the information at each moment and the search target, the weight coefficient of the search UAV track planning is established, and then the track planning of the UAV is performed by the linear function of each weight coefficient. While ensuring the priority search, the search path and search time are effectively shortened. The simulation results show that the method proposed by the present invention can improve the timeliness of the search, and is efficient and easy to implement.

[0098] The embodiment of the present invention also proposes a multi-UAV collaborative target search system, including a search initialization setting module, an area search module, a grid score acquisition module, a track planning module, a target prediction module, a target capture module and a computer program that can be run on a processor. It should be noted that each module in the above system corresponds to the specific steps of the method provided in the embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. For technical details not described in detail in this embodiment, please refer to the method provided in the embodiment of the present invention.

[0099] The embodiment of the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. It should be noted that when the processor executes the computer program, the specific steps of the method provided in the embodiment of the present invention are corresponding to the specific steps of the method provided in the embodiment of the present invention, and the processor has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided in the embodiment of the present invention.

[0100] The embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. It should be noted that when the computer program is executed by the processor, it corresponds to the specific steps of the method provided in the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided in the embodiment of the present invention.

[0101] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A multi-UAV collaborative target search method, characterized in that: include: S1. Perform discrete coordinate rasterization on the search area, set the initial parameters of the UAV, and take the position where the target P enters the area as the starting position of the UAV; S2. All drones search at the starting point at the same time, using the optimized area wandering algorithm to search the area, and synchronize the obtained search information to the search map of all drones within the communication range; the search information is the feature with the movement direction of the target; S3. According to the correlation between the search information and the target, the grid score of the grid where the information is located is obtained and synchronized to the search map of all drones within the communication range; S4, obtaining the weight coefficient of the UAV search track planning according to the grid score, and performing the UAV track planning according to the linear function of the weight coefficient in each direction; S5. If more than two pieces of search information are found during the search process, the target position is predicted and the target is captured. Otherwise, the process returns to step S4. S6. When the target appears within the visual detection range of the multiple UAVs, it indicates that the target has been successfully captured, and the search mission ends; otherwise, steps S4-S5 are repeated.

2. The multi-UAV cooperative target search method according to claim 1, characterized in that: In step S1, the search area is set to M and the size of the search area is set to L. x ×L y , the planned number of grids is N×A, then the grid size l x , l y The calculation formula is as follows: gxy={(x,y)(x,y)∈M} Where N is the number of columns of the grid, A is the number of rows of the grid, x is the horizontal distance, y is the vertical distance, g is xy Represents a grid in the search area; The initial parameters of the UAV include the lateral detection angle γ uc Maximum detection distance R uc UAV search speed, UAV, Maximum capture speed and minimum capture speed of drone.

3. The multi-UAV cooperative target search method according to claim 1, characterized in that: In step S2, performing area search includes the following sub-steps: S201. The first turning interval of the hth UAV is: in, Indicates the first turning angle range of the drone. m represents the total number of drones; S202, the UAV randomly generates a steering value within its first steering angle range to turn, moves according to a random step length and searches, and after the movement is completed, marks the coordinate point where the current UAV is located as searched and adds it to the search graph; S203, randomly generate an angle θ in the interval [0,360°], the drone rotates θ and moves a random step to the next coordinate point, and updates the search graph; S204, repeat steps S202-S203 until a feature with the target's moving direction is found in the search area.

4. The multi-UAV cooperative target search method according to claim 1, characterized in that: In step S3, obtaining the grid score includes the following: The different probabilities of marking the features with the target's moving direction found by the drone during the search process reflect the probability that the target is located in the corresponding grid, and the grid score corresponding to the grid is obtained. The specific formula is: Among them, (i,j) represents the center coordinates of the grid discovered by the current drone; S(i,j) represents the grid score of (i,j); P(i,j) represents the weight of the feature in the grid; Covered(i,j) indicates whether the grid has been searched, 0 means it has been searched, and 1 means it has not been searched; (x,y) represents the coordinates of the current drone.

5. The multi-UAV cooperative target search method according to claim 1, characterized in that: In step S4, the trajectory planning of the UAV includes the following: The grid scores of the n grids where the searched features are located are normalized to obtain the weight coefficients of the direction vectors from all drones to the center of each grid. The weight coefficient of the direction vector of the rth grid is K r ; After discovering the feature, the steering angle function D of the hth UAV h for: Among them, θ represents the direction vector randomly generated by the drone, J hr represents the direction vector from the h-th UAV to the center of the r-th grid, N represents the number of columns of the grid, and A represents the number of rows of the grid; After all drones have completed searching a grid, the searched area will be automatically updated to the search map of all drones within the communication radius. Drones will allocate search tasks based on their own positions and grid scores. Drones will prioritize searching grids with the largest grid scores.

6. The multi-UAV cooperative target search method according to claim 1, characterized in that: In step S5, the target capture includes the following contents: Perform linear fitting on the position of the feature with the target's moving direction. If the fitted straight line passes through the center point of a grid in the direction reflected by the feature, the grid is identified as the location of the target, and the target is captured in the grid. Based on the size of each grid score, the distance negotiation method is used to allocate the capture area. The flight speed of each UAV is adjusted according to the time consistency constraint so that each UAV can reach its paired capture point at the same time, forming a capture formation and conducting the capture.

7. A system for the multi-UAV cooperative target search method according to claim 1, characterized in that: include: The search initial setting module is used to rasterize the search area into discrete coordinates, set the initial parameters of the UAV, and take the position where the target P enters the area as the starting position of the UAV; The regional search module is used for all drones to search at the starting point at the same time. It uses the optimized regional wandering algorithm to search the area and synchronize the obtained search information to the search map of all drones within the communication range. The search information is the feature with the movement direction of the target. The grid score acquisition module is used to obtain the grid score of the grid where the search information is located according to the degree of correlation between the search information and the target, and synchronize it to the search map of all drones within the communication range; The trajectory planning module is used to obtain the weight coefficient of the UAV search trajectory planning according to the grid score, and to plan the UAV trajectory according to the linear function of the weight coefficient in each direction; The target prediction module is used to predict the target position and capture the target if more than two search information are found during the search process, otherwise it returns to the track planning module; The target capture module is used to terminate the search mission when the target appears within the visual detection range of multiple UAVs, indicating that the target has been successfully captured; otherwise, the trajectory planning module and target prediction module are repeated.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the multi-UAV collaborative target search method described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the multi-UAV collaborative target search method described in any one of claims 1 to 6 is executed.

Images