Unmanned aerial vehicle group task allocation and obstacle avoidance method based on Hungary-APF model

Abstract

The invention relates to an unmanned aerial vehicle group task allocation and obstacle avoidance method based on a Hungary-APF model. The method comprises the following steps: constructing a task allocation model according to an unmanned aerial vehicle task allocation strategy and a cost value of completing a task by an unmanned aerial vehicle; a cost value matrix is constructed according to cost values of all unmanned aerial vehicles in the unmanned aerial vehicle group for completing tasks, and a decision matrix is constructed according to task allocation strategies of all unmanned aerial vehicles in the unmanned aerial vehicle group; and taking the decision matrix as the input of an unmanned aerial vehicle task allocation strategy of the task allocation model, taking the cost value matrix as the input of the cost value, and solving the task allocation model by adopting a Hungary algorithm to obtain an optimal unmanned aerial vehicle task allocation result. Constructing an APF obstacle avoidance model, wherein the APF obstacle avoidance model comprises static repulsive force, dynamic repulsive force and gravitational force; and inputting the task allocation result into the APF obstacle avoidance model, and solving to obtain a flight path planning result of obstacle avoidance of each unmanned aerial vehicle in the unmanned aerial vehicle group. By adopting the method, the cargo distribution efficiency of the unmanned aerial vehicle group and the obstacle avoidance path precision can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of drone swarm express delivery control technology, and in particular to a drone swarm task allocation and obstacle avoidance method based on the Hungarian-APF model. Background Art

[0002] With the development of drone-related technologies, drone swarm delivery has attracted widespread attention from both industry and academia. Drone swarm delivery is an intelligent logistics technology based on the collaborative operation of multiple drones. It accomplishes complex delivery tasks through swarm collaboration, even when the capabilities of individual drones are limited. It has broad application prospects in urban logistics and delivery.

[0003] Drone swarm delivery can effectively improve the efficiency of material distribution through the coordinated cooperation of multiple drones and autonomous trajectory planning. However, the complex urban environment poses significant challenges to swarm delivery. On the one hand, how can the overall benefits of swarm delivery be maximized through reasonable task allocation? On the other hand, how can the swarms avoid complex urban obstacles through effective trajectory planning?

[0004] The problem of joint task allocation and trajectory planning for drone swarms has attracted widespread attention from scholars. In a scenario involving drones performing multi-target reconnaissance missions, Zheng Kai et al. employed a genetic algorithm to first allocate tasks and then the A* algorithm for trajectory planning, verifying the effectiveness of their approach through software. Li Meng et al. developed a gridded trajectory model, ingeniously combining the ant colony algorithm with the APF algorithm for trajectory planning. Fei Chen et al. used the EEFO algorithm to plan strike trajectories for drone swarms, demonstrating high mission completion rates through simulation.

[0005] Multiple time slots, multiple tasks, and complex environments are the main challenges facing the application of drone swarm delivery in cities. However, as mentioned above, previous work in this area has focused on combining heuristic algorithms with local path planning algorithms to accomplish task allocation and trajectory planning in single time slots and relatively simple environments. In terms of task allocation, heuristic algorithms suffer from high computational complexity and difficulty in reaching an optimal solution. The Hungarian algorithm, on the other hand, can achieve a strictly proven optimal solution with low computational complexity and simple operation. In terms of trajectory planning, the traditional APF algorithm suffers from the disadvantages of being prone to local minima and experiencing large trajectory fluctuations. Summary of the Invention

[0006] Based on this, it is necessary to provide a drone swarm task allocation and obstacle avoidance method based on the Hungarian-APF model, which can effectively improve the efficiency of drone swarm cargo delivery and the accuracy of obstacle avoidance paths, in order to address the above technical problems.

[0007] A method for task allocation and obstacle avoidance of a drone swarm based on the Hungarian-APF model, the method comprising:

[0008] Based on the UAV task allocation strategy and the cost of the UAV completing the task, a task allocation model is constructed. The cost is determined by the UAV's revenue from completing the task and the distance to complete the task.

[0009] According to the cost values of all drones in the drone swarm completing tasks, a cost value matrix is constructed, and according to the task allocation strategy of all drones in the drone swarm, a decision matrix is constructed.

[0010] The decision matrix is used as the input of the UAV task allocation strategy of the task allocation model, the cost matrix is used as the input of the cost value, and the Hungarian algorithm is used to solve the task allocation model to obtain the optimal UAV task allocation result.

[0011] Build an APF obstacle avoidance model. The APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. Dynamic repulsion is used when the drone approaches an obstacle. It is first subjected to dynamic repulsion, and then to dynamic repulsion and static repulsion. When the static repulsion and gravity balance, the drone's velocity vector is not zero.

[0012] The task allocation results are input into the APF obstacle avoidance model to obtain the trajectory planning result of each UAV in the UAV swarm for obstacle avoidance.

[0013] A UAV swarm task allocation and obstacle avoidance device based on the Hungarian-APF model, the device comprising:

[0014] The task allocation model building module is used to build a task allocation model based on the UAV task allocation strategy and the cost of the UAV completing the task. The cost is determined by the UAV's reward for completing the task and the distance to complete the task.

[0015] The decision matrix building module is used to build a cost matrix based on the cost of completing tasks by all drones in the drone swarm, and to build a decision matrix based on the task allocation strategy of all drones in the drone swarm.

[0016] The optimal task allocation module is used to use the decision matrix as the input of the UAV task allocation strategy of the task allocation model, the cost matrix as the input of the cost value, and use the Hungarian algorithm to solve the task allocation model to obtain the optimal UAV task allocation result.

[0017] The obstacle avoidance model building module is used to build the APF obstacle avoidance model. The APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. Dynamic repulsion is used when the drone approaches an obstacle. It is first subjected to dynamic repulsion, and then to dynamic repulsion and static repulsion. When the static repulsion and gravity balance, the drone's velocity vector is not zero.

[0018] The trajectory planning module is used to input the task allocation results into the APF obstacle avoidance model to obtain the trajectory planning results for obstacle avoidance of each UAV in the UAV swarm.

[0019] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0020] A task allocation model is constructed based on the UAV task allocation strategy and the cost value of the UAV completing the task; the cost value is determined by the revenue of the UAV completing the task and the distance to complete the task.

[0021] According to the cost values of all drones in the drone swarm completing tasks, a cost value matrix is constructed, and according to the task allocation strategy of all drones in the drone swarm, a decision matrix is constructed.

[0022] The decision matrix is used as the input of the UAV task allocation strategy of the task allocation model, the cost matrix is used as the input of the cost value, and the Hungarian algorithm is used to solve the task allocation model to obtain the optimal UAV task allocation result.

[0023] Build an APF obstacle avoidance model. The APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. Dynamic repulsion is used when the drone approaches an obstacle. It is first subjected to dynamic repulsion, and then to dynamic repulsion and static repulsion. When the static repulsion and gravity balance, the drone's velocity vector is not zero.

[0024] The task allocation results are input into the APF obstacle avoidance model to obtain the trajectory planning result of each UAV in the UAV swarm for obstacle avoidance.

[0025] The aforementioned drone swarm task allocation and obstacle avoidance method, based on the Hungarian-APF model, first constructs a task allocation model based on the drone task allocation strategy and the cost of completing the task (determined by the task reward and the distance to complete the task). Based on this model, a cost matrix and a decision matrix are further constructed. The cost matrix integrates the cost of completing the task for all drones, while the decision matrix reflects the task allocation strategy for all drones. Subsequently, the decision matrix and cost matrix serve as inputs for the task allocation strategy and cost value, respectively, and the Hungarian algorithm is introduced to solve the task allocation model. This algorithm, with its efficient computational logic, can quickly traverse multiple possible task allocation scenarios and accurately find the optimal solution. Compared to traditional heuristic algorithms, this significantly reduces the amount of computation and effectively improves the efficiency and accuracy of task allocation. Furthermore, for trajectory planning, an APF obstacle avoidance model is constructed that incorporates static repulsion, dynamic repulsion, and gravity. The innovative design of the dynamic repulsion mechanism ensures that when approaching an obstacle, the drone is first subjected to dynamic repulsion, followed by both dynamic and static repulsion. This ensures that the drone's velocity vector is non-zero when the static repulsion and gravitational forces are balanced, thus avoiding local minima and reducing track fluctuations. The optimal task allocation result is input into the APF obstacle avoidance model, ultimately yielding a trajectory planning result for each drone's obstacle avoidance. Furthermore, by combining the efficient Hungarian algorithm for task allocation with an improved APF obstacle avoidance model for trajectory planning, this two-pronged approach not only optimizes the task allocation process but also improves the accuracy of trajectory planning, effectively enhancing the efficiency of drone swarm delivery and the accuracy of obstacle avoidance paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 FIG1 is a flow chart of a method for task allocation and obstacle avoidance of a drone swarm based on the Hungarian-APF model in one embodiment;

[0027] Figure 2 A schematic diagram of a swarm delivery model system in a complex urban environment in one embodiment;

[0028] Figure 3 A schematic diagram of dynamic repulsion in one embodiment;

[0029] Figure 4 Schematic diagram of the process of dynamic repulsion and static repulsion in one embodiment;

[0030] Figure 5 A schematic diagram of setting a virtual target point in one embodiment;

[0031] Figure 6 1 is a structural block diagram of a UAV swarm task allocation and obstacle avoidance device based on the Hungarian-APF model in one embodiment;

[0032] Figure 7 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] In one embodiment, Figure 1 As shown in FIG, a method for task allocation and obstacle avoidance of a UAV swarm based on the Hungarian-APF model is provided, which includes the following steps:

[0035] Step 102: construct a task allocation model based on the UAV task allocation strategy and the cost of the UAV completing the task.

[0036] The cost is determined by the revenue gained from the drone completing the mission and the distance it travels to complete the mission.

[0037] Step 104: construct a cost matrix based on the cost of completing tasks by all drones in the drone swarm, and construct a decision matrix based on the task allocation strategies of all drones in the drone swarm.

[0038] In step 106, the decision matrix is used as the input of the UAV task allocation strategy of the task allocation model, the cost matrix is used as the input of the cost value, and the Hungarian algorithm is used to solve the task allocation model to obtain the optimal UAV task allocation result.

[0039] Step 108: Build an APF obstacle avoidance model.

[0040] The APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. Dynamic repulsion is used when the drone approaches an obstacle. It is first subjected to dynamic repulsion, and then to dynamic repulsion and static repulsion. When the static repulsion and gravity of the drone are balanced, the velocity vector of the drone is not zero.

[0041] In step 110, the task allocation result is input into the APF obstacle avoidance model to obtain the trajectory planning result of each UAV in the UAV swarm for obstacle avoidance.

[0042] In the aforementioned drone swarm task allocation and obstacle avoidance method based on the Hungarian-APF model, during the task allocation phase, a task allocation model is first constructed based on the drone task allocation strategy and the cost of completing the task (determined by the task reward and the distance to complete the task). Based on this model, a cost matrix and a decision matrix are further constructed. The cost matrix integrates the cost of completing the task for all drones, while the decision matrix reflects the task allocation strategy for all drones. Subsequently, the decision matrix and cost matrix are used as inputs for the task allocation strategy and cost value, respectively, and the Hungarian algorithm is introduced to solve the task allocation model. With its efficient computational logic, this algorithm can quickly traverse multiple possible task allocations and accurately find the optimal solution. Compared to traditional heuristic algorithms, this algorithm significantly reduces the amount of computation and effectively improves the efficiency and accuracy of task allocation. Furthermore, for trajectory planning, an APF obstacle avoidance model that incorporates static repulsion, dynamic repulsion, and gravity is constructed. The innovative design of the dynamic repulsion mechanism ensures that when approaching an obstacle, the drone is first subjected to dynamic repulsion, followed by both dynamic and static repulsion. This ensures that the drone's velocity vector is non-zero when the static repulsion and gravitational forces are balanced, thus avoiding local minima and reducing track fluctuations. The optimal task allocation result is input into the APF obstacle avoidance model, ultimately yielding a trajectory planning result for each drone's obstacle avoidance. Furthermore, by combining the efficient Hungarian algorithm for task allocation with an improved APF obstacle avoidance model for trajectory planning, this two-pronged approach not only optimizes the task allocation process but also improves the accuracy of trajectory planning, effectively enhancing the efficiency of drone swarm delivery and the accuracy of obstacle avoidance paths.

[0043] In one embodiment, the cost is determined by the revenue of the drone completing the mission and the distance the drone completes the mission, including:

[0044]

[0045] in, is the cost of the i-th UAV completing the j-th task at the initial moment of time slot t, C j is the income from completing the jth task, is the distance that the i-th UAV completes the j-th task at the initial moment of time slot t.

[0046] It is worth noting that if Figure 2 As shown, in the initial situation, all drones are concentrated in the central control station of the drone swarm, that is, Figure 2The empty circle in the figure. After receiving a delivery order, a drone's status changes to 'unavailable', indicating that it is currently executing the mission. It first flies to the order's shipping point to pick up the goods. After receiving the goods, it then flies to a pickup point connected to the order's shipping point to unload the goods. Once the drone has unloaded the goods, the order is considered fulfilled, and its status returns to 'available'. The drone continues to receive and deliver new orders until its battery runs out or there are no more orders to deliver.

[0047] In one embodiment, the task allocation model is:

[0048]

[0049] in, is the UAV task allocation strategy for the i-th UAV to receive the j-th task at the initial moment of time slot t, M is the total number of UAVs, N is the total number of tasks, is the mission time for the i-th UAV to complete the j-th mission at the initial moment of time slot t, T is the maximum flight time of a single UAV; subject to is subject to subsequent constraints.

[0050] It is worth noting that since each drone can directly continue to receive and execute new delivery tasks after completing its mission, the task allocation of the drone swarm is a multi-time slot continuous allocation problem. The expression is as follows:

[0051]

[0052]

[0053] in, It represents the distance from the i-th drone to the delivery point of the currently executed order. It represents the distance between the receiving point of the order being executed by drone i and the shipping point of order j at the initial moment of time slot t. It represents the distance from the initial position of the drone at time slot t to the delivery point of order j. represents the coordinate value of the i-th UAV at the initial moment of time slot t, It represents the coordinate value of the harvest point of the order being executed by drone i at the initial moment of time slot t, (x j ,y j ) represents the coordinates of the delivery point of order j.

[0054] From the above formula, we can see that the state of the drone determines the calculation method of the distance required for the i-th drone to receive the j-th order at the initial moment of time slot t. When the drone's state is 'available', the drone state is idle, so It is directly equal to the distance between drone i and the delivery point of task j at the initial moment of time slot t; if the drone's status is 'unavailable', it means that the drone is performing a task. This is equal to the sum of the distance a drone needs to travel to complete the current order and the distance from the current order's delivery point to the next order's delivery point. This ensures that all drones participate in each task allocation, ensuring the optimal allocation solution.

[0055] In one embodiment, a matrix consisting of the cost values of all drones in the drone swarm completing tasks is filled with zeros to obtain a cost value matrix:

[0056]

[0057] Among them, B t is the square matrix of cost values at the initial moment of time slot t, is the cost of the Nth UAV receiving the Nth task at the initial moment of time slot t, is the cost of the i-th UAV completing the j-th task at the initial moment of time slot t, M is the total number of UAVs, N is the total number of tasks; subject to is subject to subsequent constraints.

[0058] It is worth noting that when assigning tasks, the goal is to maximize the sum of the value functions of all drones. Therefore, the cost is constructed according to the following formula:

[0059]

[0060] in, M is the total number of drones, N is the total number of tasks. When the number of drones is less than the number of orders, the above matrix B is not a square matrix. The premise of the traditional Hungarian algorithm is that the cost matrix is a square matrix, and the goal is to minimize the objective function of the algorithm. Therefore, the improved Hungarian algorithm is used to solve it. First, the cost matrix B is padded with 0s to make it a square matrix.

[0061] In one embodiment, a decision matrix of all drones in the unmanned swarm is constructed based on a task allocation strategy in which each drone is assigned at most one task:

[0062]

[0063] Among them, X t is the decision matrix at the initial moment of time slot t, The task allocation strategy for the Nth drone to receive the Nth order task at the initial moment of time slot t.

[0064] It's worth noting that, since each drone can only receive a maximum of one order at a time, X represents an N*N binary decision matrix. Furthermore, each drone has a limited battery life, limiting its maximum flight range. In this article, we use T to represent the maximum flight time. Assuming the time it takes for a drone to take off, receive, and unload cargo is negligible, and the drone's motion is considered uniform, we obtain the following equation:

[0065]

[0066] T in the above formula is the maximum operating time of each drone, t t ij is the time required for drone i to complete the jth order at the initial moment of time slot t, v represents the speed of the drone, d j is the distance from the shipping point to the receiving point of the jth order. t ij It is a 0-1 variable, where 1 means that the i-th UAV receives the j-th order at the initial moment of time slot t, and 0 means that the i-th UAV does not receive the j-th order at the initial moment of time slot t.

[0067] In one embodiment, the dynamic repulsive force is:

[0068]

[0069]

[0070] in, k F exm is the dynamic repulsive force of the kth obstacle on the drone, k F ex is the static repulsion of the kth obstacle on the drone, λ is the dynamic repulsion coefficient, v is the speed of the drone, k d is the distance vector between the UAV and the kth obstacle, is the virtual velocity vector of the UAV, ρ1 is the threshold of the dynamic repulsive force, k m is the static repulsion coefficient, ρ(q, k q o ) is the distance from the UAV to the center of the kth obstacle.

[0071] It is worth noting that if Figure 3 As shown in the figure, ρ1 is the threshold of dynamic repulsion, which is non-zero only when the distance between the UAV and the obstacle is less than the threshold. is the distance vector between the UAV and the kth obstacle, with the direction from the UAV to the obstacle center and the magnitude being the Euclidean distance between the UAV and the obstacle center. β is the angle between the velocity vector and the distance vector. In addition, Figure 4 As shown in the figure, since the dynamic repulsive force range is proportional to the square of the velocity, by properly setting k m The value of can make the range of dynamic repulsion greater than the range of static repulsion. At this time, when the drone approaches an obstacle, it is first affected by dynamic repulsion and then by dynamic repulsion and static repulsion at the same time.

[0072] In one embodiment, the APF obstacle avoidance model is:

[0073]

[0074] Among them, F co is the APF obstacle avoidance model, F ex is the static repulsive force, F exm is the dynamic repulsive force, F att is the gravitational force, η is the static gravitational coefficient, ρ(q,q g ) is the distance from the target point q(x g ,y g ) points to the vector of the drone’s current time slot coordinate q(x,y), d amax The task allocation result is input into the APF obstacle avoidance model, and the coordinates of the UAV are updated according to the UAV's speed increment, the preset speed threshold, and the speed distribution ratio constraint:

[0075]

[0076]

[0077] v xm =vcosθ+k v F co cosθdt

[0078] v ym =vsinθ+k v F co sinθdt

[0079] Among them, Δx is the difference between the horizontal coordinate of the drone at the current time slot and the horizontal coordinate of the next time slot, Δy is the difference between the vertical coordinate of the drone at the current time slot and the vertical coordinate of the next time slot, v x The speed threshold is the horizontal speed constrained by the speed distribution ratio, v y is the vertical speed after the speed threshold is constrained by the speed distribution ratio, F co is the dynamic resultant force, k v is the velocity increment coefficient, v xmis the horizontal velocity after the dynamic resultant force is updated without velocity distribution proportional constraints, v ym The vertical velocity after the dynamic resultant force is updated without the velocity distribution ratio constraint. The obstacle avoidance trajectory planning result of each UAV in the UAV swarm is obtained based on the updated UAV coordinates.

[0080] It is worth noting that the dynamic repulsion vector is perpendicular to the direction of the drone's velocity vector and points in the direction away from the obstacle. When the drone is within the range of the dynamic repulsion, even if the drone's static repulsion and gravity are balanced, as long as the drone's velocity vector is not 0, the drone's dynamic repulsion will definitely not be 0, and the drone will always be able to obtain a force away from the obstacle area. By setting the speed threshold and distributing the operation in proportion, it is possible to avoid excessive speed increases that lead to sudden changes in the path, thereby preventing excessive changes in the repulsion. At the same time, by using k v F co To simulate the effect of acceleration in physics, the trajectory obtained by coordinate update is smoother.

[0081] In one embodiment, a virtual delivery point is set for tasks within a preset distance within the obstacle area where the delivery point is located. The virtual delivery point is located on the vector extension line from the delivery point to the delivery point. After the drone meets the conditions for crossing the obstacle, the delivery point is restored to the original delivery point.

[0082] It is worth noting that if Figure 5 As shown in the example, for orders whose delivery points are within a certain distance of an obstacle zone, a virtual delivery point is set. This virtual point is located along the extension of the vector from the delivery point to the delivery point. When the drone is determined to have passed the obstacle (in this context, this condition can be interpreted as having passed the line passing through the center of the obstacle zone and perpendicular to the line connecting the delivery point and the delivery point), the delivery point is restored to its original state. This method increases the gravitational pull and speeds up path planning when resolving unreachable paths, saving planning time.

[0083] In one embodiment, the pseudocode for the joint task allocation trajectory planning algorithm is shown in Table 1 below. The drone's status is first initialized to "available" and its order information to "None." Next, the server detects order information, obstacle information, and drone information. It then constructs a cost matrix and performs task planning using the modified Hungarian algorithm. Based on the planning results, the status of the drone assigned a task is set to "task." The status of previously assigned tasks is set to "disable" and removed from the task list to prevent duplicate assignments. Upon receiving a task, the drone uses the modified APF algorithm to plan an obstacle-avoiding path based on the task information and the drone's location. When a drone completes its assigned task using the planned path, its status is restored to "available," and the cost matrix is reconstructed for all drones and available orders, with the modified Hungarian algorithm used for assignments. If a drone already has a task but is assigned a new task, the new task is registered. Upon completing the existing task, the registered task is set as the current task. This cycle continues until all tasks are completed.

[0084] Table 1 Pseudo code of joint task allocation trajectory planning algorithm

[0085]

[0086]

[0087] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0088] In one embodiment, Figure 6 As shown, a UAV swarm task allocation and obstacle avoidance device based on the Hungarian-APF model is provided, comprising: a task allocation model construction module 602, a decision matrix construction module 604, an optimal task allocation module 606, an obstacle avoidance model construction module 608 and a trajectory planning module 610, wherein:

[0089] The task allocation model building module 602 is used to build a task allocation model based on the UAV task allocation strategy and the cost of the UAV completing the task. The cost is determined by the UAV's revenue from completing the task and the distance to complete the task.

[0090] The decision matrix construction module 604 is used to construct a cost matrix based on the cost of completing tasks by all drones in the drone swarm, and to construct a decision matrix based on the task allocation strategy of all drones in the drone swarm.

[0091] The optimal task allocation module 606 is used to use the decision matrix as the input of the UAV task allocation strategy of the task allocation model and the cost matrix as the input of the cost value, and use the Hungarian algorithm to solve the task allocation model to obtain the optimal UAV task allocation result.

[0092] Obstacle avoidance model construction module 608 is used to construct an APF obstacle avoidance model. The APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. Dynamic repulsion is used to first affect the drone when it approaches an obstacle, and then the dynamic repulsion and static repulsion are combined. When the static repulsion and gravity balance, the drone's velocity vector is not zero.

[0093] The trajectory planning module 610 is used to input the task allocation results into the APF obstacle avoidance model to obtain the trajectory planning results for each drone in the drone swarm to avoid obstacles.

[0094] Regarding the specific definition of the drone swarm task allocation and obstacle avoidance device based on the Hungarian-APF model, please refer to the definition of the drone swarm task allocation and obstacle avoidance method based on the Hungarian-APF model above, which will not be repeated here. Each module in the above-mentioned drone swarm task allocation and obstacle avoidance device based on the Hungarian-APF model can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0095] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 7As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for task allocation and obstacle avoidance of a drone swarm based on the Hungarian-APF model is implemented. The display screen of the computer device can be a liquid crystal display or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0096] Those skilled in the art will understand that Figure 6-7 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0097] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0098] Based on the UAV task allocation strategy and the cost of the UAV completing the task, a task allocation model is constructed. The cost is determined by the UAV's revenue from completing the task and the distance to complete the task.

[0099] According to the cost values of all drones in the drone swarm completing tasks, a cost value matrix is constructed, and according to the task allocation strategy of all drones in the drone swarm, a decision matrix is constructed.

[0100] The decision matrix is used as the input of the UAV task allocation strategy of the task allocation model, the cost matrix is used as the input of the cost value, and the Hungarian algorithm is used to solve the task allocation model to obtain the optimal UAV task allocation result.

[0101] Build an APF obstacle avoidance model. The APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. Dynamic repulsion is used when the drone approaches an obstacle. It is first subjected to dynamic repulsion, and then to dynamic repulsion and static repulsion. When the static repulsion and gravity balance, the drone's velocity vector is not zero.

[0102] The task allocation results are input into the APF obstacle avoidance model to obtain the trajectory planning result of each UAV in the UAV swarm for obstacle avoidance.

[0103] Those skilled in the art will appreciate that all or part of the processes in the above-described embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described embodiments. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0104] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0105] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. A method for task allocation and obstacle avoidance of drone swarm based on the Hungarian-APF model, characterized in that: The method comprises: Constructing a task allocation model based on the UAV task allocation strategy and the cost of the UAV completing the task; the cost is determined by the UAV's revenue from completing the task and the distance to complete the task; According to the cost values of all drones in the drone swarm completing tasks, a cost value matrix is constructed, and according to the task allocation strategy of all drones in the drone swarm, a decision matrix is constructed; The decision matrix is used as the input of the UAV task allocation strategy of the task allocation model, the cost value matrix is used as the input of the cost value, and the Hungarian algorithm is used to solve the task allocation model to obtain the optimal UAV task allocation result; Construct an APF obstacle avoidance model; the APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravitational force. The dynamic repulsion is used to firstly affect the UAV with dynamic repulsion when approaching an obstacle, and then with dynamic repulsion and static repulsion, so that when the static repulsion and gravitational force of the UAV are balanced, the UAV's velocity vector is not zero; The task allocation result is input into the APF obstacle avoidance model to obtain the trajectory planning result of each drone in the drone swarm to avoid obstacles.

2. The method according to claim 1, characterized in that The cost is determined by the revenue of the drone completing the mission and the distance the mission is completed, including: in, is the cost of the i-th UAV completing the j-th task at the initial moment of time slot t, C j is the income from completing the jth task, is the distance that the i-th UAV completes the j-th task at the initial moment of time slot t.

3. The method according to claim 1, characterized in that The task allocation model is: subject to in, is the UAV task allocation strategy for the i-th UAV to receive the j-th task at the initial moment of time slot t, M is the total number of UAVs, N is the total number of tasks, is the mission time for the i-th UAV to complete the j-th mission at the initial moment of time slot t, T is the maximum flight time of a single UAV; subject to is subject to subsequent constraints.

4. The method according to any one of claims 1 to 3, characterized in that According to the cost value of all drones in the drone swarm completing the task, a cost value matrix is constructed, including: According to the matrix composed of the cost values of all drones in the drone swarm completing the task, fill the empty values with 0 to obtain the cost value matrix: subject to Among them, B t is the square matrix of cost values at the initial moment of time slot t, is the cost of the Nth UAV receiving the Nth task at the initial moment of time slot t, is the cost of the i-th UAV completing the j-th task at the initial moment of time slot t, M is the total number of UAVs, N is the total number of tasks; subject to is subject to subsequent constraints.

5. The method according to claim 4, characterized in that According to the task allocation strategy of all drones in the drone swarm, a decision matrix is constructed, including: According to the task allocation strategy of assigning at most one task to each drone, the decision matrix of all drones in the unmanned swarm is constructed: Among them, X t is the decision matrix at the initial moment of time slot t, The task allocation strategy for the Nth drone to receive the Nth order task at the initial moment of time slot t.

6. The method according to claim 5, characterized in that The dynamic repulsive force is: ρ1=k m v 2 in, k F exm is the dynamic repulsive force of the kth obstacle on the drone, k F ex is the static repulsion of the kth obstacle on the drone, λ is the dynamic repulsion coefficient, v is the speed of the drone, k d is the distance vector between the UAV and the kth obstacle, is the virtual velocity vector of the UAV, ρ1 is the threshold of the dynamic repulsive force, k m is the static repulsion coefficient, ρ(q, k q o ) is the distance from the UAV to the center of the kth obstacle.

7. The method according to claim 6, characterized in that The APF obstacle avoidance model is: F co =F ex +F exm +F att Among them, F co is the APF obstacle avoidance model, F ex is the static repulsive force, F exm is the dynamic repulsive force, F att is the gravitational force, η is the static gravitational coefficient, ρ(q,q g ) is the distance from the target point q(x g ,y g ) points to the vector of the drone’s current time slot coordinate q(x,y), d amax is the static gravitational threshold; The task allocation result is input into the APF obstacle avoidance model to obtain the trajectory planning result of each UAV in the UAV swarm for obstacle avoidance, including: The task allocation result is input into the APF obstacle avoidance model, and the coordinates of the drone are updated according to the speed increment of the drone, the preset speed threshold and the speed distribution ratio constraint: v xm =v cosθ+k v F co cosθdt v ym =v sinθ+k v F co sinθdt Among them, Δx is the difference between the horizontal coordinate of the drone at the current time slot and the horizontal coordinate of the next time slot, Δy is the difference between the vertical coordinate of the drone at the current time slot and the vertical coordinate of the next time slot, v x The speed threshold is the horizontal speed constrained by the speed distribution ratio, v y is the vertical speed after the speed threshold is constrained by the speed distribution ratio, F co is the dynamic resultant force, k v is the velocity increment coefficient, v xm is the horizontal velocity after the dynamic resultant force is updated without velocity distribution proportional constraints, v ym It is the vertical velocity after the dynamic resultant force is updated without velocity distribution proportional constraint; Obtain the obstacle avoidance trajectory planning results for each drone in the drone swarm based on the updated drone coordinates.

8. The method according to claim 7, characterized in that A virtual delivery point is set for the task within a preset distance within the obstacle area where the delivery point is located. The virtual delivery point is located on the vector extension line from the delivery point to the delivery point. After the drone meets the conditions for crossing the obstacle, the delivery point is restored to the original delivery point.

9. A UAV swarm task allocation and obstacle avoidance device based on the Hungarian-APF model, characterized in that: The device comprises: A task allocation model construction module is used to construct a task allocation model based on the UAV task allocation strategy and the cost of the UAV completing the task; the cost is determined by the UAV's revenue from completing the task and the distance to complete the task; A decision matrix building module is used to build a cost matrix based on the cost of completing tasks by all drones in the drone swarm, and to build a decision matrix based on the task allocation strategy of all drones in the drone swarm; an optimal task allocation module, configured to use the decision matrix as an input of the UAV task allocation strategy of the task allocation model, the cost matrix as an input of the cost value, and solve the task allocation model using the Hungarian algorithm to obtain an optimal UAV task allocation result; The obstacle avoidance model construction module is used to construct an APF obstacle avoidance model; the APF obstacle avoidance model includes static repulsion, dynamic repulsion, and gravity. The dynamic repulsion is used to firstly affect the dynamic repulsion when the drone approaches an obstacle, and then affect the dynamic repulsion and static repulsion, so that when the static repulsion and gravity of the drone are balanced, the velocity vector of the drone is not zero; The trajectory planning module is used to input the task allocation result into the APF obstacle avoidance model to solve and obtain the trajectory planning result of each drone in the drone group to avoid obstacles.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images