Multi-uav cooperative task allocation and path planning method and device, equipment and medium

Abstract

This application discloses a method, apparatus, device, and medium for multi-UAV cooperative task allocation and path planning, relating to the field of UAV technology. The method includes: receiving a task instruction containing the spatial coordinates of multiple reconnaissance targets, a line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information; and acquiring environmental information including obstacle locations and weather conditions. The line-of-sight coverage scheme is a candidate coverage scheme constructed based on at least one set of multiple UAV observation positions corresponding to each reconnaissance target. A path planning model is constructed based on the task instruction and environmental information, with the goal of minimizing the total cost under the constraints of the line-of-sight coverage scheme and path planning. The path planning model is iteratively solved using an adaptive position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme. This method can collaboratively optimize position combination selection and flight path, achieving efficient large-scale path planning.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to methods, apparatus, equipment and media for multi-UAV collaborative task allocation and path planning. Background Technology

[0002] With increasing environmental complexity and the development of intelligent air defense systems, collaborative reconnaissance by multi-UAV swarms has become a key method. Compared to single-UAV operations, UAV swarms offer advantages such as wider coverage, higher fault tolerance, and better task parallelism. However, reconnaissance against camouflaged or high-value targets must meet multi-view coverage constraints (triangulation and stereo imaging), meaning a single target needs to be jointly observed by multiple UAVs in line-of-sight positions, forming a characteristic where one target corresponds to multiple feasible position combinations. Specifically, for example... Figure 1 As shown, existing UAV collaborative mission planning methods mainly suffer from two types of problems: one is that traditional methods abstract the reconnaissance target as a single fixed node, ignoring the influence of sensor field of view and terrain occlusion, and cannot optimize the selection of array positions, resulting in low solution quality or mission failure; the other is that the phased strategy first determines the optimal array position and then plans the path, which severs the strong coupling relationship between array position combination and route cost, and may cause extremely high flight costs.

[0003] Therefore, how to coordinate and optimize the selection of array positions and flight paths to achieve efficient large-scale path planning is a technical problem that needs to be solved in this field. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a method, apparatus, device, and medium for multi-UAV collaborative task allocation and path planning, which can collaboratively optimize the selection of array positions and flight paths, and achieve efficient large-scale path planning. The specific solution is as follows:

[0005] Firstly, this application discloses a method for multi-UAV cooperative task allocation and path planning, including:

[0006] The system receives a task instruction containing the spatial coordinates of multiple reconnaissance targets, a line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information, and acquires environmental information including obstacle locations and weather conditions; wherein, the line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target.

[0007] Based on the mission instructions and the environmental information, a path planning model is constructed with the goal of minimizing the total cost under the constraints of line-of-sight coverage scheme and path planning; the total cost includes the UAV deployment cost and the flight range cost.

[0008] The path planning model is iteratively solved using an adaptive array perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme.

[0009] Optionally, the step of iteratively solving the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme includes:

[0010] The large neighborhood search algorithm with adaptive array position perturbation is used to select the optimal line-of-sight coverage scheme from multiple corresponding line-of-sight coverage schemes for each reconnaissance target. Based on the optimal line-of-sight coverage scheme, the set of all visited UAV observation array positions is determined, and then the output is the initial flight path planning scheme of each UAV that meets the preset UAV condition constraints.

[0011] Multi-aircraft conflict detection is performed on the initial flight path planning scheme to obtain a cooperative flight path planning scheme that includes a set of optimized flight paths without conflicts.

[0012] Optionally, the cooperative flight path planning scheme includes an optimal line-of-sight coverage scheme determined for each reconnaissance target, an optimal UAV observation position sequence allocated to each UAV, and a flight path sequence generated based on the optimal UAV observation position sequence.

[0013] Optionally, the step of iteratively solving the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme includes:

[0014] At least one observation array access task already assigned to the UAV is removed from the current flight path planning scheme using a preset destruction operator to obtain the target observation array access task.

[0015] The target observation array access task is re-inserted into the path sequence of all UAVs in the current iteration state using a preset repair operator, so as to generate the current candidate flight path planning scheme based on the allocation position and access order that minimizes the total cost.

[0016] Based on the simulated annealing criterion, the first total cost of the current candidate flight path planning scheme is compared with the second total cost of the current flight path planning scheme, so as to update the flight path planning scheme with the lowest total cost to the optimal flight path planning scheme in the current iteration state;

[0017] If the optimal flight path planning scheme under multiple iterations does not meet the preset total cost optimization condition, then a reconnaissance target is selected based on the tabu search strategy, and the corresponding current line-of-sight coverage scheme is switched to other candidate coverage schemes. Based on the other candidate coverage schemes, the step of removing at least one observation array access task assigned to the UAV from the current flight path planning scheme using a preset destruction operator is executed.

[0018] Optionally, the multi-UAV collaborative task allocation and path planning method further includes:

[0019] During the iteration process, the control parameters are adjusted according to the current iteration state; wherein, the control parameters include the trigger probability of the array perturbation operation, the taboo period of the taboo search, and the temperature parameter in the simulated annealing criterion.

[0020] Optionally, the preset destruction operator includes any one or more of the following: random destruction operator, similar destruction operator, and drone clearing operator; the preset repair operator includes any one or two of the following: greedy repair operator and regret value repair operator.

[0021] Optionally, the line-of-sight coverage scheme constraint ensures that all reconnaissance targets are covered by the corresponding line-of-sight coverage scheme; the path planning constraints include position access consistency constraints, path flow conservation constraints, maximum range constraints, and capability matching constraints, wherein the position access consistency constraint ensures that all UAV observation positions corresponding to the line-of-sight coverage scheme are visited by at least one UAV; the path flow conservation constraint ensures that the flight paths of each UAV are continuous and that each visited UAV observation position is reached and departed by exactly one UAV; the maximum range constraint ensures that the total flight distance of each UAV does not exceed a preset range limit; and the capability matching constraint ensures that UAVs are assigned to perform observation position access tasks that match the corresponding sensor payload capabilities.

[0022] Secondly, this application discloses a multi-UAV cooperative task allocation and path planning device, comprising:

[0023] The information receiving module is used to receive mission instructions containing the spatial coordinates of multiple reconnaissance targets, the line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information, and to acquire environmental information including obstacle locations and meteorological conditions; wherein, the line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target.

[0024] The path planning model construction module is used to construct a path planning model based on the task instructions and the environmental information, with the optimization objective being to minimize the total cost under the constraints of the line-of-sight coverage scheme and path planning; the total cost includes the UAV deployment cost and the flight range cost;

[0025] The scheme output module is used to iteratively solve the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme.

[0026] Thirdly, this application discloses an electronic device, including:

[0027] Memory, used to store computer programs;

[0028] A processor is used to execute the computer program to implement the steps of the aforementioned disclosed multi-UAV cooperative task allocation and path planning method.

[0029] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed multi-UAV cooperative task allocation and path planning method.

[0030] As can be seen, this application discloses a method for receiving mission instructions containing the spatial coordinates of multiple reconnaissance targets, a line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information, and for acquiring environmental information including obstacle locations and weather conditions. The line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target. A path planning model is constructed based on the mission instructions and the environmental information, with the goal of minimizing the total cost under line-of-sight coverage scheme constraints and path planning constraints. The total cost includes UAV deployment cost and range cost. The path planning model is iteratively solved using an adaptive position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme. Therefore, by acquiring mission instructions and environmental information, the sensor field of view and terrain visibility related to the need for multiple specific array positions to collaboratively observe a target are transformed into decision variables to be optimized. In this way, it can be seen that the planning is for an entity that needs to be covered by multiple array positions. Since all candidate array position schemes are visibility schemes obtained in advance based on terrain and field of view analysis, regardless of which one is selected, successful reconnaissance of the target can be geometrically guaranteed, avoiding mission failure due to line of sight obstruction. Furthermore, the constructed path planning model is a joint optimization model. The objective function of the model minimizes the total cost while being constrained by two types of constraints. Then, the path planning model is iteratively solved using an adaptive array perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme. During the optimization process, the selected array coverage scheme can be dynamically changed to the target, and the path cost under the new scheme can be evaluated. Finally, the scheme with the best overall coverage quality and route cost is obtained, avoiding the problems of local optima and high global cost caused by phased strategies. Moreover, since the array selection and path planning are carried out simultaneously during the iteration process, the path cost information can be used to guide the array selection in real time, resulting in higher search efficiency and a converged solution that is closer to the practically executable and economically optimal path planning scheme. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0032] Figure 1 This is a schematic diagram illustrating the UAV mission allocation problem under the traditional multi-view array configuration disclosed in this application;

[0033] Figure 2 This is a flowchart of a multi-UAV collaborative task allocation and path planning method disclosed in this application;

[0034] Figure 3 This application discloses a specific method for multi-UAV collaborative task allocation and path planning.

[0035] Figure 4(a) is a schematic diagram of the output of the AP-ALNS algorithm in a small-scale scenario with randomly distributed targets as disclosed in this application;

[0036] Figure 4(b) is a schematic diagram of the output of the GA algorithm in a small-scale scenario with randomly distributed targets as disclosed in this application;

[0037] Figure 5(a) is a schematic diagram of the output results of the AP-ALNS algorithm in a small-scale scenario with mixed distributed targets as disclosed in this application;

[0038] Figure 5(b) is a schematic diagram of the output of the GA algorithm in a small-scale scenario with mixed distributed targets as disclosed in this application;

[0039] Figure 6(a) is a schematic diagram of the output results of the AP-ALNS algorithm in a small-scale scenario of clustering distribution targets disclosed in this application;

[0040] Figure 6(b) is a schematic diagram of the output of the GA algorithm in a small-scale scenario of clustering distribution targets disclosed in this application;

[0041] Figure 7 This application discloses a comparison chart of algorithm convergence curves in a large-scale scenario;

[0042] Figure 8 This is a schematic diagram of a multi-UAV collaborative task allocation and path planning device disclosed in this application;

[0043] Figure 9 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0044] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0045] With increasing environmental complexity and the development of intelligent air defense systems, collaborative reconnaissance by multi-UAV swarms has become a key method. Compared to single-UAV operations, UAV swarms offer advantages such as wider coverage, higher fault tolerance, and better task parallelism. However, reconnaissance against camouflaged or high-value targets must meet multi-view coverage constraints (triangulation and stereo imaging), meaning a single target needs to be jointly observed by multiple UAVs in line-of-sight positions, forming a characteristic where one target corresponds to multiple feasible position combinations. Specifically, for example... Figure 1 As shown, existing UAV collaborative mission planning methods mainly suffer from two types of problems: one is that traditional methods abstract the reconnaissance target as a single fixed node, ignoring the influence of sensor field of view and terrain occlusion, and cannot optimize the selection of array positions, resulting in low solution quality or mission failure; the other is that the phased strategy first determines the optimal array position and then plans the path, which severs the strong coupling relationship between array position combination and route cost, and may cause extremely high flight costs.

[0046] To this end, the present invention provides a multi-UAV collaborative task allocation and path planning scheme, which can collaboratively optimize the selection of array positions and flight paths, and achieve efficient large-scale path planning.

[0047] like Figure 2 As shown, this invention provides a method for multi-UAV collaborative task allocation and path planning, including:

[0048] Step S11: Receive a task instruction containing the spatial coordinates of multiple reconnaissance targets, a line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information, and obtain environmental information containing obstacle locations and weather conditions; wherein, the line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target.

[0049] In this embodiment, a coordinated reconnaissance mission command issued by the command system is received, and real-time environmental information is acquired simultaneously. The coordinated reconnaissance mission command includes the spatial coordinates (latitude, longitude, and elevation) of multiple reconnaissance targets (vehicles, facilities). Each target constitutes an element in the target set. For each reconnaissance target, at least one set of candidate coverage schemes is generated in advance through terrain analysis and field-of-view calculation, forming a scheme set. Each scheme is a combination of multiple (2 to 4) specific UAV observation positions. For example, reconnaissance target 1 corresponds to scheme 1, which requires simultaneous observation from positions P1 and P2; scheme 2 requires observation from positions P3 and P4. Time constraint information refers to the time window (earliest start time and latest end time) required for each reconnaissance target to be reconnaissance completed, used to meet the mission coordination rhythm or utilize specific time. Priority information assigns a priority level to each reconnaissance target, used to guide decision-making in situations of resource scarcity or conflict, ensuring that high-priority reconnaissance targets are prioritized.

[0050] Environmental information is obtained from sensing units or databases at the planning time, mainly including: obstacle locations: referring to the spatial distribution of static terrain features (mountains, buildings), used to determine no-fly zones and calculate avoidance paths. In the simulation scenario of path planning, simulated mountains are generated through a sine curve function, forming natural obstacles and line-of-sight obstructions; meteorological conditions: including real-time wind speed, wind direction, visibility, etc., which are environmental parameters for planning feasibility assessment. The generation of the line-of-sight coverage scheme is the result of the analysis of the target's spatial coordinates and the locations of environmental obstacles (terrain). Only when a specific position maintains a line of sight (unobstructed) to the reconnaissance target can that position be included in the candidate coverage schemes for covering the reconnaissance target. Feasible mission schemes (line-of-sight coverage schemes) are defined using environmental information. Based on the received information, the input of a collaborative reconnaissance planning problem with multiple feasible solution spaces under complex physical constraints can be obtained.

[0051] Step S12: Based on the task instructions and the environmental information, construct a path planning model with the goal of minimizing the total cost under the constraints of the line-of-sight coverage scheme and path planning; the total cost includes the UAV deployment cost and the flight cost.

[0052] In this embodiment, the line-of-sight coverage scheme constraint ensures that all reconnaissance targets are covered by the corresponding line-of-sight coverage scheme; the path planning constraints include array access consistency constraints, path flow conservation constraints, maximum range constraints, and capability matching constraints. Specifically, the array access consistency constraint ensures that all UAV observation arrays corresponding to the line-of-sight coverage scheme are visited by at least one UAV; the path flow conservation constraint ensures that the flight paths of each UAV are continuous and that only one UAV arrives at and departs from each visited UAV observation array; the maximum range constraint ensures that the total flight distance of each UAV does not exceed a preset range limit; and the capability matching constraint ensures that UAVs are assigned to perform observation array access tasks with corresponding sensor payload capability matching. Based on the above line-of-sight coverage scheme constraints and path planning constraints, and according to task instructions and environmental information, a path planning model is constructed with minimizing the total cost as the optimization objective. Specifically, the total cost function of the path planning model is:

[0053] ;

[0054] ;

[0055] in, This represents the total system cost, which includes the cost of deploying the drone and the cost of its flight path. Drones that do not deploy (with a path length of 0) do not incur fixed costs. For drones Total flight distance consumed (including flight and hovering). As the dispatch indicator variable, Indicates if drone If the flight travels from node m to node n, the value is 1; otherwise, it is 0. Represented as drone Fixed deployment costs Represented as drone The cost per unit distance of flight Indicates a collection of drones. This represents the set of all nodes (including airports and positions).

[0056] Furthermore, the line-of-sight coverage scheme constraint requires that for each reconnaissance target, exactly one scheme must be selected from its set of available coverage schemes. That is, a combination of positions that can fully cover the reconnaissance target must be selected, and the corresponding multiple positions must be added to the list of positions to be visited by the UAV, specifically expressed as follows:

[0057] ;

[0058] in, For 0-1 variables, if the reconnaissance target Selected formation combination scheme If it is 1, then it is 1; otherwise it is 0. Indicate reconnaissance target The set of feasible array combination schemes, where T is the set of reconnaissance targets.

[0059] The consistency constraint for array access is that once the target... Selected Plan Then all the array positions included in this scheme All nodes must be accessed, therefore the array access consistency constraint is expressed as follows:

[0060] ;

[0061] in, Represents a 0-1 variable; if position j is assigned to the UAV... If executed, the value is 1; otherwise, it is 0. This represents the set of candidate positions.

[0062] The path flow conservation constraint ensures the continuity of the UAV path, and that exactly one UAV arrives at and departs from each assigned task position, specifically expressed as:

[0063] ;

[0064] in, Represents the set of candidate positions. Indicates drone Fly from node m to node j Indicates drone Fly to node n via node j.

[0065] The maximum range constraint is the sum of the actual flight distance of the UAV and the equivalent hovering distance generated during the mission, which cannot exceed its physical maximum range, specifically expressed as:

[0066] ;

[0067] in, Indicates drone Maximum range Indicates drone Execute reconnaissance targets The equivalent hovering distance during a mission (determined by the mission type and the capabilities of the drone). This represents the Euclidean distance from node m to node n.

[0068] Capability matching constraints require specific payloads for reconnaissance of each target, and these payload requirements are mapped to positions within the corresponding coverage array combination for the target. Therefore, UAVs can only access positions with the appropriate sensor capabilities, as detailed below:

[0069] ;

[0070] in, Represents a binary parameter, if the drone Have execution objectives The ability is 1 if it is positive, otherwise it is 0.

[0071] In addition, there are variable constraints, meaning all variables are binary variables, and their values ​​can only be 0 or 1, as specifically stated below:

[0072] .

[0073] Step S13: Iteratively solve the path planning model using the large neighborhood search algorithm with adaptive array position perturbation to obtain a cooperative flight path planning scheme.

[0074] In this embodiment, an adaptive array perturbation large neighborhood search algorithm is used to select the optimal line-of-sight coverage scheme from multiple corresponding line-of-sight coverage schemes for each reconnaissance target. Based on the optimal line-of-sight coverage scheme, a set of all visited UAV observation array positions is determined, and then an initial flight path planning scheme that satisfies preset UAV condition constraints is output. Multi-aircraft conflict detection is performed on the initial flight path planning scheme to obtain a cooperative flight path planning scheme that includes a set of conflict-free optimized flight paths. The cooperative flight path planning scheme includes the optimal line-of-sight coverage scheme determined for each reconnaissance target, the optimal UAV observation array position sequence allocated to each UAV, and the flight path sequence generated based on the optimal UAV observation array position sequence. Understandably, AP-ALNS (ArrayPerturbation Adaptive Large Neighborhood Search) is used to find a preliminary solution to the entire reconnaissance mission problem through an optimization mechanism. This preliminary flight path planning scheme includes a target coverage scheme, a set of observation positions to be visited, and an initial flight path. Specifically, the target coverage scheme selects the optimal line-of-sight coverage scheme from multiple candidate schemes for each reconnaissance target. The set of observation positions to be visited summarizes all selected optimal line-of-sight coverage schemes, resulting in a set of observation positions that all UAVs need to visit together. Based on the above set of observation positions, and under the premise of satisfying the UAV's maximum range, capability matching, and other preset UAV constraints, the preliminary flight path plans a flight path for each UAV that starts from the takeoff airport, visits the assigned observation positions in an orderly manner, and finally returns. However, the initial flight path planning scheme described above mainly addresses the allocation and coarse-grained path problem of which positions to cover and who will access them, without considering the potential spatiotemporal conflicts that may occur when multiple UAVs simultaneously perform tasks in a shared airspace. Therefore, conflict detection is implemented. Specifically, the flight paths of all UAVs in the initial scheme are analyzed to detect whether two or more UAVs appear in the airspace within a safe interval at the same or similar time, or whether there are time window conflicts at path intersections. When a conflict is detected, the paths of the relevant UAVs are locally adjusted based on the right-of-way according to task priority. This involves time-series fine-tuning of a UAV's waypoints (brief hovering and waiting) or triggering small-scale local path replanning to ensure that all paths are completely free from spatiotemporal conflicts.

[0075] In this embodiment, the AP-ALNS algorithm is designed, specifically employing a master-slave two-layer architecture. The inner ALNS subprocess is responsible for path optimization, while the outer adaptive position control is responsible for scheme adjustment. The initial solution of the algorithm is generated through greedy allocation and the shortest path method, and key parameters such as the maximum number of iterations, initial temperature, and perturbation threshold are set. The neighborhood search operator is implemented as follows: preset destruction operators include RandomDestroy (randomly removing tasks), Shaw Destroy (based on spatiotemporal similarity), and UAVDestroy (clearing single-machine tasks); preset repair operators include Greedy Repair (minimizing cost increment) and Regret-k Repair (avoiding short-sightedness). The operator weights are adaptively adjusted according to historical performance, and operators that generate the globally optimal solution are given high rewards. The adaptive position perturbation mechanism is implemented as follows: when the number of consecutive times without improvement exceeds the threshold Θ, non-taboo targets are screened, a target set is selected according to the perturbation ratio, their position combinations are switched to alternative schemes, tasks are reassigned, and unchanged task relationships are inherited. A two-tiered tabu system is implemented based on the perturbation effect: effective perturbations have long-term tabu periods, while ineffective perturbations have short-term tabu periods, preventing search loops. The dynamic balancing strategy involves linearly adjusting parameters as the iteration progresses. The perturbation ratio decreases linearly from its initial value (small-scale: from 0.40 to 0.15; large-scale: from 0.20 to 0.08), while the tabu period increases linearly from its initial value (small-scale: from 10 to 40; large-scale: from 20 to 80). The simulated annealing temperature decays with a cooling rate α, and a reheating operation (β>1) is performed after array perturbation to increase the probability of accepting inferior solutions.

[0076] like Figure 3As shown, after initializing the task instructions, environmental information, and fixed parameters of the model, at least one observation array access task already assigned to the UAV is removed from the current flight path planning scheme using a preset destruction operator to obtain the target observation array access task. The target observation array access task is then reinserted into the path sequence of all UAVs in the current iteration state using a preset repair operator to generate the current candidate flight path planning scheme based on the allocation position and access order that minimizes the total cost. The first total cost of the current candidate flight path planning scheme is compared with the second total cost of the current flight path planning scheme based on the simulated annealing criterion, and the flight path planning scheme with the lowest total cost is updated as the optimal flight path planning scheme in the current iteration state. If the optimal flight path planning scheme under multiple consecutive iterations does not meet the preset total cost optimization condition, a reconnaissance target is selected based on a tabu search strategy, and the corresponding current line-of-sight coverage scheme is switched to another candidate coverage scheme. Based on the other candidate coverage scheme, the step of removing at least one observation array access task already assigned to the UAV from the current flight path planning scheme using the preset destruction operator is executed. The preset destruction operator includes any one or more of the following: random destruction operator, similar destruction operator, and drone clearing operator; the preset repair operator includes any one or two of the following: greedy repair operator and regret value repair operator.

[0077] It should be noted that the control parameters are adjusted according to the current iteration state during the iteration process; the control parameters include the trigger probability of the array perturbation operation, the taboo period of the taboo search, and the temperature parameter in the simulated annealing criterion.

[0078] It can be seen that this application discloses receiving a task instruction including the spatial coordinates of multiple reconnaissance targets, a visibility coverage plan for each reconnaissance target, time constraint information, and reconnaissance target priority information, and obtaining environmental information including obstacle positions and meteorological conditions; wherein, the visibility coverage plan is at least one candidate coverage plan based on multiple UAV observation positions corresponding to each reconnaissance target; constructing a path planning model with the minimization of the total cost under the constraints of the visibility coverage plan and path planning constraints as the optimization objective based on the task instruction and the environmental information; the total cost includes UAV deployment cost and voyage cost; using a large neighborhood search algorithm with adaptive position perturbation to iteratively solve the path planning model to obtain a cooperative flight path planning plan. Thus, by obtaining the task instruction and environmental information, the sensor fields of view and terrain visibility related to the need for multiple specific positions to cooperate in observing one target are transformed into decision variables to be optimized. In this way, it can be seen that it is for entities that need to be covered by multiple positions. Since all candidate position plans are pre-obtained visibility plans based on terrain and field of view analysis, no matter which one is selected, it can geometrically ensure the successful reconnaissance of the target and avoid mission failure caused by line-of-sight occlusion. Further, the constructed path planning model is a joint optimization model. The objective function of the model minimizes the total cost while being restricted by two types of constraints. Then, using a large neighborhood search algorithm with adaptive position perturbation to iteratively solve the path planning model to obtain a cooperative flight path planning plan can dynamically change the position coverage plan selected for the target during the optimization process and evaluate the path cost under the new plan. The finally obtained plan is the one with the overall optimal coverage quality and route cost, avoiding the problems of local optimality and high global cost caused by the phased strategy. And during the iterative process, since position selection and path planning are carried out simultaneously, it can use the path cost information for real-time feedback to guide position selection, with higher search efficiency and the solution converging to a path planning plan that is closer to the actually executable and economically optimal one.

[0079] Experimental simulation analysis and verification are carried out on the path planning result of the present invention. First, in order to simulate a real border reconnaissance environment, the experimental scenario uses a sine superposition function to simulate the meandering mountain curve, and the combat area is dynamically divided into three functional areas: Target Zone: Located below the mountain curve on our side (y < f(x)), simulating the hinterland where enemy hidden targets are distributed. Position Zone: Located between the mountain curve and a certain buffer distance (0.2×MapSize) above it, simulating the ridges or highlands suitable for deploying reconnaissance positions. Airport Zone: Located above the Position Zone, simulating the take-off and landing base in the rear. The mountain curve function is defined as, where both the amplitude A and the phase ϕ are introduced with random perturbations to increase the unstructured characteristics of the terrain.

[0080] Entity generation: Two sets of calculation examples were set up: a small-scale (250×250 km², 10 reconnaissance targets, 5 UAVs, 11 positions) and a large-scale (600×600 km², 50 reconnaissance targets, 15 UAVs, ≈60 positions). The target distribution included three types: uniform, clustered, and mixed. The heterogeneous UAVs were divided into three categories: multi-rotor, fixed-wing, and reconnaissance-strike integrated. The performance parameters are shown in Table 1.

[0081] Table 1 Performance parameters of heterogeneous UAVs

[0082]

[0083] The experimental environment was as follows: hardware configuration: Core (TM) i7-13700KF processor, 32G memory; software environment: Python language; algorithm parameter settings are shown in Table 2, where the maximum number of iterations is 300 for small-scale scenarios and 800 for large-scale scenarios, the simulated annealing initial temperature ratio is 0.1 (small-scale) / 0.06 (large-scale), and the cooling rate is 0.985 (small-scale) / 0.992 (large-scale).

[0084] Table 2 Algorithm Parameter Settings

[0085]

[0086] The effectiveness verification in a small-scale scenario is as follows: Three cases are generated, with a reconnaissance scenario size of 200km×200km. Each case contains 10 targets and 3 airports. The positions are randomly generated, and the target distribution types are uniform distribution, clustered distribution, and mixed distribution, respectively. The task allocation path diagrams and task allocation Gantt charts for three small-scale scenarios obtained by AP-ALNS and GA algorithms are shown in Figures 4(a) and 4(b), respectively. These figures represent the results of the AP-ALNS algorithm output in the first example (randomly distributed targets) and the results of the GA algorithm output in the same example. T0 to T10 represent the observation tasks of the reconnaissance targets. When a task block is marked as Ti (i∈[0,10]), it means that the UAV is performing reconnaissance of target Ti. Since each target needs to be observed from multiple line-of-sight positions, one Ti may correspond to multiple task blocks, which are accessed by different UAVs (or the same UAV at different times) to form different positions that constitute the target coverage scheme. For example, if the coverage scheme of target T1 selects positions {P1, P2}, then in the Gantt chart, there may be two task blocks marked as T1, which are executed by two different UAVs (or the same UAV at different time periods), representing that they each access positions P1 and P2 to jointly complete the reconnaissance of T1. A2_U2 represents U2 taking off or landing at airport A2, A2_U1 represents U1 taking off or landing at airport A2, and A3_U1 represents U1 taking off or landing at airport A3. Figures 5(a) and 5(b) show the output results of the AP-ALNS algorithm in a small-scale scenario under the second example (mixed distribution target) and the GA algorithm in the same example, respectively. A1_U1 represents U1 taking off or landing at airport A1. Figures 6(a) and 6(b) show the output results of the AP-ALNS algorithm in a small-scale scenario under the third example (clustering target) and the GA algorithm in the same example, respectively. Therefore, all tasks were successfully assigned and the multi-view coverage constraint was satisfied. Furthermore, the AP-ALNS algorithm reduced the cost by 46.75% compared to the GA algorithm in small-scale scenarios.

[0087] To verify the performance advantages of the AP-ALNS algorithm in handling highly complex problems with multiple constraints, a large-scale simulation example involving 10 airports, 50 target points, 15 heterogeneous UAVs, and approximately 60 selectable array positions was selected for comparative experiments. The experiments compared the proposed AP-ALNS algorithm with the classic Genetic Algorithm (GA) under the same computational resource constraints, focusing on four dimensions: optimization quality, convergence characteristics, robustness, and computational efficiency.

[0088] In large-scale scenarios, the number of possible array combination schemes increases exponentially with the increase in the number of targets, posing a severe challenge to the global optimization capability of the algorithm. Table 3 lists the statistical results of the two algorithms in 30 independent runs.

[0089] Table 3 Results of Large-Scale Case Study

[0090]

[0091] The experimental results show that the AP-ALNS algorithm significantly outperforms the GA algorithm in all metrics:

[0092] Solution quality: The average total cost of the system obtained by AP-ALNS is significantly lower than that of GA, and the average improvement rate reaches 40%. This shows that in the complex coupled space of "position selection-route planning", AP-ALNS can more effectively find high-quality solutions that balance coverage quality and flight cost.

[0093] Optimization depth: GA is often limited by the loss of population diversity, making it difficult to finely adjust the path under specific array combinations in the later stage; AP-ALNS, through its unique greedy distance repair and Regret-k repair operator, can perform deep optimization of local paths, thereby discovering lower-cost flight schemes.

[0094] To delve deeper into the dynamic optimization process of the algorithm, Figure 7 The paper presents a comparison of the convergence curves of the two algorithms under typical large-scale tests, as well as box plots of the final cost distribution of the two algorithms after 30 independent runs.

[0095] By analyzing the convergence curves, we can observe the distinctly different search behaviors of the two algorithms:

[0096] GA (Green Curve): The cost decreases rapidly in the early stages of iteration (first 100-200 generations), but then quickly levels off, exhibiting a clear "premature" convergence characteristic. This is because in large-scale discrete spaces, the crossover and mutation of genetic operators are difficult to effectively break the local extremum traps of matrix combinations.

[0097] AP-ALNS (blue curve): The curve exhibits a unique "step-like descent" characteristic. In the early stages of iteration, it rapidly decreases using the acceptance criterion of simulated annealing; when the search enters a stagnation period, the curve shows several distinct secondary descents. This is mainly attributed to the adaptive position perturbation mechanism and temperature reheating strategy introduced in the algorithm. When the algorithm detects that it is trapped in a local optimum, it forcibly changes the target position mapping scheme and instantly increases the system temperature, giving the algorithm the ability to escape deep local optima and explore new solution spaces.

[0098] The box plots reveal the following: Distribution range: The box position of AP-ALNS is significantly lower than that of GA, and they do not overlap, indicating that its advantage is statistically significant. Fluctuation range: AP-ALNS has a narrower box height (interquartile range) and fewer outliers. This is attributed to a dynamic balancing strategy, which reduces the proportion of perturbations in the later stages of iteration. and extending the taboo period This effectively locks in high-quality solution structures, reduces performance fluctuations caused by randomness, and demonstrates the strong robustness of the algorithm in handling large-scale random terrain and unstructured tasks.

[0099] In a scenario where a large bridge structure requires regular full-structure health inspections, traditional manual inspections are time-consuming, labor-intensive, and pose safety risks. Furthermore, certain areas (tower tops, cables, bridge decks) are difficult to inspect up close. This solution proposes using a cluster of multiple heterogeneous drones to perform collaborative inspections.

[0100] Inspection targets: Multiple reconnaissance locations on the bridge are defined as the reconnaissance target set T, such as: the top of main tower 1, the top of main tower 2, the anchor points of the 3rd to 6th groups of stay cables, the bottom surface of the central section of the bridge deck, etc., totaling 10 reconnaissance targets. Each reconnaissance target requires high-resolution images or 3D point cloud data from multiple angles (front, side, and bottom views) to complete the full damage identification and condition assessment. Therefore, a single reconnaissance target (one anchor point) corresponds to multiple necessary observation positions (array positions). Thus, the path planning process for bridge inspection using this scheme is as follows:

[0101] Mission instructions received: The reconnaissance targets are the aforementioned 10 bridge sections, each with its corresponding three-dimensional coordinates.

[0102] Line-of-sight coverage scheme: For each section of the bridge, multiple feasible combinations of shooting points were pre-planned. For example, for the anchor point of cable-stayed bridge No. 3, feasible schemes include: Scheme A = {Observation point P5 (side), Observation point P6 (below)}, Scheme B = {Observation point P7 (side-front), Observation point P8 (directly below)}. Each observation point is a line-of-sight position where a drone can safely hover and take pictures.

[0103] Time constraints: The entire inspection must be completed within 2 hours during off-peak traffic hours; at the same time, the top of the No. 2 main tower, which is suspected of having problems, should be prioritized for inspection within the first hour.

[0104] Priority information: Set the top of the main tower and the bottom of the bridge as high-priority targets.

[0105] Environmental Information Acquisition: Obstacle Location: The bridge's own structure, surrounding buildings, high-voltage lines, and coordinates of no-fly zones constitute the main obstacle information. Meteorological Conditions: Real-time wind speed and direction data.

[0106] Based on the above inputs, a collaborative inspection path planning model is constructed to minimize total power consumption and time cost. The model must satisfy the following requirements: select a complete combination of shooting points (coverage scheme) for each key location; plan a path for each drone while meeting its endurance and payload (sensor type) constraints. The adaptive array perturbation large neighborhood search (AP-ALNS) algorithm is used for solving the problem. During optimization, the algorithm dynamically attempts to switch shooting point combinations for different targets (switching anchor points from scheme A to scheme B) and simultaneously recalculates and optimizes the flight paths of all drones to seek the globally optimal combination selection and path arrangement scheme. Conflict detection is performed on the solved preliminary paths, for example, to avoid two drones simultaneously arriving at the same side of the bridge tower's airspace. By fine-tuning the time or path, a final safe collaborative flight plan is generated. The system ultimately outputs a bridge collaborative inspection scheme, which includes which specific combination of shooting points is used for each bridge location, the specific task sequence for each drone (drone 1 equipped with a high-definition camera, drone 2 equipped with a lidar), i.e., which shooting points are visited in sequence, the flight path of each drone, and the estimated arrival and operation time for each shooting point. The overall operation time of the cluster is expected to be 1 hour and 45 minutes, with the shortest total flight distance, and it completely avoids no-fly zones and weather risks.

[0107] like Figure 8 As shown, the present invention also discloses a multi-UAV collaborative task allocation and path planning device, comprising:

[0108] The information receiving module 11 is used to receive a task instruction containing the spatial coordinates of multiple reconnaissance targets, the line-of-sight coverage scheme for each reconnaissance target, time constraint information and reconnaissance target priority information, and to acquire environmental information containing obstacle locations and meteorological conditions; wherein, the line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target.

[0109] The path planning model construction module 12 is used to construct a path planning model based on the task instructions and the environmental information, with the optimization objective of minimizing the total cost under the constraints of the line-of-sight coverage scheme and path planning; the total cost includes the UAV deployment cost and the flight cost;

[0110] The scheme output module 13 is used to iteratively solve the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme.

[0111] Therefore, incorporating multiple target coverage schemes into the optimization framework enables joint optimization of coverage quality and route cost, avoiding the limitations of traditional phased strategies. The structural optimization search strategy for target-multiple array mapping improves the efficiency of discrete space optimization and reduces the total system cost.

[0112] Furthermore, embodiments of this application also disclose an electronic device, Figure 9 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0113] Figure 9 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the multi-UAV cooperative task allocation and path planning method disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be a computer.

[0114] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0115] The processor 21 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 21 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 21 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 21 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 21 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0116] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0117] The operating system 221 manages and controls the various hardware devices and computer programs 222 on the electronic device 20 to enable the processor 21 to perform calculations and processing on the massive amounts of data 223 in the memory 22. It can be Windows Server, Netware, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the multi-UAV collaborative task allocation and path planning method disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the electronic device from external devices, as well as data collected by its own input / output interface 25.

[0118] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned multi-UAV cooperative task allocation and path planning method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0119] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0120] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly in hardware, software modules executed by a processor, or a combination of both. The software module may be located in random access memory (RAM), memory, read-only memory (ROM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, removable disks, CD-ROMs (Compact Disc-Read Only Memory), or any other form of storage medium known in the art.

[0121] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0122] The solution provided by the present invention has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for multi-UAV collaborative task allocation and path planning, characterized in that, include: The system receives a task instruction containing the spatial coordinates of multiple reconnaissance targets, a line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information, and acquires environmental information including obstacle locations and weather conditions; wherein, the line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target. A path planning model is constructed based on the task instructions and environmental information, with the goal of minimizing the total cost under the constraints of line-of-sight coverage scheme and path planning. The total cost includes UAV deployment cost and range cost. The line-of-sight coverage scheme constraint ensures that all reconnaissance targets are covered by the corresponding line-of-sight coverage scheme. The path planning constraints include array access consistency constraint, path flow conservation constraint, maximum range constraint, and capability matching constraint. Specifically, the array access consistency constraint ensures that all UAV observation arrays corresponding to the line-of-sight coverage scheme are visited by at least one UAV; the path flow conservation constraint ensures that the flight paths of each UAV are continuous and that each visited UAV observation array is reached and departed by exactly one UAV; the maximum range constraint ensures that the total flight distance of each UAV does not exceed a preset range limit; and the capability matching constraint ensures that UAVs are assigned to perform observation array access tasks with corresponding sensor payload capability matching. The path planning model is iteratively solved using an adaptive array perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme.

2. The multi-UAV cooperative task allocation and path planning method according to claim 1, characterized in that, The method of iteratively solving the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme includes: The large neighborhood search algorithm with adaptive array position perturbation is used to select the optimal line-of-sight coverage scheme from multiple corresponding line-of-sight coverage schemes for each reconnaissance target. Based on the optimal line-of-sight coverage scheme, the set of all visited UAV observation array positions is determined, and then the output is the initial flight path planning scheme of each UAV that meets the preset UAV condition constraints. Multi-aircraft conflict detection is performed on the initial flight path planning scheme to obtain a cooperative flight path planning scheme that includes a set of optimized flight paths without conflicts.

3. The multi-UAV collaborative task allocation and path planning method according to claim 2, characterized in that, The cooperative flight path planning scheme includes an optimal line-of-sight coverage scheme determined for each reconnaissance target, an optimal UAV observation position sequence allocated to each UAV, and a flight path sequence generated based on the optimal UAV observation position sequence.

4. The multi-UAV cooperative task allocation and path planning method according to claim 1, characterized in that, The method of iteratively solving the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme includes: At least one observation array access task already assigned to the UAV is removed from the current flight path planning scheme using a preset destruction operator to obtain the target observation array access task. The target observation array access task is re-inserted into the path sequence of all UAVs in the current iteration state using a preset repair operator, so as to generate the current candidate flight path planning scheme based on the allocation position and access order with the minimum total cost. Based on the simulated annealing criterion, the first total cost of the current candidate flight path planning scheme is compared with the second total cost of the current flight path planning scheme, so as to update the flight path planning scheme with the lowest total cost to the optimal flight path planning scheme in the current iteration state; If the optimal flight path planning scheme under multiple iterations does not meet the preset total cost optimization condition, then a reconnaissance target is selected based on the tabu search strategy, and the corresponding current line-of-sight coverage scheme is switched to other candidate coverage schemes. Based on the other candidate coverage schemes, the step of removing at least one observation array access task assigned to the UAV from the current flight path planning scheme using a preset destruction operator is executed.

5. The multi-UAV cooperative task allocation and path planning method according to claim 4, characterized in that, Also includes: During the iteration process, the control parameters are adjusted according to the current iteration state; wherein, the control parameters include the trigger probability of the array perturbation operation, the taboo period of the taboo search, and the temperature parameter in the simulated annealing criterion.

6. The multi-UAV cooperative task allocation and path planning method according to claim 4, characterized in that, The preset destruction operator includes any one or more of the following: random destruction operator, similar destruction operator, and drone clearing operator; the preset repair operator includes any one or two of the following: greedy repair operator and regret value repair operator.

7. A multi-UAV collaborative task allocation and path planning device, characterized in that, include: The information receiving module is used to receive mission instructions containing the spatial coordinates of multiple reconnaissance targets, the line-of-sight coverage scheme for each reconnaissance target, time constraint information, and reconnaissance target priority information, and to acquire environmental information including obstacle locations and meteorological conditions; wherein, the line-of-sight coverage scheme is a candidate coverage scheme constructed based on multiple UAV observation positions, corresponding to at least one set for each reconnaissance target. The path planning model construction module is used to construct a path planning model based on the task instructions and environmental information, with the optimization objective being to minimize the total cost under the constraints of the line-of-sight coverage scheme and path planning. The total cost includes UAV deployment cost and range cost. The line-of-sight coverage scheme constraint ensures that all reconnaissance targets are covered by the corresponding line-of-sight coverage scheme. The path planning constraints include position access consistency constraints, path flow conservation constraints, maximum range constraints, and capability matching constraints. Specifically, the position access consistency constraint ensures that all UAV observation positions corresponding to the line-of-sight coverage scheme are visited by at least one UAV; the path flow conservation constraint ensures that the flight paths of each UAV are continuous and that each visited UAV observation position is reached and departed by exactly one UAV; the maximum range constraint ensures that the total flight distance of each UAV does not exceed a preset range limit; and the capability matching constraint ensures that UAVs are assigned to perform observation position access tasks with corresponding sensor payload capabilities matched. The scheme output module is used to iteratively solve the path planning model using an adaptive array position perturbation large neighborhood search algorithm to obtain a cooperative flight path planning scheme.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the multi-UAV cooperative task allocation and path planning method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when the computer programs are executed by a processor, they implement the steps of the multi-UAV cooperative task allocation and path planning method as described in any one of claims 1 to 6.

Images