Collaborative path planning method and system for vehicle-assisted multi-UAV mission execution

Abstract

The present invention provides a method and system for collaborative path planning for multiple unmanned aerial vehicles (UAVs) using a vehicle-assisted task. This method relates to the technical field of vehicle-assisted path planning for multiple unmanned aerial vehicles (UAVs). The method utilizes a set covering model to plan vehicle paths based on all task points. Then, based on these vehicle paths, a simulated annealing algorithm with a hybrid neighborhood search is used to solve the collaborative path planning for multiple unmanned aerial vehicles (UAVs) using the vehicle-assisted task. Compared to existing technologies, the present invention achieves higher solution efficiency and quality when solving collaborative path planning for multiple unmanned aerial vehicles using the vehicle-assisted task.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle-assisted multi-UAV path planning, and in particular to a collaborative path planning method and system for vehicle-assisted multi-UAV mission execution. Background Art

[0002] Due to the limited battery capacity of drones, it's difficult for them to complete missions within a large area within their flight time. Therefore, vehicles can be used as mobile base stations to assist drones in their movement, thereby expanding the range of their missions. Specifically, the vehicle drives on the road network while the drone flies in the air. The vehicle launches the drone at a suitable location on the road network. After the drone completes its nearby mission, the vehicle retrieves it at the appropriate location. This process repeats until all missions within the target area are completed, and the vehicle finally returns to the warehouse with all the drones. If the paths used by vehicles to assist multiple drones in performing monitoring missions can be scientifically and rationally planned, it will greatly facilitate the efficient execution of related monitoring tasks in military and civilian fields, such as intelligence reconnaissance and power inspections, and its significance is significant.

[0003] Currently, a commonly used method for path planning in vehicle-assisted multi-UAV inspection missions is the VAMU algorithm (joint routing and scheduling algorithm for Vehicle-Assisted Multi-UAV inspection). The VAMU algorithm solves the problem through three steps: solving vehicle paths, planning UAV paths, and locally optimizing the UAV paths.

[0004] However, although the VAMU algorithm can obtain a feasible solution, it still has problems with low solution quality and efficiency. Summary of the Invention

[0005] (1) Technical problems solved

[0006] In response to the shortcomings of the existing technology, the present invention provides a collaborative path planning method and system for vehicle-assisted multi-UAV mission execution, which solves the problems of low solution quality and efficiency in the existing technology when facing the collaborative path planning problem of vehicle-assisted multi-UAV mission execution.

[0007] (2) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] In a first aspect, the present invention first proposes a collaborative path planning method for a vehicle to assist multiple UAVs in performing a task, the method comprising:

[0010] Vehicle path planning is performed based on all mission points and using the set coverage model;

[0011] Based on the vehicle path, a simulated annealing algorithm with hybrid neighborhood search is used to solve the collaborative path planning solution for vehicle-assisted multi-UAV missions.

[0012] Preferably, the vehicle path planning based on all task points and using a set coverage model includes:

[0013] S11. Determine the mission points that can be covered on each road section in the road network based on the flight radius of the drone;

[0014] S12. Acquire a key road that can cover all the task points based on a set coverage model;

[0015] S13. Starting from the starting point, all the key roads are connected based on the principle of preferentially connecting the key road with the shortest distance to the current position to form a vehicle path.

[0016] Preferably, the method of solving the collaborative path planning of vehicle-assisted multi-UAV missions based on the vehicle path and using a simulated annealing algorithm with a hybrid neighborhood search includes:

[0017] S21, taking the projection points of all the task points on the vehicle path as candidate vehicle stop points, and sorting the task points according to the order of the candidate vehicle stop points on the vehicle path;

[0018] S22. Using the task point sorting result as the initial solution of the simulated annealing algorithm based on the hybrid neighborhood search, and using the simulated annealing algorithm based on the hybrid neighborhood search to solve the solution of the collaborative path planning of the vehicle-assisted multi-UAV mission.

[0019] Preferably, the S22 includes:

[0020] S221, setting and initializing relevant parameters of the simulated annealing algorithm, and using the task point sorting result as the initial solution input of the simulated annealing algorithm, and calculating the objective function value;

[0021] S222. Perform neighborhood transformation on the current solution w according to the preset neighborhood selection rule and the preset neighborhood transformation rule to generate a new solution w'. If the new solution w' has not been stored, calculate the objective function value f(w') of the new solution w'. Otherwise, continue to perform neighborhood transformation according to the preset neighborhood selection rule and the preset neighborhood transformation rule until a new unstored solution w' is generated, and calculate the objective function value f(w').

[0022] S223. Compare f(w') and f(w). If f(w') ≤ f(w), accept the new solution w'. Otherwise, determine whether to accept the new solution w' according to the Metropolis criterion. If the new solution w' is accepted, store the new solution w'.

[0023] S224, repeating steps S222 and S223 at the current temperature T, and saving the best solution corresponding to the objective function value iterated in this state; and executing step S225 until the current number of iterations reaches the preset maximum number of iterations;

[0024] S225, determine whether the termination condition of the algorithm execution is met, if so, execute S227; otherwise, go to S226;

[0025] S226, the temperature T gradually decreases, the score of each neighborhood transformation rule and the weight of each neighborhood transformation rule are updated, and S222 is executed;

[0026] S227: The algorithm is executed and the current solution is output as the optimal solution.

[0027] Preferably, the preset neighborhood selection rules include:

[0028] updating the weight of each neighborhood transformation rule, and then adopting roulette and selecting a neighborhood transformation rule according to the weight of each neighborhood transformation rule;

[0029] The weight of each neighborhood transformation rule is updated as follows:

[0030]

[0031] Among them, w' i represents the weight of the updated i-th neighborhood transformation rule; w i represents the initial weight of the i-th neighborhood transformation rule; s i represents the score of the i-th neighborhood transformation rule in the same state; n represents the number of times the i-th neighborhood transformation rule is used in the same state; ρ∈(0,1) is the coefficient for recalculating the weight;

[0032] Preferably, the score s of each neighborhood transformation rule is i The update principle is:

[0033] If the new solution is accepted and is better than the current optimal solution, the score is +σ1; if it is a new solution accepted according to the Metropolis criterion, the score is +σ2; otherwise, the score is +σ3;

[0034] Preferably, the preset neighborhood transformation rules include:

[0035] Two-point exchange neighborhood transformation rule: randomly select two points in the coding sequence, exchange the positions of the two points, and generate a new coding sequence;

[0036] Two-pair point exchange neighborhood transformation rule: randomly select two groups of adjacent points in the coding sequence, exchange the positions of the two groups of points, and generate a new coding sequence;

[0037] Reversing the sequence fragment neighborhood transformation rule: that is, intercepting a small part of the coding sequence, reversing the fragment, and generating a new coding sequence;

[0038] Re-insert neighborhood transformation rule: randomly select a point and move it from its original position to the back of another point closest to it to generate a new coding sequence.

[0039] Preferably, the termination conditions for the execution of the algorithm include:

[0040] When the temperature drops to the minimum temperature; or the number of solutions that continuously produce the same objective function value reaches k max Second-rate.

[0041] Preferably, the objective function value f(w) is:

[0042]

[0043]

[0044]

[0045]

[0046]

[0047] Where P represents the set of candidate stops for the vehicle; T uav Indicates the maximum flight time of the drone, that is, the endurance of the drone; r i represents the number of drones when the vehicle arrives at the stop i, i∈P, when r i <0 indicates an infeasible solution; d i(i+1) represents the driving distance between two adjacent stops, i∈P; δ represents the time it takes for the UAV to visit a task; TA i represents the time when the vehicle arrives at the stop i, i∈P; TL i represents the time when the vehicle leaves the stop i, i∈P; Indicates that the drone takes off from the stop point j and passes through a series of mission points Then return to a complete path of stop point i, i∈P,j∈P; Indicates the flight distance of the drone to complete the path and the time to complete each section of the drone path Must be less than or equal to the drone's endurance T uav ; Indicates that the drone completes the path The time of arrival at stop i, j∈P; Represents the drone execution path The time of launch from stop j, j∈P; v gv represents the vehicle speed; v uav Represents the speed of the drone; M represents an infinite number.

[0048] In the second aspect, the present invention also proposes a collaborative path planning system for vehicles to assist multiple UAVs in performing tasks. The system includes: a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, the steps of the above-mentioned method are implemented.

[0049] (3) Beneficial effects

[0050] The present invention provides a method and system for collaborative path planning for multiple unmanned aerial vehicles (UAVs) to assist with mission execution. Compared with existing technologies, this method has the following advantages:

[0051] 1. This invention plans vehicle paths based on all mission points using a set covering model. Based on these vehicle paths, a simulated annealing algorithm with a hybrid neighborhood search is used to solve for collaborative path planning for multiple unmanned aerial vehicles (UAVs) performing missions. Compared to existing technologies, this invention achieves higher efficiency and higher quality solutions for collaborative path planning for multiple UAVs performing missions.

[0052] 2. The present invention designs an adaptive hybrid neighborhood search rule including neighborhood selection rule and neighborhood transformation rule, which changes the solution search process of the traditional simulated annealing algorithm and improves the calculation and solution efficiency of the simulated annealing algorithm.

[0053] 3. The present invention adds a storage link (memory function) to save the neighborhood solutions so far. If the same solution is generated, the neighborhood transformation will be re-applied until a new and different solution is generated. This avoids the problem of repeated calculation caused by the same neighborhood solution generated during the neighborhood transformation process, and improves the solution efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 A schematic diagram of the collaborative path planning of multiple UAVs assisted by a vehicle in performing a task according to the present invention;

[0056] Figure 2 Flowchart of a collaborative path planning method for vehicle-assisted multi-UAV mission execution according to an embodiment of the present invention;

[0057] Figure 3 is a schematic diagram of a reversal path in an embodiment of the present invention;

[0058] Figure 4 Schematic diagram of a situation in which the number of available drones at a launch point or a recovery point is negative in an embodiment of the present invention;

[0059] Figure 5 Schematic diagram of four preset neighborhood transformation rules in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0061] The embodiments of the present application provide a collaborative path planning method and system for vehicle-assisted multi-UAV mission execution, thereby solving the problems of poor solution quality and low solution efficiency in the prior art when facing the collaborative path planning problem of vehicle-assisted multi-UAV mission execution. It realizes the scientific and reasonable optimization of the collaborative path planning technology for vehicle-assisted multi-UAV mission execution, thereby assisting the efficient execution of related monitoring tasks in military and civilian fields such as intelligence reconnaissance and power inspection.

[0062] The technical solution in the embodiments of the present application is to solve the above technical problems, and the overall idea is as follows:

[0063] In order to solve the problems and defects of the existing technology in the face of the collaborative path planning problem of vehicle-assisted multi-UAV tasks, which has poor solution quality and low solution efficiency, the present application first reasonably divides the path planning problem of vehicle-assisted multi-UAV monitoring tasks into two stages: the vehicle path planning problem affected by the belt network constraints and the distribution of task points; the UAV path planning problem constrained by endurance and the joint planning problem of allocating the planned UAV path to the vehicle path. Then, for the first stage problem, a vehicle path construction sub-algorithm based on greedy thinking is proposed to solve it, and for the second stage problem, a UAV path construction sub-algorithm based on a simulated annealing algorithm with a hybrid neighborhood search is proposed to solve it, so as to obtain the optimal collaborative path planning solution for vehicle-assisted multi-UAV tasks. Compared with the existing technology, the technical solution of the present application has higher solution quality and higher solution efficiency in solving the vehicle-machine routing problem of collaborative tasks of increasing scale.

[0064] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0065] The technical solution of the present application studies the path planning problem of vehicle-assisted multi-UAVs performing monitoring tasks, where a group of UAVs perform corresponding tasks along the planned routes. In order to solve the problem that the UAVs are unable to complete tasks in a large area within the flight time due to the limited battery capacity of the UAVs, the technical solution of the present application uses vehicles as mobile base stations to expand the scope of the UAVs' tasks. In the technical solution of the present application, the process of vehicle-assisted multi-UAVs performing monitoring tasks can be described as follows: assuming that the vehicle is driving on the road network, and the UAVs are not restricted when flying in the air, the vehicle launches the UAV at a suitable position on the road network, the UAV performs nearby tasks, and then recovers the UAV at a suitable position, repeating the process until all tasks in the target area are completed, and finally the vehicle carries all the UAVs back to the warehouse. The ultimate goal of the path planning of vehicle-assisted multi-UAVs performing monitoring tasks is to solve the vehicle driving path that minimizes the time it takes for the UAVs to complete all tasks and return to the warehouse. As Figure 1 As shown, Figure 1 In the experiment, a vehicle assisted multiple drones in performing a monitoring mission, involving 30 target points and 3 drones, with the goal of minimizing the time it took for the drones to complete all tasks and return to the warehouse.

[0066] In order to solve the above-mentioned ultimate goal, the technical solution of this application breaks down the path planning problem of vehicle-assisted multi-UAV monitoring tasks into several sub-problems. The first is the vehicle path planning problem affected by the belt network constraints and the distribution of task points; the second is the UAV path planning problem subject to the endurance constraint; and the third is the joint planning problem of allocating the planned UAV path to the vehicle path. Based on this, the technical solution of this application specifically includes:

[0067] Example 1:

[0068] In the first aspect, the present invention first proposes a collaborative path planning method for vehicle-assisted multi-UAV mission execution, see Figure 2 , the method comprising:

[0069] S1. Vehicle path planning is performed based on all task points and using the set coverage model;

[0070] S2. Based on the vehicle path, a simulated annealing algorithm with hybrid neighborhood search is used to solve the collaborative path planning solution for vehicle-assisted multi-UAV mission execution.

[0071] As can be seen, this embodiment uses a set covering model to plan vehicle paths based on all mission points. Then, based on this vehicle path, a simulated annealing algorithm with a hybrid neighborhood search is used to solve the problem of collaborative path planning for multiple UAVs assisted by vehicles. Compared to existing technologies, this embodiment achieves higher solution efficiency and quality when solving collaborative path planning for multiple UAVs assisted by vehicles.

[0072] The following is combined with Figure 1-5 , and an explanation of the specific steps S1-S2, to describe in detail the implementation process of an embodiment of the present invention.

[0073] S1. Vehicle path planning is performed based on all task points and using the set coverage model.

[0074] To address the vehicle path planning problem in the first phase affected by the road network constraints and the distribution of mission points, this embodiment proposes a vehicle path construction method based on greedy thinking. The specific steps of this method include:

[0075] S11. Determine the mission points that can be covered by each road section in the road network based on the flight radius of the drone.

[0076] When a vehicle travels along a road network segment, it must traverse the entire segment, thus providing coverage for each segment. However, drones, due to battery capacity limitations, have a fixed flight radius. Based on the drone's flight radius (i.e., battery capacity limitations), the mission points that can be covered on each segment within the road network are determined, ensuring that each segment covers a certain range of mission points.

[0077] S12. Acquire a key road that can cover all the task points based on a set coverage model.

[0078] Based on the concept of set covering, we determine the critical roads that the vehicle must traverse to ensure that all mission points can be visited within the drone's cruising time. Specifically, we calculate the cost of each road segment based on the length of the critical road and the number and distance of the mission points covered. This is then formulated as a set covering problem, solved using a greedy approach with the goal of minimizing the total cost. Finally, we determine the critical roads that the vehicle must traverse and that cover all mission points. When calculating the cost of each road segment, we use the average distance from all mission points covered by the road to represent the cost of the vehicle on that road.

[0079] S13. Starting from the starting point, all the key roads are connected based on the principle of preferentially connecting the key road with the shortest distance to the current position to form a vehicle path.

[0080] All key roads are connected from the starting point. The next key road connected each time is the key road with the shortest distance to the end point of the previous road, so as to construct a vehicle path starting from the starting point and returning to the starting point.

[0081] S2. Based on the vehicle path, a simulated annealing algorithm with hybrid neighborhood search is used to solve the collaborative path planning solution for vehicle-assisted multi-UAV mission execution.

[0082] For the second stage, we first solve the UAV path planning problem subject to the endurance constraint, and then assign the planned UAV path to the vehicle path for joint planning, that is, the vehicle-assisted collaborative path planning of multiple UAVs performing tasks. In order to solve the problem of solving the second stage, this embodiment proposes a simulated annealing algorithm based on hybrid neighborhood search for vehicle-assisted collaborative path planning of multiple UAVs performing tasks. The specific process is as follows:

[0083] S21. Sort the projection points of all task points on the vehicle path in order of priority.

[0084] First, determine the projected position of the task point on the vehicle path, and use the projected point as a candidate stop point for the vehicle;

[0085] The projection points are then numbered according to their order on the vehicle path, and the task points are sorted according to the projection point numbers.

[0086] S22. Using the sorting result as the initial solution of the simulated annealing algorithm based on hybrid neighborhood search, and using the simulated annealing algorithm based on hybrid neighborhood search to solve the solution of the collaborative path planning of the vehicle-assisted multi-UAV mission.

[0087] Among them, the specific process of solving the collaborative path planning of vehicle-assisted multi-UAV missions based on the simulated annealing algorithm of hybrid neighborhood search is as follows:

[0088] S221 , setting and initializing relevant parameters of the simulated annealing algorithm, and inputting the sorting result as the initial solution w of the simulated annealing algorithm, and calculating the objective function value f(w).

[0089] Set the initial temperature T0 of the simulated annealing algorithm, and set the current temperature T = T0; set the annealing rate α; the minimum temperature T min ; The number of iterations of the neighborhood transformation and solution acceptance cycle L max ; The number of times the solution appears k max ; Initialize weight w i ; Weight coefficient ρ; Scoring rules in different situations δ i .

[0090] The above sorting result is input as the initial solution w of the simulated annealing algorithm, and the objective function value f(w) is calculated.

[0091] When sorting, each task point is given a number and encoded using natural numbers. Based on the drone's endurance, the initial encoding can be cut into several feasible drone paths; these feasible drone paths are then assigned to the vehicle path, and the objective function calculation formula is used to calculate the objective function value.

[0092] In this embodiment, the goal of collaborative path planning for a vehicle-assisted multi-UAV mission is to minimize the time it takes for the vehicle to carry all UAVs back to the warehouse. We can calculate the objective function based on the encoding. However, in our encoding, we cannot directly obtain the corresponding UAV path based on encoding and decoding. To this end, we first divide the encoding sequence according to the UAV's endurance, ensuring that each UAV path is feasible. We then design rules based on the number of available UAVs and discard infeasible solutions. Finally, we design a method for calculating the time required for vehicle-UAV synchronization. The specific ideas are as follows:

[0093] 1) Considering only the drone's endurance T w Here, path cutting means that if the entire mission has 1-90 mission points, the drone may need to change the battery after completing 1-3 mission points, then the path where the drone executes 1-3 mission points will be cut out as a sub-path. The total time it takes for the drone to take off from the launch point, visit a series of mission points, and return to the recovery point must be less than the drone's endurance. Based on this, the stop point corresponding to the first point of the sub-path is used as the launch point, and the stop point corresponding to the last point is used as the recovery point to determine the drone sub-path. Several drone sub-paths are obtained; each drone sub-path is then assigned to a vehicle path. Note that for a single drone sub-path, the launch point of the drone must be before or at the same point as the recovery point. Therefore, if the recovery point is before the launch point, the path is reversed. This is because in reality, drones must be launched first and then recovered. If, in a complete sub-path, the drone is recovered first and launched later, the order is incorrect, so simply reverse and swap. Figure 3 As shown in the figure, the stop point corresponding to the nth mission point is represented by n'. Since the stop point order is consistent with the mission point order, n' represents the nth stop point. Then this path represents launching the drone from the 8th stop point and recovering the drone at the 6th stop point. This is obviously unrealistic, so the drone sub-path is reversed, making the 6th stop point the launch point and the 8th stop point the recovery point.

[0094] 2) Based on the drone sub-paths cut above, calculate the number of available drones ri at the drone launch point and the recovery point. Calculate in the order that the vehicles pass. The vehicle carries N drones at the start. The number of available drones when arriving at the vehicle stop is equal to the number of drones remaining on the vehicle when the vehicle left the previous stop. The number of available drones when the vehicle leaves the stop is equal to the number of drones when the vehicle arrived at the stop minus the number of launched drones plus the number of recovered drones. If the number of available drones at a certain point is negative, it means that an infeasible solution has been generated, such as Figure 4 As shown in the figure, assume that there are 2 drones in total, which are launched at the 1st and 3rd stop points respectively, and no drone is recovered before the 4th stop point. Therefore, there are no free drones available at the 4th stop point, and no free drones can complete the path. This means that an infeasible solution has been generated. At this time, the objective function value corresponding to the solution is set to an infinite number (theoretically, it is an infinite number, but in practice, a relatively large number that meets the calculation requirements will be set).

[0095] 3) If the number of available drones calculated at the drone launch and recovery points in 2) above is not a negative number, then further calculate the time when the vehicle arrives at and leaves each stop.

[0096] The time it takes for a vehicle to arrive at a stop is equal to the time it leaves the previous stop plus the time it travels between the two stops, as shown in formula (1).

[0097] The time it takes for a vehicle to leave a stop is divided into two cases, as shown in formula (2): a. If there is no drone recovered at the point, the time it takes to leave the point is equal to the time it takes to arrive at the point (launching a drone does not consume time); b. If there is a drone recovered at the point, it is necessary to recover all drones before leaving. The time it takes for the drone to return to the stop can be calculated using formulas (3) and (4).

[0098] The time it takes for the vehicle to finally return to the starting point is taken as the objective function value, as shown in formula (5): represents the vehicle's travel time, ∑ i∈P (TL i -TA i ) represents the time the vehicle and the drone wait for each other.

[0099]

[0100]

[0101]

[0102]

[0103]

[0104] Where P represents the set of candidate stops for the vehicle; T uav Indicates the maximum flight time of the drone, that is, the endurance of the drone; r i represents the number of drones when the vehicle arrives at the stop i, i∈P, when r i <0 indicates an infeasible solution; d i(i+1) represents the driving distance between two adjacent stops, i∈P; δ represents the time it takes for the UAV to visit a task; TA i represents the time when the vehicle arrives at the stop i, i∈P; TL i represents the time when the vehicle leaves the stop i, i∈P; Indicates that the drone takes off from the stop point j and passes through a series of mission points Then return to a complete path of stop point i, i∈P,j∈P; Indicates the flight distance of the drone to complete the path and the time to complete each section of the drone path Must be less than or equal to the drone's endurance T uav ; It represents the time it takes to launch a drone from any stop j and return to stop i. Represents the drone execution path The time of launch from stop j; v gv represents the vehicle speed; vuav Represents the speed of the drone; M represents an infinite number.

[0105] S222. Perform neighborhood transformation on the current solution w according to the preset neighborhood selection rule and the preset neighborhood transformation rule to generate a new solution w'. If the new solution w' has not been stored, calculate the objective function value f(w') of the new solution w'; otherwise, continue to perform neighborhood transformation on the preset neighborhood transformation rule according to the preset neighborhood selection rule until a new solution w' that has not been stored is generated, and calculate the objective function value f(w').

[0106] During the execution of the simulated annealing algorithm, several neighborhood transformation rules can be used to alter the neighborhood to adjust the current solution, thus constructing a simulated annealing algorithm with a hybrid neighborhood search. In traditional simulated annealing, each new solution requires a comparison of the objective function value. To improve the computational efficiency of simulated annealing, we adopted a multi-search strategy and designed an adaptive hybrid neighborhood selection rule. Finally, to avoid repeated calculations due to the generation of the same neighborhood solution during neighborhood transformation, a storage step (memory) was added to save the neighborhood solutions to date. Based on this approach, we designed four neighborhood transformation rules and one neighborhood selection rule for neighborhood adjustment, thus increasing the algorithm's memory function.

[0107] The current solution w is transformed according to the preset neighborhood transformation rules. The preset neighborhood transformation rules include two-point exchange neighborhood transformation rule, two-pair point exchange neighborhood transformation rule, reverse sequence fragment neighborhood transformation rule, and reinsert neighborhood transformation rule. Specifically,

[0108] ①See Figure 5 a, is the two-point exchange neighborhood transformation rule, that is, randomly select two points in the coding sequence, exchange the positions of the two points, and generate a new coding sequence;

[0109] ②See Figure 5 b, is the neighborhood transformation rule for exchanging two pairs of points, that is, randomly selecting two groups of adjacent points in the coding sequence, exchanging the positions of the two groups of points, and generating a new coding sequence;

[0110] ③See Figure 5 c, is the neighborhood transformation rule for reversing sequence fragments, that is, intercepting a small part of the coding sequence, reversing the fragment, and generating a new coding sequence;

[0111] ④See Figure 5 d is the reinsertion neighborhood transformation rule, which randomly selects a point and inserts it behind another point closest to it to generate a new coding sequence.

[0112] In this embodiment, when the current solution w is transformed according to the preset neighborhood transformation rule, the neighborhood transformation rule is selected based on the preset neighborhood selection rule. Specifically, we designed an adaptive hybrid neighborhood selection rule that uses a multiple search strategy in each state instead of the single comparison method in the traditional simulated annealing algorithm. In our method, each neighborhood transformation rule has a weight, and the initial weight w i The weights of each neighborhood transformation rule are updated in an adaptive manner:

[0113]

[0114] Among them, w' i represents the weight of the updated i-th neighborhood transformation rule; s i represents the score of the i-th neighborhood transformation rule in the same state, n represents the number of times the i-th neighborhood transformation rule is used in the same state, and ρ∈(0,1) is the coefficient for recalculating the weight. Roulette is used and several neighborhood transformation rules are selected based on the weight of each neighborhood transformation rule to perform neighborhood transformation operations on the current solution w to obtain a new solution w'.

[0115] The score of each neighborhood transformation rule (s i ) The update principle is: if the new solution is accepted and the resulting new solution is better than the current optimal solution, the score is +σ1; if it is a new solution accepted according to the Metropolis criterion, the score is +σ2; otherwise, the score is +σ3.

[0116] Roulette wheel selection: We use a roulette wheel to select neighborhood transformation rules. For example, if the current weights are [w1,w2,w3,w4], calculate the cumulative weight of each rule, that is, [w1,w1+w2,w1+w2+w3,w1+w2+w3+w4], and then generate a random number between 0 and (w1+w2+w3+w4). If the weight w of a neighborhood transformation rule is i If it is greater than this random number, the neighborhood transformation rule is selected.

[0117] Since the above neighborhood transformation is random, in order to avoid the same solution during the search process, a storage link can be added to record the solutions generated so far. If the same solution is generated, the neighborhood transformation will be re-applied until a new and different solution is generated.

[0118] S223. Compare f(w') and f(w). If f(w') ≤ f(w), accept the new solution w'. Otherwise, determine whether to accept the new solution w' according to the Metropolis criterion. If the new solution w' is accepted, store the new solution w'.

[0119] S224, repeat steps S222 and S223 at the current temperature T to iterate, and save the best solution of the objective function value obtained by iteration to prevent it from being lost during the iteration process, until the current number of iterations l reaches the preset maximum number of iterations L max Execute step S225.

[0120] Repeat L at current temperature T max After the process of neighborhood transformation and solution acceptance (i.e., executing steps S222 and S223) (iteration number), step S225 is executed.

[0121] S225. Determine whether the termination condition of the algorithm execution is met. If so, execute S227; otherwise, go to S226.

[0122] Among them, the termination condition of the algorithm is: when the temperature drops to the lowest temperature T min , or the number of solutions that continuously produce the same objective function value reaches k max Second-rate.

[0123] When the algorithm meets any of the above termination conditions, the algorithm will stop and the current solution will be output as the optimal solution, that is, step S227 will be executed; otherwise, the process will go to S226.

[0124] S226 , the temperature T gradually decreases, the score of each neighborhood transformation rule and the weight of each neighborhood transformation rule are updated, and S222 is executed.

[0125] S227: The algorithm is executed and the current solution is output as the optimal solution.

[0126] At the end of the algorithm execution, the current solution is output as the optimal solution, which is the final path planning scheme for the collaborative path planning of vehicle-assisted multi-UAV missions.

[0127] At this point, the entire process of the collaborative path planning method for vehicle-assisted multi-UAV mission execution in this embodiment is completed.

[0128] In order to verify that the technical solution of this embodiment has higher solution quality and efficiency than the existing technology when facing the collaborative path planning problem of vehicle-assisted multi-UAV tasks, we have proved it through experiments.

[0129] We used the method (algorithm) of this application and the VAMU algorithm in the prior art to solve the path planning problem of large-scale vehicle-assisted multi-UAV monitoring tasks, and compared the results.

[0130] There are five groups of problem instances, A1-A5, representing five different road networks, all with a regional dimension of 10,000. Within each group, we designed five scales. For example, A1-30 represents 30 randomly generated task points in the first network. For each scale, we generated data 10 times. The VAMU algorithm was run only once, while the algorithm was run 10 times to ensure algorithm validity, and the average results were taken.

[0131] The relevant parameters for all experiments are as follows: the maximum flight time of the drone is 1500 seconds, the drone speed is 5 units per second, and the vehicle speed is 30 units per second. The first column lists the numbers of all scales in the five groups of examples. Columns 2 to 7 show the results of the VAMU and our method: Obj is the objective function value, and Time is the computation time in seconds. The specific comparison results are shown in Table 1.

[0132] Table 1 Experimental comparison results between this algorithm and VAMU algorithm

[0133]

[0134] The results in Table 1 show that this algorithm achieves higher solution quality and faster solution efficiency than the VAMU algorithm. This shows that compared to existing VAMU methods, this method has a significant performance advantage in solving the vehicle-machine routing problem for collaborative missions between vehicles and multiple UAVs.

[0135] Example 2:

[0136] In a second aspect, the present invention further provides a vehicle-assisted collaborative path planning system for multiple UAVs to perform tasks, the system comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method, which mainly comprises:

[0137] Vehicle path planning is performed based on all mission points and using the set coverage model;

[0138] Based on the vehicle path, a simulated annealing algorithm with hybrid neighborhood search is used to solve the collaborative path planning solution for vehicle-assisted multi-UAV missions.

[0139] Optionally, the vehicle path planning based on all task points and using a set coverage model includes:

[0140] S11. Determine the mission points that can be covered on each road section in the road network based on the flight radius of the drone;

[0141] S12. Acquire a key road that can cover all the task points based on a set coverage model;

[0142] S13. Starting from the starting point, all the key roads are connected based on the principle of preferentially connecting the key road with the shortest distance to the current position to form a vehicle path.

[0143] Optionally, the method of solving the collaborative path planning for vehicle-assisted multi-UAV missions using a simulated annealing algorithm with a hybrid neighborhood search based on the vehicle path includes:

[0144] S21, taking the projection points of all the task points on the vehicle path as candidate vehicle stop points, and sorting the task points according to the order of the candidate vehicle stop points on the vehicle path;

[0145] S22. Using the task point sorting result as the initial solution of the simulated annealing algorithm based on the hybrid neighborhood search, and using the simulated annealing algorithm based on the hybrid neighborhood search to solve the solution of the collaborative path planning of the vehicle-assisted multi-UAV mission.

[0146] Optionally, the S22 includes:

[0147] S221, setting and initializing relevant parameters of the simulated annealing algorithm, and using the task point sorting result as the initial solution input of the simulated annealing algorithm, and calculating the objective function value;

[0148] S222. Perform neighborhood transformation on the current solution w according to the preset neighborhood selection rule and the preset neighborhood transformation rule to generate a new solution w'. If the new solution w' has not been stored, calculate the objective function value f(w') of the new solution w'. Otherwise, continue to perform neighborhood transformation according to the preset neighborhood selection rule and the preset neighborhood transformation rule until a new unstored solution w' is generated, and calculate the objective function value f(w').

[0149] S223. Compare f(w') and f(w). If f(w') ≤ f(w), accept the new solution w'. Otherwise, determine whether to accept the new solution w' according to the Metropolis criterion. If the new solution w' is accepted, store the new solution w'.

[0150] S224, repeating steps S222 and S223 at the current temperature T, and saving the best solution corresponding to the objective function value iterated in this state; and executing step S225 until the current number of iterations reaches the preset maximum number of iterations;

[0151] S225, determine whether the termination condition of the algorithm execution is met, if so, execute S227; otherwise, go to S226;

[0152] S226, the temperature T gradually decreases, the score of each neighborhood transformation rule and the weight of each neighborhood transformation rule are updated, and S222 is executed;

[0153] S227: The algorithm is executed and the current solution is output as the optimal solution.

[0154] Optionally, the preset neighborhood selection rule includes:

[0155] updating the weight of each neighborhood transformation rule, and then adopting roulette and selecting a neighborhood transformation rule according to the weight of each neighborhood transformation rule;

[0156] The weight of each neighborhood transformation rule is updated as follows:

[0157]

[0158] Among them, w' i represents the weight of the updated i-th neighborhood transformation rule; w i represents the initial weight of the i-th neighborhood transformation rule; s i represents the score of the i-th neighborhood transformation rule in the same state; n represents the number of times the i-th neighborhood transformation rule is used in the same state; ρ∈(0,1) is the coefficient for recalculating the weight;

[0159] Optionally, the score s for each neighborhood transformation rule i The update principle is:

[0160] If the new solution is accepted and is better than the current optimal solution, the score is +σ1; if it is a new solution accepted according to the Metropolis criterion, the score is +σ2; otherwise, the score is +σ3;

[0161] Optionally, the preset neighborhood transformation rule includes:

[0162] Two-point exchange neighborhood transformation rule: randomly select two points in the coding sequence, exchange the positions of the two points, and generate a new coding sequence;

[0163] Two-pair point exchange neighborhood transformation rule: randomly select two groups of adjacent points in the coding sequence, exchange the positions of the two groups of points, and generate a new coding sequence;

[0164] Reversing the sequence fragment neighborhood transformation rule: that is, intercepting a small part of the coding sequence, reversing the fragment, and generating a new coding sequence;

[0165] Re-insert neighborhood transformation rule: randomly select a point and move it from its original position to the back of another point closest to it to generate a new coding sequence.

[0166] Optionally, the termination conditions of the algorithm include: when the temperature drops to the minimum temperature; or when the number of solutions that continuously produce the same objective function value reaches kmax Second-rate.

[0167] Optionally, the objective function value f(w) is:

[0168]

[0169]

[0170]

[0171]

[0172]

[0173] Where P represents the set of candidate stops for the vehicle; T uav Indicates the maximum flight time of the drone, that is, the endurance of the drone; r i represents the number of drones when the vehicle arrives at the stop i, i∈P, when r i <0 indicates an infeasible solution; d i(i+1) represents the driving distance between two adjacent stops, i∈P; δ represents the time it takes for the UAV to visit a task; TA i represents the time when the vehicle arrives at the stop i, i∈P; TL i represents the time when the vehicle leaves the stop i, i∈P; Indicates that the drone takes off from the stop point j and passes through a series of mission points Then return to a complete path of stop point i, i∈P,j∈P; Indicates the flight distance of the drone to complete the path and the time to complete each section of the drone path Must be less than or equal to the drone's endurance T uav ; Indicates that the drone completes the path The time of arrival at stop i, j∈P; Represents the drone execution path The time of launch from stop j, j∈P; v gv represents the vehicle speed; v uav Represents the speed of the drone; M represents an infinite number.

[0174] It can be understood that the collaborative path planning system for vehicle-assisted multi-UAV mission execution provided in the embodiment of the present invention corresponds to the above-mentioned collaborative path planning method for vehicle-assisted multi-UAV mission execution. The explanation, examples, beneficial effects, etc. of the relevant contents can refer to the corresponding contents in the collaborative path planning method for vehicle-assisted multi-UAV mission execution, and will not be repeated here.

[0175] In summary, compared with the existing technology, the present invention has the following beneficial effects:

[0176] 1. This invention plans vehicle paths based on all mission points using a set covering model. Based on these vehicle paths, a simulated annealing algorithm with a hybrid neighborhood search is used to solve for collaborative path planning for multiple unmanned aerial vehicles (UAVs) performing missions. Compared to existing technologies, this invention achieves higher efficiency and higher quality solutions for collaborative path planning for multiple UAVs performing missions.

[0177] 2. The present invention designs an adaptive hybrid neighborhood search rule including neighborhood selection rule and neighborhood transformation rule, which changes the solution search process of the traditional simulated annealing algorithm and improves the calculation and solution efficiency of the simulated annealing algorithm.

[0178] 3. The present invention adds a storage link (memory function) to save the neighborhood solutions so far. If the same solution is generated, the neighborhood transformation will be re-applied until a new and different solution is generated. This avoids the problem of repeated calculation caused by the same neighborhood solution generated during the neighborhood transformation process, and improves the solution efficiency.

[0179] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0180] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A collaborative path planning method for vehicle-assisted multi-UAV mission execution, characterized in that: The method comprises: Vehicle path planning is performed based on all mission points and using a set coverage model; Based on the vehicle path, a simulated annealing algorithm with hybrid neighborhood search is used to solve the collaborative path planning solution for vehicle-assisted multi-UAV missions, including: S21, taking the projection points of all the task points on the vehicle path as candidate vehicle stop points, and sorting the task points according to the order of the candidate vehicle stop points on the vehicle path; S22. Using the task point sorting result as the initial solution of a simulated annealing algorithm based on a hybrid neighborhood search, and using the simulated annealing algorithm based on the hybrid neighborhood search to solve the solution for the collaborative path planning of the vehicle-assisted multi-UAV mission, including: S221, setting and initializing relevant parameters of the simulated annealing algorithm, and using the task point sorting result as the initial solution input of the simulated annealing algorithm, and calculating the objective function value; S222. Perform neighborhood transformation on the current solution w according to the preset neighborhood selection rule and the preset neighborhood transformation rule to generate a new solution w'. If the new solution w' has not been stored, calculate the objective function value f(w') of the new solution w'. Otherwise, continue to perform neighborhood transformation according to the preset neighborhood selection rule and the preset neighborhood transformation rule until a new unstored solution w' is generated, and calculate the objective function value f(w'). S223. Compare f(w') and f(w). If f(w') ≤ f(w), accept the new solution w'. Otherwise, determine whether to accept the new solution w' according to the Metropolis criterion. If the new solution w' is accepted, store the new solution w'. S224, repeating steps S222 and S223 at the current temperature T, and saving the best solution corresponding to the objective function value obtained in this state as the optimal solution at the current temperature, until the current number of iterations reaches the preset maximum number of iterations, and executing step S225; S225, determine whether the termination condition of the algorithm execution is met, if so, execute S227; otherwise, go to S226; S226, the temperature T gradually decreases, the score of each neighborhood transformation rule and the weight of each neighborhood transformation rule are updated, and S222 is executed; S227: The algorithm is executed and the current solution is output as the optimal solution.

2. The method according to claim 1, wherein The vehicle path planning based on all task points and using the set covering model includes: S11. Determine the mission points that can be covered on each road section in the road network based on the flight radius of the drone; S12. Acquire a key road that can cover all the task points based on a set coverage model; S13. Starting from the starting point, all the key roads are connected based on the principle of preferentially connecting the key road with the shortest distance to the current position to form a vehicle path.

3. The method according to claim 1, wherein The preset neighborhood selection rules include: updating the weight of each neighborhood transformation rule, and then adopting roulette and selecting a neighborhood transformation rule according to the weight of each neighborhood transformation rule; The weight of each neighborhood transformation rule is updated as follows: Among them, w ′ i represents the weight of the updated i-th neighborhood transformation rule; w i represents the initial weight of the i-th neighborhood transformation rule; s i represents the score of the i-th neighborhood transformation rule under the same state; n represents the number of times the i-th neighborhood transformation rule is used under the same state; ρ∈(0,1) is the coefficient for recalculating the weight.

4. The method according to claim 3, wherein The score s of each neighborhood transformation rule i The update principle is: If the new solution is accepted and the resulting new solution is better than the current optimal solution, the score is +σ1; if it is a new solution accepted according to the Metropolis criterion, the score is +σ2; otherwise, the score is +σ3.

5. The method according to claim 1, wherein The preset neighborhood transformation rules include: Two-point exchange neighborhood transformation rule: randomly select two points in the coding sequence, exchange the positions of the two points, and generate a new coding sequence; Two-pair point exchange neighborhood transformation rule: randomly select two groups of adjacent points in the coding sequence, exchange the positions of the two groups of points, and generate a new coding sequence; Reversing the sequence fragment neighborhood transformation rule: that is, intercepting a small part of the coding sequence, reversing the fragment, and generating a new coding sequence; Re-insert neighborhood transformation rule: randomly select a point and move it from its original position to the back of another point closest to it to generate a new coding sequence.

6. The method according to claim 1, wherein The termination conditions for the execution of the algorithm include: When the temperature drops to the minimum temperature; or the number of solutions that continuously produce the same objective function value reaches k max Second-rate.

7. The method according to claim 1, wherein The objective function value f(w) is: Where P represents the set of candidate stops for the vehicle; T uav Indicates the maximum flight time of the drone, that is, the endurance of the drone; r i represents the number of drones when the vehicle arrives at the stop i, i∈P, when r i <0 indicates an infeasible solution; d i(i+1) represents the driving distance between two adjacent stops, i∈P; δ represents the time it takes for the UAV to visit a task; TA i represents the time when the vehicle arrives at the stop i, i∈P; TL i represents the time when the vehicle leaves the stop i, i∈P; Indicates that the drone takes off from the stop point j and passes through a series of mission points Then return to a complete path of stop point i, i∈P,j∈P; Indicates the flight distance of the drone to complete the path and the time to complete each section of the drone path Must be less than or equal to the drone's endurance T uav ; Indicates that the drone completes the path The time of arrival at stop i, j∈P; Represents the drone execution path The time of launch from stop j, j∈P; v gv represents the vehicle speed; v uav Represents the speed of the drone; M represents an infinite number.

8. A vehicle-assisted collaborative path planning system for multiple unmanned aerial vehicles to perform tasks, the system comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.

Images