Multi-unmanned aerial vehicle path efficient reconstruction method for complex terrains and sudden tasks

Abstract

The technical problems that task redistribution and path replanning of dynamic burst tasks cannot be achieved through an existing path replanning algorithm, dynamic redistribution of the tasks is difficult to conduct in real time in the face of the dynamic burst tasks through an existing task redistribution algorithm, and the processing capacity is insufficient in the face of complex terrains are solved. The invention provides a multi-unmanned aerial vehicle path efficient reconstruction method oriented to complex terrains and sudden tasks, which considers factors such as a large number of task demand constraints, fuel constraints, complex elevation terrains and sudden tasks. Low-altitude real-time local path re-planning which is used for reasonably evaluating and sorting sudden task values, optimally selecting unmanned aerial vehicles and considering threat degrees and digital terrain representation is designed for an existing unmanned aerial vehicle task sequence so as to realize efficient path reconstruction; when the burst task is inserted, the optimal insertion point strategy adopted by the invention reduces the repeated calculation process of the insertion task point cost, the time complexity of calculation is effectively reduced, and the real-time index of the algorithm is ensured.

Description

Technical Field

[0001] The present invention relates to a method for reconstructing paths of multiple unmanned aerial vehicles. Background Art

[0002] Due to its unique characteristics of vertical takeoff and landing, hovering, and flexibility, unmanned helicopters have important application values when performing tasks such as reconnaissance and surveillance, target tracking, ground attack, material transportation, and casualty evacuation. When multiple unmanned helicopters are applied to low-altitude environments, they need to plan a safe path from the starting point to the target point for each unmanned helicopter through path planning under conditions such as low-altitude complex digital elevation model (DEM) environments and networked radar detection, and the fuel constraints of the unmanned helicopters are satisfied. However, during the process of an unmanned helicopter performing a series of tasks, sudden task requirements may occur. How to efficiently complete real-time path reconstruction based on the paths planned for existing tasks will greatly improve the flexibility and survivability of multiple unmanned helicopters in handling tasks.

[0003] Different from traditional path replanning algorithms for dealing with dynamically changing environments, path reconstruction algorithms emphasize the task reallocation and path replanning processes for dynamic and sudden tasks. The main problems that path reconstruction algorithms need to solve include: (1) coordinating the task allocation of multiple unmanned helicopters under fuel constraints and task requirement constraints to ensure that the fuel of the unmanned helicopters can support them to meet flight tasks, and multiple unmanned helicopters can complete all tasks almost simultaneously; (2) identifying the optimal insertion position of a sudden task in the existing task sequence of the selected unmanned helicopter; (3) considering finding a locally optimal replanned path on a digital elevation map.

[0004] Therefore, when multiple unmanned helicopters perform path reconstruction for sudden tasks, they not only need to consider the problem of local safe path replanning with the shortest possible path-finding time for a single unmanned helicopter under complex terrain and fire threat conditions, but also need to solve the task allocation problem of multiple unmanned helicopters under fuel constraints in the face of sudden tasks, and consider the time coordination problem among multiple unmanned helicopters to ensure that multiple unmanned helicopters can execute all tasks almost simultaneously in the shortest possible time.

[0005] Currently, for the distributed real-time task reallocation algorithms of multiple unmanned helicopters in emergency tasks, there are market auction algorithms, consensus-based bundling algorithms (CBBA), heuristic performance impact algorithms (PI), Markov decision processes (MDP), etc. Among them, when the market auction algorithm is used for large-scale task bidding, on the one hand, it is difficult to reasonably estimate the bidding cost, and on the other hand, it insufficiently considers the task demand constraints in the dynamic environment and the fuel constraints of unmanned helicopters, and ignores the impact of complex terrain and emergency tasks on path reconstruction, resulting in difficulty in obtaining a real-time high-quality bidding scheme. Both the CBBA algorithm and the PI algorithm need to repeatedly calculate the cost after inserting new tasks into the original task sequence, and more computational effort will be generated in the task conflict resolution calculation, resulting in low computational efficiency and poor real-time performance, which is not suitable for the dynamic reallocation process. The MDP algorithm is difficult to handle the high-dimensional space and complex environment in the case of a large number of unmanned helicopters and tasks. It usually needs to use the deep reinforcement learning (DRL) method for repeated training, and the training quality is difficult to guarantee, and its ability to handle sudden threats and tasks is insufficient. Summary of the Invention

[0006] In order to solve the technical problems that the existing path replanning algorithms cannot achieve task reallocation and path replanning for dynamic emergency tasks, and the existing task reallocation algorithms are difficult to perform dynamic reallocation of tasks in real time in the face of dynamic emergency tasks and have insufficient processing capabilities in the face of complex terrain, the present invention provides a method for efficient path reconstruction of multiple unmanned aerial vehicles for complex terrain and emergency tasks. The present invention can enable multiple unmanned helicopters to perform real-time and high-quality low-altitude task reallocation and path replanning processes when facing a large number of emergency task requirements and complex terrain, while satisfying task demand constraints and fuel constraints.

[0007] The technical solution adopted by the present invention is as follows:

[0008] A method for efficient path reconstruction of multiple unmanned aerial vehicles for complex terrain and emergency tasks is characterized by including the following steps:

[0009] Step 1: Evaluate the value of each emergency task point in the set of emergency task points to obtain the value of each emergency task point;

[0010] Step 2: Sort the values in descending order to obtain a value set;

[0011] Step 3: For the emergency task point Ta corresponding to the first value in the current value set * , based on the improved A* algorithm, calculate the optimal path list P corresponding to each unmanned aerial vehicle i if the emergency task point Ta i is inserted into the task sequence T currently executed by each unmanned aerial vehicle i * , i* and the cumulative distance change Δd(P i ); i takes 1, 2, …, N respectively; N is the total number of UAVs performing tasks at the initial moment; the cost function of the improved A* algorithm is designed based on the digital elevation map, including longitude, latitude and altitude information, and converts the three-dimensional node search of the original A* algorithm into a two-dimensional node search;

[0012] Step 4: Screen out the UAV set F that still satisfies the fuel constraint after inserting the emergency task point Ta * :

[0013] F = {i|Δd(P i ) ≤ u i δ i - d(P i )};

[0014] Step 5: Judge whether the UAV set F is empty. If so, the current emergency task point Ta * is not executed, and go to Step 8; if not, go to Step 6;

[0015] Step 6: Screen out the UAV i * from the UAV set F, such that after inserting the emergency task point Ta * into the task sequence of the UAV i * , the flight time difference between any two of the N UAVs is as small as possible;

[0016] Step 7: Arrange the current emergency task point Ta * for the UAV i * , and update the path list and the shortest flight length of the task sequence currently to be executed by the UAV i * ;

[0017] Step 8: Delete the first value in the current value set;

[0018] Step 9: Judge whether the current value set is empty. If so, output the path lists of the N UAVs, and the process ends; otherwise, return to Step 3.

[0019] Furthermore, the value of each emergency task point in Step 1 is calculated according to the following formula:

[0020]

[0021] In the formula,

[0022] val k represents the value of the emergency task point Ta k ; k = 1, 2...|S|; |S| is the total number of emergency tasks;

[0023] Function J k is the task demand for the emergency task point Ta k ;

[0024] m i is the mass of UAV i;

[0025] k i is the contribution of unit mass of UAV i to any task;

[0026] X ik is a Boolean decision variable. If UAV i has selected the emergency task point Ta k , then X ik = 1, otherwise X ik = 0;

[0027] ρ i (Ta k ) represents the shortest distance from the emergency task point Ta k to all the task points currently accepted by UAV i, represents the straight-line distance on the digital elevation map between the position of the task point and the position of the emergency task point Ta k ; is the j-th task in the task sequence currently executed by UAV i ;

[0028] Furthermore, the principle of descending order in step 2 is: First, sort in descending order according to the size of the first element k in the list of Val . If the sizes of the first elements are the same, then sort in descending order according to the size of the second element ;

[0029] Furthermore, the cost function of the improved A* algorithm in step 3 is:

[0030] F n = ω G G n + ω H H n + ω I I n + ω c C n ;

[0031]

[0032] H n = k lon |lon n - lon Ta | + k lat |latn -lat Ta |;

[0033]

[0034]

[0035] In the formula,

[0036] G n , H n , I n , C n are respectively the cumulative map distance cost, heuristic cost, height cost and threat degree cost of the improved A* algorithm at node n, and their initial values are all 0;

[0037] ω G , ω H , ω I , ω C are respectively n , H n , I n , C n weight coefficients;

[0038] G neighbor(n) is the cumulative map distance cost of the improved A* algorithm at the neighboring node of node n;

[0039] k lon and k lat respectively represent the map resolutions along the longitude and latitude directions;

[0040] lon n and lat n are respectively the longitude and latitude coordinates of the current node n; lon Ta and lat Ta are respectively the longitude and latitude coordinates of the target point Ta;

[0041] hei n is the terrain height of the current node n, hei min is the lowest altitude of the entire digital elevation terrain, hei max is the highest altitude of the entire digital elevation terrain, α hei is the equivalent altitude coefficient,

[0042] C 0 is the intensity of the threat source, h(n, radar) is the straight-line distance from the current node n to the threat source, and R(radar) represents the influence range of the threat source.

[0043] Furthermore, step 3 is specifically as follows:

[0044] Step 3.1: Initialize the index i of the UAVs to be traversed, i = 1;

[0045] Step 3.2: Traverse the task sequence currently executed by UAV i Calculate the best insertion position j of the emergency task point Ta * using the following formula * :

[0046]

[0047] where represents the optimal distance between the task point and the emergency task point Ta * considering the complex terrain factor and threat factor, compared with the distance the increased percentage;

[0048] respectively represent the optimal distance between the emergency task point Ta * and the task point considering the complex terrain factor and threat factor, compared with the distance the increased percentage;

[0049] and are respectively the prior estimates of and ;

[0050] represents the straight-line distance between the position of the k-th task point in the task sequence of UAV i on the digital elevation map and the position of the emergency task point Ta * ;

[0051] represents the straight-line distance between the position of the emergency task point Ta * and the position of the (k + 1)-th task point in the task sequence of UAV i on the digital elevation map;

[0052] In the first iteration, i.e., when i = 1, the same prior estimation method as is used for calculation; when i > 1 it has been calculated by the improved A* algorithm in the previous task sequence and is a known value, which can be directly called;

[0053] When k = L i is the assumed task point; ​

[0054] Step 3.3: Calculate the optimal path list * after inserting the burst task point Ta into the current task sequence executed by UAV i and the cumulative distance change Δd(P i );

[0055] Step 3.3.1: Use the improved A* algorithm to calculate the current best insertion position j * and the optimal path * between the burst task point Ta optimal distance the burst task point Ta * and the position j * +1 optimal distance and the current best insertion position j * and its next task point j * +1

[0056] Step 3.3.2: Calculate the cumulative distance change Δd(P * when the burst task point Ta * is inserted at the current best insertion position j i ):

[0057]

[0058] In the formula, when the burst task point Ta * is inserted after the last task point , and are both 0;

[0059] Step 3.3.3: Calculate the formula for the optimal path list of UAV i * when the burst task point Ta is inserted into the current task sequence executed by UAV i * as follows:

[0060]

[0061] In the formula, when the burst task point Ta * is inserted after the last task point , and are both empty sets;

[0062] ​Step 3.4: Determine whether all drones have been traversed. If not, set i ← i + 1 and return to Step 3.1; if so, proceed to Step 4.

[0063] Further, in Step 6, drone i * is determined using the following formula:

[0064]

[0065] In the formula,

[0066] is the average flight speed of drone i;

[0067] is the average flight speed of drone j;

[0068] d(P i ) is the path length of the original path list of drone i before tentatively inserting the emergency task point Ta * ; The path length;

[0069] d(P j ) is the path length of the original path list of drone j before tentatively inserting the emergency task point Ta * ; The path length.

[0070] Further, Step 7 is specifically: The new path list obtained by inserting the emergency task point Ta * into the path list of the task sequence currently being executed by drone i * is updated to the new path list that drone i currently needs to execute. The new shortest flight length obtained by inserting the emergency task point Ta * into the path list of the task sequence currently being executed by drone i * is updated to the shortest flight length of the new path list that drone i * currently needs to execute. is updated to the shortest flight length of the new path list that drone i * currently needs to execute.

[0071] The beneficial effects of the present invention are:

[0072] 1. The present invention comprehensively considers a large number of task requirement constraints, fuel constraints, complex elevation terrain, and sudden tasks. For the existing UAV task sequences, it designs a method for reasonably evaluating and sorting the value of sudden tasks, selecting the optimal UAV, and performing low-altitude real-time local path replanning considering threat degree and digital terrain representation to achieve efficient path reconstruction. It solves the problems of the allocation of multiple UAVs for a large number of task orders, and the difficulty in balancing computational efficiency and task allocation quality under complex terrain, threat degree constraints, fuel constraints, and task constraints, improving the survival ability and task execution efficiency of multiple UAVs when facing sudden situations during the execution of complex tasks.

[0073] 2. Compared with traditional market auction algorithms and the CBBA algorithm, the optimal insertion point strategy proposed by the present invention greatly reduces the repeated calculation process of the cost of inserting task points, can effectively reduce the time complexity of calculation, improve computational efficiency, and ensure the real-time performance index of the algorithm.

[0074] 3. When dealing with new tasks, the PI algorithm is difficult to accurately evaluate the impact of factors such as complex terrain and sudden task requirements on the performance of UAVs, and its computational complexity is greater, which results in lower processing efficiency for large-scale UAV clusters and a large number of tasks; while the present invention fully considers the impact of complex terrain factors, threat degree, fuel constraints, task requirement constraints, etc. on path reconstruction, and can effectively improve the solution efficiency while taking into account the solution quality.

[0075] 4. Compared with the MDP method, the present invention has the advantages of simple operation, being able to adapt to a larger number of UAVs and greater task volume requirements, and being easily extended with more task constraints.

[0076] 5. Traditional algorithms such as elastic band and D* algorithm are difficult to handle path rapid replanning problems in unstructured environments such as digital elevation terrain (DEM), and such algorithms rarely consider the impact of factors such as threat degree, fuel constraints, and task requirements on the path replanning results; the original A* algorithm has the following disadvantages: 1) It is only applicable to structured terrain and some obstacle terrains, and is not applicable to digital elevation terrain; 2) The original A* algorithm has a large space for node expansion when dealing with 3D path planning, which will lead to a reduction in computational efficiency. The present invention improves the cost function of the original A* algorithm to make it suitable for optimal path search of digital elevation terrain, thereby reducing the 3D grid search to 2D search. Since the improved cost function proposed by the present invention contains both longitude and latitude information and height information, the 3D node search of the original A* algorithm can be transformed into a 2D node search problem, improving computational efficiency. In addition, the improved cost function of the present invention is tailored for digital elevation maps and has been experimentally verified to have high search efficiency on common digital elevation maps.

[0077] 6. When calculating the best insertion position of the emergency mission points into the existing mission sequence of the UAV using the formula, the present invention uses prior estimation to calculate the and In this way, it is not necessary to repeatedly call the improved A* algorithm for search and solution, which can effectively reduce the path finding time when inserting a large number of emergency missions and improve the real-time performance of the algorithm.

[0078] 7. The method of the present invention is applicable not only to unmanned helicopters, but also to other types of UAVs operating at low altitudes (near the ground), such as flapping-wing UAVs, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 is the overall flowchart of the present invention.

[0080] Figure 2 is a schematic diagram of the position for inserting the emergency mission points. (a) is a schematic diagram of inserting the mission point into the middle of the mission sequence, and (b) is a schematic diagram of inserting the mission point into the end of the mission sequence.

[0081] Figure 3 is a schematic diagram of the positional relationship between adjacent nodes.

[0082] Figure 4 is a schematic diagram of node cost calculation.

[0083] Figure 5 is a diagram of the UAV coordinates and the coordinates of the emergency missions after value sorting. The red circles in the figure are UAVs, and the black plus signs are emergency mission points.

[0084] Figure 6 is a diagram of the time-sequence path reconstruction process of three UAVs in dealing with emergency missions in the embodiment of the present invention. (a), (b), …, (i) are the time-sequence path reconstruction results when dealing with emergency missions 1, 2, …, 9 respectively.

[0085] Figure 7 is a result diagram of three UAVs dealing with 10 emergency missions in the embodiment of the present invention. (a) is a top view, and (b) is a side view.

[0086] Figure 8 is a time-consuming curve of three UAVs dealing with 10 different emergency missions in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0087] The present invention will be further described in detail below with reference to the drawings and embodiments.

[0088] Assume there are N UAVs, and the current mission sequence executed by UAV i is which are the coordinates of the 1st, 2nd, …, Lth i mission points to be executed by UAV i respectively, Denote the coordinates of the j-th task point to be executed by UAV i, including the longitude coordinate lon j , the latitude coordinate lat j , and the altitude coordinate hei j , L i is the task sequence T of UAV i i The current number of existing tasks in it, i is the index of the UAV, and i takes 1, 2,..., N respectively; the task sequence T currently executed by UAV i i The path list is Denote that when UAV i executes the task from the coordinate point to The set of path points including longitude, latitude, and altitude coordinates, Denote the path The straight-line length of, P i The path length of is Perform a discrete rasterization operation on the digital elevation model (DEM). The straight-line distance between any two points x i =(lon i , lat i , hei i ) and x j =(lon j , lat j , hei j ) is h(x i , x j ), and the calculation formula is k lon and k lat Represent the map resolutions in the longitude and latitude directions respectively.

[0089] As Figure 1 shown, when facing complex terrains and sudden tasks, the method for efficiently reconstructing paths for these N UAVs in the present invention includes the following steps:

[0090] Step 1: Evaluate the value of each sudden task point Ta in the set of sudden task points Task k to obtain the value val k of each sudden task point Ta k ;

[0091] Assume that the set of sudden task points Task = {Ta 1 , Ta 2 ,... Ta |S|}; for each sudden task point Ta in the set of sudden task points Task k ={lon k , lat k , hei k}, k = 1, 2... |S|, lon k , lat k , hei k are respectively the longitude coordinate, latitude coordinate and altitude coordinate of the emergency mission point Ta k . For each emergency mission point Ta k , its value val k is calculated using the following formula:

[0092]

[0093] In formula (1), the function J k is the mission demand of the emergency mission point Ta k . Taking a transport helicopter as an example, the mission demand can be understood as the weight of the transported materials; taking an armed helicopter as an example, the mission demand can be understood as the amount of ammunition for carrying out a strike mission at a certain mission point; N is the total number of UAVs; m i is the mass of UAV i; k i is the unit mass contribution of UAV i to any mission; X ik is a Boolean decision variable. If UAV i has selected the emergency mission point Ta k , then X ik = 1, otherwise X ik = 0. This decision variable is used to mark whether UAV i executes the emergency mission point Ta k ; ρ i (Ta k ) represents the shortest distance from the emergency mission point Ta k to all the mission points that UAV i has currently accepted. The calculation formula is as follows:

[0094]

[0095] Step 2: Sort the values val k of each emergency mission point Ta k obtained in Step 1 in descending order to obtain the value set Value;

[0096] Sort the values of each emergency mission point in the emergency mission point set Task in descending order to obtain the value set Value = (Val 1 , Val 2 ,... Val |S| ). The sorting rule is: First, sort in descending order according to the first element in the list of formula (1) Val k : If the first elements are the same (this may happen), then sort according to the second element: Sort it in descending order according to its size. The val calculated by formula (1) k is actually a list containing two elements, and these two elements jointly characterize the importance of the task. The difference is that the importance of the first element is stronger than that of the second element. For each sudden task point, calculate a set of such lists, and at the same time compare the first elements in the values of each sudden task point. The sudden task point with the larger first element has greater importance. If the first elements are the same, then compare the second elements in the values of each sudden task point. At this time, the task with the larger second element has greater importance.

[0097] Step 3: For the first item Val in the current value set Value 1 corresponding to the sudden task point Ta * , traverse the UAVs and calculate the optimal path and the change in cumulative distance corresponding to each UAV after inserting the sudden task point Ta into the current task sequence * of each UAV.

[0098] The position of inserting the sudden task point Ta * is shown as Figure 2 shown.

[0099] Initialize the index i of the UAVs to be traversed as i = 1. The specific execution process is as follows:

[0100] Step 3.1: Start to find the best insertion position j * of the UAV i for the sudden task point Ta * .

[0101] Step 3.2: Traverse the current task sequence of the UAV i to obtain the best insertion position j * of the sudden task point Ta * . The calculation formula is as follows:[[]]

[0102]

[0103] In formula (3), and respectively represent the optimal distance between the task point * and the sudden task point Ta considering the complex terrain factor and the threat degree factor compared with the straight-line distance between the task point * and the sudden task point Ta on the digital elevation map and the relative percentage increase, and the sudden task point Ta * and the task point Optimal distance between Compared with the sudden task point Ta on the digital elevation map * And the task point The straight-line distance between The increased relative percentage, which can be approximately considered to be only related to the terrain and threats at the current location, and has nothing to do with the specific task point coordinates; And The value of is based on the digital elevation map used in the specific implementation of the method of the present invention, and is determined through multiple experiments (experiments of taking the ratio of the optimal distance and the straight-line distance between two task points actually obtained by using the improved A* algorithm); And Are respectively the coordinates of the kth and the (k + 1)th task points in the task sequence T currently executed by the unmanned aerial vehicle i i Where k takes 1, 2,..., L i -1; Represents the optimal (shortest) distance between the sudden task point Ta * And the task point Considering complex terrain and threat environment; Represents the optimal (shortest) distance between the sudden task point Ta * And the task point Considering complex terrain and threat environment; And Are respectively And The prior estimates of, which are only used to obtain the position j of the best insertion point in step 3.2 * ; Represents the straight-line distance between the position of the kth task point In the task sequence of the unmanned aerial vehicle i on the digital elevation map and the position of the sudden task point Ta * ; Represents the straight-line distance between the position of the sudden task point Ta * On the digital elevation map and the position of the (k + 1)th task point In the task sequence of the unmanned aerial vehicle i; in the first iteration, that is, when i = 1, Also uses the prior estimate; when i > 1 Has been calculated by the improved A* algorithm in the previous task sequence and is a known value, which can be directly called. In addition, in formula (3), when k takes L i There will be a situation where a sudden task is inserted at the end of the task sequence T of the unmanned aerial vehicle i as shown in Figure 2 -(b), and at this time, in order to ensure that formula (3) is valid for 1 ≤ k ≤ L i At the end of the task sequence T of the unmanned aerial vehicle i, when a sudden task is inserted, in order to ensure that formula (3) is valid for 1 ≤ k ≤ Li The generality at this time is stipulated here Here is the assumed task point (which does not actually exist).

[0104] Step 3.3: Calculate the optimal path list P * after inserting the burst task point Ta into the task sequence currently being executed by the UAV i i * and the cumulative distance change Δd(P i );

[0105] Step 3.3.1: Use the improved A* algorithm to calculate the current best insertion position j * and the optimal path between the burst task point Ta * and the optimal distance The optimal distance between the burst task point Ta * and the position j * +1 and the optimal distance The optimal distance and the current best insertion position j * and the optimal distance between it and its next task point j * +1

[0106] The cost function of the node n estimated by the adopted improved A* algorithm is:

[0107] F n = ω G G n + ω H H n + ω I I n + ω c C n (4)

[0108] In formula (4), G n , H n , I n , C n respectively represent the cumulative map distance cost, heuristic cost, height cost and threat degree cost of the improved A* algorithm at the node n, and their initial values are all 0. The calculation schematic diagrams of these four costs are as Figure 4 shown, where n - 1 represents the previous adjacent node of the expanded node n, and so on, and n + 1 represents the previous adjacent node of the expanded node n + 2. Or, the adjacent node n is expanded from the node n - 1, then the adjacent node n + 1 is expanded from the node n, and then the adjacent node n + 2 is expanded from the node n + 1; ω G , ω H , ωI , ω C are the weight coefficients of the cumulative map distance cost, heuristic cost, height cost, and threat cost given artificially. Since G n , H n , I n , C n have different dimensions, the sum of the weight coefficients does not have to be 1; generally speaking, if you want to increase the success rate of finding the shortest path, you should increase ω G , but this will increase the pathfinding time; if you want to reduce the pathfinding time, you should increase the coefficient ω H of the heuristic cost term, but this may cause the algorithm to be difficult to find the global shortest path and easily fall into the local shortest path; if you want to increase the influence of the mountain height, you should increase the weight coefficient ω I , however, this will cause the final path to easily fall into the mountain valley bottom and increase the pathfinding time; if you want to increase the influence of the threat source, you should increase the weight coefficient ω C , however, this may cause additional tortuous path consumption and it is difficult to ensure that the final path is the shortest and optimal. In practical applications, you can select according to the needs and the pros and cons brought by the numerical values of each weight coefficient.

[0109] The iterative calculation formula for the cumulative map distance cost G neighbor(n) between node n and its neighbor neighbor(n) is:

[0110]

[0111] where r(n, neighbor(n)) represents the positional relationship between node n and its neighbor neighbor(n); the positional relationship between node neighbor(n) and node n is as Figure 3 shown. Taking 2 and 6 respectively represent moving left along the longitude direction and moving right along the longitude direction, taking 4 and 8 respectively represent moving up along the latitude direction and moving down along the latitude direction, and taking 1, 3, 5, 7 respectively represent moving in the south - west 45 - degree angle direction, north - west 45 - degree angle direction, north - east 45 - degree angle direction, and south - east 45 - degree angle direction; G n is the cumulative map distance cost at node n, and the value of G n at the initial node is 0; k lon and k lat represent the map resolutions along the longitude and latitude directions respectively.

[0112] The calculation formula for the heuristic cost H n at the current node n is:

[0113] H n = k lon |lon n - lon Ta |+ klat |lat n -lat Ta | (6)

[0114] Among them, lon n and lat n are the longitude and latitude coordinates of the current node n respectively; lon Ta and lat Ta are the longitude and latitude coordinates of the target point Ta respectively.

[0115] The height cost I at the current node n n is calculated by the formula:

[0116]

[0117] Among them, hei n is the terrain height of the current node n, hei min is the lowest altitude of the entire digital elevation terrain, hei max is the highest altitude of the entire digital elevation terrain, α hei is the equivalent altitude coefficient, and by default it takes

[0118] The threat degree cost C at the current node n n is calculated by the formula:

[0119]

[0120] Among them, when the threat source is a radar, C 0 is the intensity of the radar source, h(n, radar) is the straight-line distance from the current node n to the radar radar, and R(radar) represents the detection range of the radar.

[0121] Step 3.3.2: Calculate the cumulative distance change Δd(P * if the emergency task point Ta is inserted at the current best insertion position j * ), and the calculation formula is as follows: i ) is as follows:

[0122]

[0123] Among them, and are obtained through iterative search by the improved A* algorithm in Step 3.3.1; represents the actual shortest (optimal) path between the task point and the task point , which is calculated by using the improved A* algorithm in Step 3.1.1; when the emergency task point Ta * is inserted at the last task point After that, since the task point does not exist, so at this time, in formula (9), and are both 0.

[0124] Step 3.3.3: Calculate the optimal path list * of the unmanned aerial vehicle i after inserting the burst task point Ta into the current task sequence * executed by the unmanned aerial vehicle i at the current optimal insertion position j The calculation formula is as follows:

[0125]

[0126] In the formula, is the optimal path between the current optimal insertion position j * and the burst task point Ta * , which is obtained by step 3.3.1 through configuring the starting point coordinates and the ending point coordinates Ta * ; is the optimal path between the burst task point Ta * and the position j * + 1, which is obtained by step 3.3.1 through configuring the starting point coordinates Ta * , and the ending point coordinates ; is the optimal path between the task point and the current optimal insertion position j * , which is a known path and does not need to be reconstructed; is the optimal path between the position j * + 1 and the task point , which is a known path and does not need to be reconstructed; when the burst task point Ta * is inserted after the last task point , since the task point does not exist, so at this time, in formula (10), and are both empty sets.

[0127] Step 3.4: Determine whether all unmanned aerial vehicles have been traversed. If not, let i ← i + 1 and return to step 3.1; if so, enter step 4.

[0128] Step 4: Screen out the set F of unmanned aerial vehicles that meet the fuel constraint after inserting the burst task point Ta * from all the traversed unmanned aerial vehicles:

[0129] F = {i|△d(P i ) ≤ u i δi -d(P i )} (11)

[0130] In the formula, i is the index of the UAV that meets the fuel constraint; δ i is the flight distance per unit fuel consumption of UAV i; u i is the maximum fuel quantity of UAV i; d(P i ) is the original path list of UAV i before tentatively inserting the emergency mission point Ta * of the path length, where, represents the path the sum of the lengths of the line segments connecting all the path points in a straight line in, which can be described as: where, represents the set of path points including longitude, latitude, and altitude coordinates when UAV i executes the mission from the coordinate point to .

[0131] Step 5: Determine whether the set F of UAVs that meet the fuel constraint is empty. If F = φ, it means that the emergency mission does not meet the fuel constraint for all UAVs, and the emergency mission is not executed, then go to Step 8; otherwise, go to Step 6.

[0132] Step 6: Select the UAV that can best ensure the timing synchronization of all UAVs from the set F of UAVs that meet the fuel constraint as the UAV i * finally assigned to the emergency mission point Ta * , and the calculation formula is as follows:

[0133]

[0134] In the formula, i * represents the index of the UAV assigned to the emergency mission point Ta * ; is the average flight speed of UAV i; is the average flight speed of UAV j; d(P i ) is the path length of the original path list of UAV i before tentatively inserting the emergency mission point Ta * ; d(P ) is the path length of the original path list of UAV j before tentatively inserting the emergency mission point Ta j ; * of the path length.

[0135] The most suitable UAV i * to execute the emergency mission point Ta can be selected through Equation (12)​​* such that for the burst task point Ta * inserted into the task sequence of UAV i * for the UAV swarm, the difference in flight time between any two UAVs is as small as possible, that is, the difference between the two UAVs with the largest flight time difference should be as small as possible.

[0136] Step 7: Assign the burst task point Ta * to UAV i * and update the path list and the shortest flight length of the task sequence that UAV i * currently needs to execute;

[0137] For the burst task point Ta * after finding the optimal UAV i * and its optimal insertion position, assign the burst task point Ta * to UAV i * and update (that is, the optimal path list obtained after inserting the burst task point Ta * into the path list of the task sequence currently executed by UAV i * is updated to the new path list that UAV i currently needs to execute), * (that is, the new shortest flight length obtained after inserting the burst task point Ta into the path list of the task sequence currently executed by UAV i * is updated to the shortest flight length of the new path list that UAV i * currently needs to execute). Update the new path list that UAV i * currently needs to execute to the new shortest flight length).

[0138] Step 8: Update the value set;

[0139] Pop the first item Val from the value set Value 1 to get Value ← Value - Val 1 (that is, delete the first item Val in the current value set, update the value set, and use the original second item value as the first item value in the current value set). 1 Delete the first item Val in the current value set, update the value set, and use the original second item value as the first item value in the current value set).

[0140] Step 9: Determine whether the current value set Value is empty. If Value = φ, output the path list P i of the N UAVs that need to execute tasks at the initial moment, i = 1, 2... N, and the process ends; otherwise, return to Step 3.

[0141] Example:

[0142] Preparation: Perform a discrete rasterization operation on the Digital Elevation Map (DEM) for subsequent path replanning using the improved A* algorithm. For example, discretize the map within the range of 128.00°E - 128.32°E in longitude and 44.68°N - 45.00°N in latitude into a 100×100 raster map. The resolution k lon along the longitude direction of the map is 352 m, and the resolution along the latitude direction is k lat which is 362 m. Here, the number of UAVs N for performing the task is 5, the number of emergency tasks to be performed is 10, and the position coordinates of the emergency tasks are shown in Table 1 below. Assume that the average flight speed of UAV i (i = 1, 2... N) is 25 m / s, the mass m i is 5092 kg for all, the maximum range is 482 km for all, the maximum endurance time is 3 hours and 9 minutes for all, and the fuel quantity u i carried by UAV i (i = 1, 2... N) is 1500 kg. Then the flight distance δ i per unit fuel consumption is 0.3213 km / kg. The position of the threat source is (128.3°E, 45°N), and the threat source is a radar with a threat intensity C 0 of 15000 m 4 .

[0143] Table 1 Longitude and latitude coordinates of emergency task points

[0144]

[0145] The specific process of this embodiment is as follows:

[0146] Step 1: Calculate the value val k of each emergency task point, and the calculation formula is

[0147] Step 2: Sort the values of each emergency task point in descending order to obtain the value set Value = (Val 1 , Val 2 ,... Val 10 ). After sorting in descending order, the priority order of the coordinates of the emergency task points is as Figure 5 shown. The red circles in the figure are UAVs, and the black plus signs are emergency task points. The smaller the number corresponding to the black plus sign, the higher the priority of the task corresponding to the plus sign.

[0148] Step 3: For the emergency task point Ta 1 corresponding to the first item Val * in the value set Value, traverse the UAVs and calculate that if in the task sequence of each UAV i Execute the sudden task point Ta * After the insertion operation, the optimal path and the cumulative distance change amount corresponding to each UAV.

[0149] Initialize the UAV index i = 1 for subsequent traversal. The specific execution process is as follows:

[0150] Step 3.1: Start to find the best insertion position j of UAV i for the sudden task point Ta * * .

[0151] Step 3.2: Traverse the task sequence T currently executed by UAV i i , and obtain the best insertion position j of the current sudden task point Ta * in the task sequence , and the calculation formula is as follows: *

[0152]

[0153] In the formula, represents the optimal distance between the task point Ta * and considering the complex terrain and threat environment; represents the optimal distance between the task point Ta * and considering the complex terrain and threat environment; represents the optimal distance between the task points and considering the complex terrain and threat environment.

[0154] Here, the following formula is used to calculate and

[0155]

[0156] Among them, are the longitude, latitude and altitude coordinates of UAV i at the kth task point respectively; are the longitude, latitude and altitude coordinates of UAV i at the (k + 1)th task point respectively; are the longitude, latitude and altitude coordinates of the sudden task Ta * respectively; determined through multiple experiments based on the digital elevation map used in this embodiment, the value of is between 0.1 - 0.3, and here take ​​Used to roughly estimate the relative increase in the optimal distance between two mission points considering complex terrain factors and threat level factors with respect to the straight-line distance between the two mission points on the digital elevation map.

[0157] Step 3.3: Calculate if the sudden mission point Ta * is inserted into the mission sequence of UAV i the resulting optimal path list and the cumulative distance change Δd(P i );

[0158] Step 3.3.1: Taking the best insertion position j * as the starting point and the sudden mission point Ta * as the ending point, use the improved A* algorithm to calculate the optimal path * between the best insertion position j * and the sudden mission point Ta optimal distance and the optimal path * between the sudden mission point Ta * and position j + 1 optimal distance

[0159] The following takes the specific process of obtaining and as an example for illustration:

[0160] Sub-step 1): Construct an OPEN table to store the nodes to be expanded, and construct a CLOSED table to store the explored nodes. To avoid duplication with the previous UAV set label F, Fu is used to represent the cost function F. Here, the cost function Fu of each node is calculated using the aforementioned formula (4), and the initial value of this cost function is set to 0. Initialize and include the mission nodes ( representing the coordinates of the j * th mission point to be executed by UAV i) into the OPEN table, and the CLOSED table is empty.

[0161] Sub-step 2): Pop the node n with the minimum cost function Fu value from the OPEN table, and at the same time put this node n into the CLOSED table, and execute Sub-step 3); if the CLOSED table has expanded to the end mission Ta * , then execute Sub-step 5).

[0162] Sub-step 3): Use the following formula to calculate the cost values corresponding to the 8 adjacent nodes neighbor(n) of node n (relative positions as Figure 3 shown):

[0163] Funeighbor(n) = ω G G neighbor(n) + ω H H neighbor(n) + ω I I neighbor(n) + ω c C neighbor(n) (16)

[0164] Wherein, G neighbor(n) , H neighbor(n) , I neighbor(n) , C neighbor(n) respectively represent the cumulative map distance cost, heuristic cost, height cost, and threat cost when the improved A* algorithm is at the neighbor node neighbor(n) of node n. The calculation schematic diagrams of these four costs are as shown in Figure 4 shown; ω G , ω H , ω I , ω C are the weight coefficients of the cumulative map distance cost, heuristic cost, height cost, and threat cost respectively. Here, the weight coefficient ω G = 0.4, ω H = 0.5, ω I = 10, ω c = 10.

[0165] The cumulative map distance cost G neighbor(n) between node n and its neighbor node neighbor(n) is calculated as follows:

[0166]

[0167] Among them, r(neighbor(n), n) represents the positional relationship between node n and its neighbor node neighbor(n). Taking 2 and 6 respectively represents moving left along the longitude direction and moving right along the longitude direction. Taking 4 and 8 respectively represents moving up along the latitude direction and moving down along the latitude direction. Taking 1, 3, 5, and 7 respectively represents moving in the south-west 45-degree angle direction, north-west 45-degree angle direction, north-east 45-degree angle direction, and south-east 45-degree angle direction.

[0168] The heuristic cost H neighbor(n) of the neighbor node neighbor(n) of node n is calculated as follows:

[0169] H neighbor(n) = k lon |lon neighbor(n) - lon Ta | + k lat |lat neighbor(n) - lat Ta | (18)

[0170] Among them, lon neighbor(n) and lat neighbor(n) are the longitude and latitude coordinates of the neighbor node neighbor(n), and lon Ta and lat Ta are the longitude and latitude coordinates of the target point Ta.

[0171] The height cost I of the neighbor node neighbor(n) of node n neighbor(n) is calculated by the formula:

[0172]

[0173] Among them, hei neighbor(n) is the terrain height of the neighbor node neighbor(n), hei min is the lowest altitude of the entire digital elevation terrain, hei max is the highest altitude of the entire digital elevation terrain, α hei is the equivalent altitude coefficient, and by default, it is taken as

[0174] The threat cost C of the neighbor node neighbor(n) of node n neighbor(n) is calculated by the formula:

[0175]

[0176] Among them, in this embodiment, the situation with radar threat is considered, so the threat intensity C 0 is the intensity of the radar source; generally speaking, the closer to the radar source, the higher the threat level to the UAV, and the threat level is inversely proportional to the fourth power of the straight-line distance from the radar source. When exceeding a certain distance, it can be considered that the radar has no influence on the UAV. h(neighbor(n), radar) is the straight-line distance from the neighbor node neighbor(n) to the radar radar, and R(radar) represents the detection range of the radar.

[0177] Sub-step 4): If the current neighbor node neighbor(n) already exists in the OPEN list, or the current neighbor node neighbor(n) already exists in the CLOSED list, record the cost function values of this neighbor node neighbor(n) in the OPEN list and the CLOSED list as At this time, compare the cost function value Fu of this neighbor node calculated by the above formula (16) neighbor(n) with If a certain neighbor node satisfies then re - substitute the cost function value of this neighbor node neighbor(n) with Fu neighbor(n) , and go to sub - step 2).

[0178] Sub-step 5): Backtrack from the emergency task point Ta * and search backward for forward nodes to obtain the optimal path and calculate the optimal cost function value, i.e., the optimal distance

[0179] Optimal path and optimal distance are obtained in the same way as and and will not be elaborated here.

[0180] Step 3.3.2: On the basis of obtaining the optimal distance and calculate the cumulative distance change amount △d(P * ) that will be generated if the emergency task point Ta * is inserted at the current best insertion position j i of the UAV i. The calculation formula is as follows:

[0181]

[0182] Step 3.3.3: Calculate the data of the optimal path list of the UAV i if the emergency task point Ta * is inserted into the current task sequence executed by the UAV i * . The calculation formula is as follows: The calculation formula is as follows:

[0183]

[0184] Step 3.4: Determine whether all UAVs have been traversed. If not, let i←i + 1 and return to Step 3.1; if so, go to Step 4.

[0185] Step 4: Screen out the set F of UAVs that meet the fuel constraint after inserting the emergency task point Ta * from all the traversed UAVs:

[0186] F = {i|△d(P i ) ≤ u i δ i - d(P i )} (23)

[0187] In the formula, u i δ i is the total flight range of the UAV at the current fuel level, approximately equal to 482 km, and d(P i ) is the total distance of the UAV i under the current path P i .

[0188] Step 5: Determine whether the set F of UAVs that meet the fuel constraint is empty. If F = φ, it indicates that no UAV can execute the emergency mission point Ta * , go to Step 8; otherwise, go to Step 6.

[0189] Step 6: Select the UAV from the set F of UAVs that meet the fuel constraint that can best ensure the timing synchronization of all UAVs as the UAV i * finally assigned to the emergency mission point Ta * , and the calculation formula is as follows:

[0190]

[0191] In the formula, i * is the index of the UAV assigned to the emergency mission point Ta * ; is the average flight speed of UAV i; is the average flight speed of UAV j; d(P i ) is the path length of the original path list of UAV i before tentatively inserting the emergency mission point Ta * ; d(P j ) is the path length of the original path list of UAV j before tentatively inserting the emergency mission point Ta * ;

[0192] Step 7: Arrange the emergency mission point Ta * for UAV i * , and update

[0193] Step 8: Update the value set Value, and pop the first item Val 1 from the value set Value to get Value ← Value - Val 1 .

[0194] Step 9: Determine whether the current value set Value is empty. If Value = φ, the algorithm ends, and the path lists P of all N UAVs are output i , i = 1, 2... N; otherwise, return to Step 3.

[0195] Through the above Steps 1 - 8, using the emergency missions 1 - 10 to perform successive task insertions on the existing tasks and path sequences of UAVs 1 - 3, the dynamic path reconstruction results obtained step by step are as Figure 6 shown (the red circles in the figure are UAVs, and the black plus signs are emergency mission points), where the reconstructed path of the last emergency mission 10 is shown in Figure 7 ​In the (a) figure (the red circles in the figure are drones, and the black plus signs are the emergency task points). It can be seen from the figure that when 10 emergency tasks appear simultaneously, the algorithm calculates the descending order of their values according to the attributes such as the importance of each emergency task and the location of each emergency task as emergency tasks 1, 2,..., 10, and performs a tentative insertion operation on the task sequences of all current drones in this order. And all drones will measure the value of the task and calculate the optimal drone and its optimal task point interval that meet the constraints for subsequent insertion operations. It should be noted that in the process from Figure 6 the (c) figure to Figure 6 the (d) figure, when the No. 2 drone processes the task sequence insertion operation of emergency task 4, it realizes the efficient reconstruction of the original task path. It should be noted that this process takes less than 3s and has high timeliness.

[0196] The top view and side view of the final path reconstruction result are as shown in Figure 7 shown, where the time curve for each emergency task insertion is as shown in Figure 8 shown. From Figure 7 and Figure 8 it can be seen that for 10 emergency tasks, in the case of 3 drones performing tasks, the total time for each time to solve the optimal drone, the best insertion position, and the reconstructed path corresponding to the emergency task is about 3s on average. And the reconstructed paths of all drones finally also have good coordination under the constraints of terrain, fuel, and time, etc.

[0197] The flight times corresponding to the three drones are shown in Table 2.

[0198] Table 2 Flight times corresponding to the three drones

[0199]

[0200] It can be seen from Table 2 that the total flight times of the three drones are all about 100s. The difference in the total flight times between drone 1 and drone 2 is the largest, which is 19s, and the difference between drone 1 and drone 3 is the smallest, about 8s. And the proportion of the difference in the maximum total flight time to the longest flight time of the drone is about 16.3%. This shows that for a large number of drones performing emergency tasks, on the basis of ensuring the local optimal reconstructed path and high calculation efficiency, the drones have realized the coordination of the flight times of multiple drones as much as possible, and this coordination makes the total task completion time of all drones reach the shortest 116.3695s.

Claims

1. An efficient multi-UAV path reconstruction method for complex terrain and emergency tasks, characterized by: The following steps are involved: Step 1: Perform value evaluation on each burst task point in the burst task point set to obtain the value of each burst task point; Step 2: Arrange the values ​​in descending order to obtain a value set; Step 3: For the burst task point Ta corresponding to the first value in the current value set * , based on the improved A* algorithm, if the task sequence T currently executed by each UAV i is calculated i Insert a burst task point Ta * After that, the optimal path list P corresponding to each UAV i i * and the cumulative distance change △d(P i ); i is 1, 2, ..., N respectively; N is the total number of drones performing tasks at the initial moment; the cost function of the improved A* algorithm is designed based on the digital elevation map, including longitude, latitude and altitude information, and converts the three-dimensional node search of the original A* algorithm into a two-dimensional node search; Step 4: Filter out the insertion burst task point Ta * The set of drones that still meet the fuel constraints is F: F={i|△d(P i )≤u i δ i -d(P i )}; Step 5: Determine whether the drone set F is empty. If so, the current emergency task point Ta * If not, go to step 6; Step 6: Select drone i from the drone set F * , meet the emergency task point Ta * Insert drone i * After the task sequence, the flight time difference between any two of the N UAVs is as small as possible; Step 7: Set the current emergency task point Ta * Arrange for drone i * , Update UAV i * The path list and the shortest flight length of the task sequence currently to be executed; Step 8: Delete the first value in the current value set; Step 9: Determine whether the current value set is empty. If so, output the path list of N drones and the process ends; Otherwise, return to step 3.

2. The method for efficient multi-UAV path reconstruction for complex terrain and emergency tasks according to claim 1 is characterized in that: The value of each emergency task point in step 1 is calculated according to the following formula: In the formula, val k Indicates the burst task point Ta k The value of; k = 1, 2...|S|; |S| is the total number of burst tasks; function J k For the emergency task point Ta k The task requirements; m i is the mass of UAV i; k i is the unit mass contribution of UAV i to any task; X ik is a Boolean decision variable. If UAV i has selected the emergency task point Ta k , then X ik =1, otherwise X ik =0; ρ i (Ta k ) represents the burst task point Ta k The shortest distance to all mission points currently accepted by drone i, Indicates the task point The location and the emergency task point Ta k The straight-line distance of the location on the digital elevation map; The mission sequence currently being executed by UAV i The jth task in .

3. The efficient multi-UAV path reconstruction method for complex terrain and emergency tasks according to claim 2 is characterized in that: The principle of descending order in step 2 is: first sort by Val k The first element in the list If the first elements are the same size, they are sorted in descending order according to the second element. Sort by size in descending order.

4. The method for efficient multi-UAV path reconstruction for complex terrain and emergency tasks according to claim 3 is characterized in that: The cost function of the improved A* algorithm described in step 3 is: F n =ω G G n +oh H H n +oh I I n +oh c C n ; H n =k lon |lon n -lon Ta |+k lat |years n -years Ta |; In the formula, G n ,H n ,I n ,C n They are respectively the accumulated map distance cost, heuristic cost, height cost and threat cost of the improved A* algorithm at node n, and their initial values ​​are all 0; ω G ,ω H ,ω I ,ω C G n ,H n ,I n ,C n The weight coefficient of G neighbor(n) is the cumulative map distance cost of the improved A* algorithm at the neighboring nodes of node n; k lon and k lat Represents the map resolution along longitude and latitude respectively; lon n and lat n are the longitude and latitude coordinates of the current node n; lon Ta and lat Ta are the longitude and latitude coordinates of the target point Ta respectively; hei n is the terrain height of the current node n, hei min is the lowest altitude of the entire digital elevation terrain, hei max is the highest altitude of the entire digital elevation terrain, α hei is the equivalent altitude coefficient, C0 is the intensity of the threat source, h(n, radar) is the straight-line distance from the current node n to the threat source, and R(radar) represents the influence range of the threat source.

5. The method for efficient multi-UAV path reconstruction for complex terrain and emergency tasks according to claim 4 is characterized in that: Step 3 is as follows: Step 3.1: Initialize the index i=1 of the drone to be traversed; Step 3.2: Traverse the mission sequence currently executed by drone i The burst task point Ta is calculated using the following formula: * The best insertion position j * : In the formula, Considering the complex terrain and threat factors, the mission point With the sudden task point Ta * The optimal distance between Compared to distance the percentage increase; They represent the sudden mission points Ta when considering complex terrain factors and threat factors. * With mission points The optimal distance between Compared to distance the percentage increase; and They are and A priori estimate of ; represents the kth mission point in the mission sequence of UAV i on the digital elevation map The location and emergency task point Ta * The straight-line distance between the locations; Represents the sudden task point Ta on the digital elevation map * The position of and the k+1th task point in the task sequence of UAV i The straight-line distance between the locations; The first iteration, when i=1, Adopt and The same prior estimation method is used for calculation; when i>1 It has been calculated by the improved A* algorithm in the previous task sequence and is a known value, so it can be called directly; k=L i hour, is the assumed mission point; Step 3.3: Calculate the burst task point Ta * Insert the mission sequence currently executed by drone i The optimal path list P i * and the cumulative distance change △d(P i ); Step 3.3.1: Use the improved A* algorithm to calculate the current best insertion position j * and the sudden task point Ta * The best path between Optimal distance Emergency Task Point Ta * and position j * +1 Best path between Optimal distance and the current best insertion position j * Its next task point * Optimal distance between +1 Step 3.3.2: Calculate the best insertion position j for drone i * If we insert a burst task point Ta * The cumulative distance change △d(P i ): In the formula, when the burst task point Ta * It is inserted at the last task point. Afterwards, and All are 0; Step 3.3.3: Calculate the best insertion position j at the current time * If the task sequence currently executed by UAV i is Insert a burst task point Ta * After that, the optimal path list P of UAV i i * , the calculation formula is as follows: In the formula, when the burst task point Ta * It is inserted at the last task point. Afterwards, and are all empty sets; Step 3.4: Determine whether all drones have been traversed. If not, set i←i+1 and return to step 3.1; if so, go to step 4.

6. The method for efficient multi-UAV path reconstruction for complex terrain and emergency tasks according to claim 5 is characterized in that: Step 6 Drone i * Determine using the following formula: In the formula, is the average flight speed of UAV i; is the average flight speed of UAV j; d(P i ) is the point where UAV i tentatively inserts a burst task point Ta * The original path list before The path length; d(P j ) is the point where UAV j tentatively inserts a burst task point Ta * The previous list of existing paths The path length.

7. The method for efficient multi-UAV path reconstruction for complex terrain and emergency tasks according to claim 6 is characterized in that: Step 7 is as follows: * Insert drone i * The new path list obtained after the path list of the currently executed task sequence Updated to dronei * The new path list that needs to be executed currently, the burst task point Ta * Insert drone i * The new shortest flight length obtained after the path list of the currently executed task sequence Updated to dronei * The shortest flight length of the current list of new paths that need to be executed.

Images