Post-disaster city full coverage detection method based on heterogeneous unmanned system

Abstract

The application discloses a kind of post-disaster city full coverage detection methods based on heterogeneous unmanned system, belong to rescue technical field, including: unmanned aerial vehicle and unmanned vehicle are as heterogeneous unmanned system, construct heterogeneous unmanned system full coverage path planning model;Discretization is carried out to heterogeneous unmanned system full coverage path planning model, and the discrete optimization model of heterogeneous unmanned system full coverage path planning is obtained;The switch state of path point in discrete optimization model is optimized using improved gene algorithm, and the optimal open point set is obtained, the shortest full coverage detection path is obtained by traversing the optimal open point set using lawnmower algorithm;Construct charging point dynamic adjustment mechanism, based on shortest full coverage detection path and charging point dynamic adjustment mechanism carries out full coverage detection.The application effectively improves the global search capability and convergence speed of gene algorithm, designs the dynamic adjustment strategy of two-way meeting charging point, effectively reduces charging waiting time.

Description

Technical Field

[0001] The present invention belongs to the field of rescue technology, and in particular relates to a post-disaster city full coverage detection method based on a heterogeneous unmanned system. Background Art

[0002] In the field of post-disaster rescue, achieving comprehensive coverage of the affected area is a key prerequisite for rescue decision-making. After a disaster, timely and comprehensive understanding of the distribution of people, road damage, collapsed buildings, and other conditions in the disaster area is crucial for developing scientific and reasonable rescue plans. Full coverage detection ensures that rescuers do not miss any potential trapped areas, thereby improving rescue efficiency and minimizing casualties and property losses. In this context, studying the problem of full coverage detection path planning is of great significance. Traditional manual or single-robot detection methods have problems such as coverage blind spots and low efficiency in complex and dynamic environments, making it difficult to achieve comprehensive and rapid information collection.

[0003] In recent years, the rapid development of drone technology has made it an ideal tool for post-disaster information collection. With their flexible maneuverability and efficient aerial reconnaissance capabilities, drones can quickly conduct full coverage of disaster areas and promptly locate the positions of trapped people and road damage. The successful application of drones in power inspections, high-rise building inspections, fire rescue, and exploration and mapping missions has verified their multifunctional advantages. Nowadays, various research contents aim to enhance the intelligence, autonomy, and adaptability of drones to cope with the complex and changing needs and challenges in post-disaster scenarios. For example, by combining deep learning, drones can achieve autonomous flight and target recognition. By adding multi-sensor fusion technology, drones can achieve environmental perception and obstacle avoidance. These studies have provided more possibilities for the application of drones in post-disaster rescue.

[0004] However, the limited endurance of drones limits their ability to sustain operations. Unmanned vehicles, as ground-based mobile platforms, can provide charging services for drones, effectively resolving the endurance issue. The two work together to form a heterogeneous unmanned system, with the drone responsible for aerial reconnaissance and positioning, and the unmanned vehicle responsible for drone endurance, providing continuous and accurate data support for rescue deployment. Therefore, a full-coverage detection path planning method for heterogeneous unmanned systems in post-disaster urban environments is designed. By leveraging the advantages of global optimization and local refinement of an improved memetic algorithm, efficient coverage of the entire disaster area is achieved. The goal is to find the shortest paths for drones and unmanned vehicles to achieve full coverage and access charging points, respectively, ensuring comprehensive collection of disaster information and improving post-disaster search and rescue efficiency and response speed. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention proposes a post-disaster city full coverage detection method based on a heterogeneous unmanned system to solve the problems existing in the above-mentioned existing technologies.

[0006] To achieve the above objectives, the present invention provides a method for full coverage detection of cities after disasters based on heterogeneous unmanned systems, comprising:

[0007] Considering drones and unmanned vehicles as heterogeneous unmanned systems, a full-coverage path planning model for heterogeneous unmanned systems is constructed;

[0008] Discretizing the heterogeneous unmanned system full coverage path planning model to obtain a discrete optimization model for the heterogeneous unmanned system full coverage path planning;

[0009] An improved memetic algorithm is used to optimize the switch states of path points in the discrete optimization model to obtain the optimal open point set, and a lawnmower algorithm is used to traverse the optimal open point set to obtain the shortest full coverage detection path;

[0010] A dynamic adjustment mechanism for charging points is constructed, and full coverage detection is performed based on the shortest full coverage detection path and the dynamic adjustment mechanism for charging points.

[0011] Optionally, the objective function of the heterogeneous unmanned system full coverage path planning model is:

[0012] ;

[0013] Where, The time consumed by the i-th UAV to perform the coverage mission, is the charging time of the i-th drone, is the hovering waiting time of the i-th UAV.

[0014] Optionally, the constraints of the heterogeneous unmanned system full coverage path planning model include power constraints, kinematic constraints, communication constraints, mission constraints, safety constraints, environmental constraints, and collaborative constraints.

[0015] Optionally, the low-altitude airspace in the post-disaster city is divided into countless cubic discrete points to form a three-dimensional position array, where each element in the array represents a discrete path point; these discrete path points are encoded, and the path planning problems of drones and unmanned vehicles are discretized to obtain a discrete optimization model for full-coverage path planning of heterogeneous unmanned systems.

[0016] Optionally, the process of optimizing the on-off state of the path points in the discrete optimization model by using the improved meme algorithm comprises: randomly generating an initial population and repairing, evaluating the fitness of individuals in the population, screening individuals with fitness meeting the requirements as parents through selection operation, generating offspring through crossover and mutation operation by using dynamic self-adaptive crossover rate and mutation rate, and optimizing the offspring through 2-opt local search; when the number of iterations reaches half, introducing evolutionary comparison operation, repeatedly updating the population until the termination condition is met, and obtaining the optimal open point set.

[0017] Optionally, the process of repairing the initial population comprises: assuming that all path points are open points, randomly selecting an open point; converting the state of the selected open point to closed, checking whether the remaining open points can achieve complete coverage, if the complete coverage can be achieved, closing the selected open point, otherwise, keeping the selected open point open; randomly selecting the next open point which has not been closed to repeat the above operation until all open points have been closed, and outputting a feasible solution.

[0018] Optionally, the relative diversity index is obtained by calculating the ratio of the standard deviation to the average of the fitness value, if the diversity index is lower than the threshold value, the crossover rate is reduced and the mutation rate is increased, if the diversity index is higher than the threshold value, the high crossover rate and the low mutation rate are maintained.

[0019] Optionally, when the power of the unmanned aerial vehicle is reduced to a threshold value, the dynamic adjustment mechanism of the charging point is started, the unmanned aerial vehicle dynamically adjusts the position of the charging point according to the current power and the obstacle condition between the unmanned vehicle, when there is an obstacle on the straight line path between the unmanned aerial vehicle and the unmanned vehicle, if the remaining power of the unmanned aerial vehicle is sufficient to fly over the obstacle, the position after flying over the obstacle is taken as a new charging point, if the remaining power is insufficient or there is no obstacle on the path, the unmanned aerial vehicle flies to a low-power hovering protection position in the direction of the unmanned vehicle, and the corresponding position is taken as a new charging point.

[0020] The application further provides a computer device, comprising a memory, a processor to store a computer program on the memory and run the computer program on the processor, and the processor executes the computer program to realize the steps of the above method.

[0021] The application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the above method.

[0022] Compared with the prior art, the application has the following advantages and technical effects:

[0023] (1) The application can perform large-scale full coverage detection tasks in complex urban environments, and introduce a mobile charging operation of an unmanned vehicle for the problem of insufficient endurance of the unmanned aerial vehicle, so as to ensure that the unmanned aerial vehicle can complete long-time and large-scale scene coverage.

[0024] (2) The method used in the present invention has a wide range of applications and can perform full coverage detection path planning for scenes of different sizes.

[0025] (3) The path planning quality obtained by the method used in the present invention is better. The improved meme algorithm effectively enhances the search efficiency of paths in large and complex environments by introducing dynamic parameter adjustment and 2-opt local optimization, and significantly improves the quality and convergence speed of path planning.

[0026] (4) The method used in the present invention has higher flexibility. A dynamic charging point adjustment mechanism based on obstacles and the remaining power of the UAV is designed. By optimizing the collaborative path of the UAV and the unmanned vehicle, the charging waiting time is significantly reduced, ensuring that the mission completion time is minimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0028] Figure 1 This is a schematic diagram of a full coverage scenario of drone and unmanned vehicle collaboration according to an embodiment of the present invention;

[0029] Figure 2 This is a schematic diagram of the kinematic constraints of a UAV according to an embodiment of the present invention;

[0030] Figure 3 A schematic diagram of kinematic constraints for an unmanned vehicle according to an embodiment of the present invention;

[0031] Figure 4 This is a schematic diagram of the coverage range of a drone according to an embodiment of the present invention when the flight altitude is 1;

[0032] Figure 5 This is a schematic diagram of the coverage range of the drone according to an embodiment of the present invention when the flight altitude is 21;

[0033] Figure 6 This is a flowchart of solving a full coverage detection path according to an embodiment of the present invention;

[0034] Figure 7 2-opt local search schematic diagram of an embodiment of the present invention;

[0035] Figure 8 A schematic diagram of a lawn mower algorithm according to an embodiment of the present invention;

[0036] Figure 9 This is a schematic diagram of a drone charging point update scenario according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0038] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0039] Example 1

[0040] This embodiment provides a method for full coverage detection of a city after a disaster based on a heterogeneous unmanned system, including:

[0041] 1. Description of the problem and environment;

[0042] 1.1. Environment Description:

[0043] In order to effectively solve the technical problems faced, without loss of generality, the following assumptions are made:

[0044] 1. The power of all unmanned vehicles is sufficient for the entire rescue operation;

[0045] 2. Know the maximum flight distance of the drone;

[0046] 3. The speed of each robot is constant during its driving process;

[0047] 4. The drone needs to retain some power to hover and wait for the unmanned vehicle to arrive for charging;

[0048] Based on the above assumptions, the problem to be solved in this embodiment is as follows: in a post-disaster urban scenario, drones and unmanned vehicles are used in a collaborative manner to perform tasks, where the drone performs full-coverage detection tasks in a designated area, and the unmanned vehicle plans the charging path according to the drone's flight route; the drone is responsible for locating the affected people in the target area and detecting road accessibility, and must cover all grid areas except buildings and no-fly zones. In key areas, the flight altitude must be lowered to obtain detailed information; when the drone's battery power drops below a certain level during the mission, it will descend and hover at the current position, and use this position as the charging point. The unmanned vehicle goes to the charging point to provide charging services for the drone, thereby optimizing the collaborative full-coverage path of the drone and the unmanned vehicle, enabling the drone to complete the full-coverage detection mission in the shortest time and improve rescue efficiency.

[0049] 1.2, scene description;

[0050] Figure 1This is a schematic diagram of a scenario where drones and unmanned vehicles collaborate to provide full coverage. As can be seen from the figure, buildings of different heights are set up in the urban scene, and the ground area is represented by a two-dimensional grid. The dark gray path in the air represents the drone coverage path, and the light gray path on the ground represents the unmanned vehicle path. At the same time, three special types of two-dimensional grids are introduced in the scene to be close to the constraints that may exist in real scenarios.

[0051] (1) Drone-restricted zones (light gray grids): Drones are strictly prohibited from entering these grids (lakes, military management areas, etc.). Drones must bypass these grids to perform coverage tasks.

[0052] (2) Unmanned vehicle prohibited areas (black grids): grids occupied by collapsed buildings or obstacles. Unmanned vehicles must bypass the obstacles and are strictly prohibited from entering such grids.

[0053] (3) Key areas (dark gray grids): This type of coverage grid (shelters, schools, parks, etc.) has special requirements for shooting resolution, that is, the drone must lower its flight altitude to achieve effective coverage.

[0054] In addition, UAVs and unmanned vehicles are also subject to constraints such as kinematics, endurance, and communication. In order to model the above-mentioned heterogeneous unmanned system collaborative full coverage path planning problem in detail, Table 1 gives the relevant symbols and their definitions.

[0055] Table 1

[0056]

[0057] 2. Model construction;

[0058] Taking into account the particularity of the rescue problem, the goal to be optimized in this embodiment is the completion time of the full coverage detection mission, and the total completion time is composed of the drone coverage flight time, charging time and hovering waiting time; optimizing the total mission completion time is to optimize the three times separately. Using the above information, a mathematical model of the above three goals is given.

[0059] 2.1、Full coverage flight time;

[0060] Based on the above analysis, the flight time of each drone covering its own area is given. The flight speed of each drone is a constant flight speed. The full coverage flight time of the drone is the area of ​​the area covered by each drone divided by the drone's flight speed. The specific calculation formula is shown in formula (1).

[0061] (1)

[0062] in, is the full coverage flight time of the i-th UAV, is the area covered by the i-th drone, is the flight speed of the drone.

[0063] 2.2 Charging time;

[0064] The charging time of a drone can be calculated by the ratio of its remaining power to the charging rate. Assuming that the maximum power of each drone is 100 units, the drone's power is gradually consumed as time goes by during flight. When charging, the charging time of the drone is equal to the current remaining power divided by the charging rate. Based on the above analysis, the mathematical expression of the charging time is given as shown in formula (2):

[0065] (2)

[0066] in, is the charging time of the i-th drone, is the remaining power of the i-th drone, The charging rate of the drone.

[0067] 2.3. Hover waiting time;

[0068] In this embodiment, the hovering waiting time is the time the drone hovers in place after reaching the low battery threshold, waiting for the unmanned vehicle to arrive. The time required for the unmanned vehicle to travel from the current position to the drone charging point is calculated based on the three-dimensional spatial distance between the drone's current hovering position and the unmanned vehicle and the unmanned vehicle's driving speed, as shown in formula (3):

[0069] (3)

[0070] in, is the charging hovering waiting time of the i-th drone, is the distance between the i-th UAV and the j-th UAV, The speed of the unmanned vehicle.

[0071] 2.4, mathematical model;

[0072] From the above analysis, it can be seen that the completion time of the full coverage detection task is composed of the full coverage flight time, charging time and hovering waiting time. Since the detection work of each drone is carried out in parallel, the total task completion time is the maximum value of the flight time of all drones. Based on this, the objective function of the detection problem of this embodiment is given, as shown in formula (4):

[0073] (4)

[0074] st

[0075] (5)

[0076] (6)

[0077] (7)

[0078] (8)

[0079] (9)

[0080] (10)

[0081] (11)

[0082] (12)

[0083] (13)

[0084] (14)

[0085] (15)

[0086] (16)

[0087] (17)

[0088] (18)

[0089] (19)

[0090] (20)

[0091] (twenty one)

[0092] (twenty two)

[0093] (twenty three)

[0094] (twenty four)

[0095] Among them, T represents the objective function, N A is the total number of drones; formulas (5)-(9) represent the kinematic constraints of drones, such as Figure 2 As shown, a UAV at a vertex can choose the vertex of the connected edge, face diagonal or body diagonal in the adjacent cube as the next path point, and the distance between each edge is l; is the horizontal coordinate of the i-th UAV at time t, is the vertical coordinate of the i-th UAV at time t, is the vertical coordinate of the i-th UAV at time t; is the displacement step length of the i-th UAV on the horizontal axis at time t, is the displacement step length of the i-th UAV on the ordinate at time t, is the displacement step length of the i-th UAV in the vertical coordinate at time t; , and are the maximum boundaries of the horizontal, vertical and vertical coordinates respectively; formulas (10)-(14) represent the kinematic constraints of the unmanned vehicle, such as Figure 3 As shown, the unmanned vehicle can select the vertex of the connected edge or the vertex of the diagonal line in the adjacent two-dimensional grid as the next path point, and the distance between each edge is l, where is the horizontal coordinate of the j-th autonomous vehicle at time t, is the ordinate of the j-th autonomous vehicle at time t, is the vertical coordinate of the j-th unmanned vehicle at time t; is the displacement step length of the j-th unmanned vehicle on the horizontal axis at time t, is the displacement step length of the j-th unmanned vehicle on the ordinate at time t, is the displacement step length of the jth unmanned vehicle on the vertical coordinate at time t; Formulas (15)-(17) indicate that when the UAV battery power is less than a certain threshold, the UAV will descend to a height of l and hover, waiting for the UAV to charge. is the power threshold at which the drone must hover and wait for charging; Formula (18) indicates that the drone cannot collide with buildings and cannot fly in no-fly zones, where is the no-fly zone for drones; Formula (19) indicates that the unmanned vehicle cannot collide with buildings and obstacles, where is the unmanned vehicle restricted zone; formulas (20)-(21), (22)-(23) represent the communication constraints between UAVs and unmanned vehicles and the communication constraints between UAVs, respectively. is the i-th drone, is the communication range of the jth unmanned vehicle, The maximum communication range between the UAV and the unmanned vehicle, is the communication range of the mth UAV, is the maximum communication range between UAVs; Formula (24) shows that the UAV coverage range is a rectangle, where , is the left and right coverage of the drone on the x-axis, , is the upper and lower coverage of the drone on the y-axis; Figure 4As shown, the flying height of the UAV is l, the length of each grid is l, X max =Y max =20l, if the drone is located at ( , )=(10l,10l), according to formula (24), we have = =9l, = =11l, so when the drone is flying at a height of l, its coverage area is a square rectangle of 2l*2l; when the drone is flying at a height of 2l, such as Figure 5 As shown, each grid length is l, X max =Y max =20l, if the drone is located at ( , )=(10l,10l), according to formula (24), we have = =8l, = =12l, so when the drone's flight altitude is 2l, its coverage area is a square rectangle of 4l*4l. It is not difficult to find that as the drone's flight altitude increases, the drone's coverage area will gradually increase.

[0096] 2.5, Discrete Problem Formulation;

[0097] Since the low-altitude airspace in an urban environment can be represented by multiple discrete points in a cube, the collaborative path planning problem for the full coverage mission can be converted into a 0-1 optimization problem. Taking the three-dimensional position array P as the optimization variable, we have:

[0098] (25)

[0099] in, is the set of three-dimensional positions of drones, and D is the set of three-dimensional coordinates of all buildings.

[0100] In equation (25), p(x, y, z) = -1 indicates that the point is located on a building, obstacle, or inaccessible location. In this case, it is defined as a non-pathpoint, and such pathpoints are always prohibited from passing. Other points are defined as pathpoints, and their on and off states are represented by 1 and 0. Each pathpoint needs to be optimized. To minimize the mission time, the goal is to find the minimum number of open points to cover the entire target area. The remaining pathpoints are set as closed points, and the open points that have been covered are also converted to closed points. UAVs and unmanned vehicles do not need to pass through closed points during driving. For all open points, the drone must visit all of them to achieve complete coverage.

[0101] 3. Solve;

[0102] 3.1, Repair strategy;

[0103] For this type of large-scale 0-1 optimization problem, due to its complex grid coverage judgment conditions, there is a high probability that infeasible solutions will appear when generating new solutions. To ensure that all solutions are feasible, the method of opening all path points and then closing them point by point is adopted to make the solution feasible. The specific operation is as follows:

[0104] Step 1: Assuming that all path points are open points, randomly select an open point;

[0105] Step 2: Change the state of the selected open point to closed, and check whether the remaining open points can still achieve complete coverage. If so, close the selected open point; otherwise, keep the point open.

[0106] Step 3: Randomly select another open point that has not been closed and repeat the above operation;

[0107] Step 4: Determine whether all open points have been closed. If so, end and output the feasible solution; otherwise, return to step 3.

[0108] All infeasible solutions are repaired in the above way, and the fitness value of each solution is calculated based on the number of open points when full coverage is achieved.

[0109] 3.2. Improve the meme algorithm to calculate the path point status;

[0110] This embodiment uses the improved meme algorithm (IMA) and the lawnmower algorithm to solve the full coverage detection path. The IMA is responsible for calculating the minimum set of open points to achieve full coverage, and the lawnmower algorithm is responsible for further obtaining the shortest full coverage path that traverses each open point through the open point set. The full coverage detection path solution flow chart is shown in the figure below. Figure 6 As shown, first, the improved memetic algorithm IMA is used to solve the optimal open point set, which includes the following steps:

[0111] Step 1: Initialize the population, set the population size, individual code length and value range, and randomly generate each individual in the initial population;

[0112] Step 2: Perform repair operations on each individual to ensure that all solutions are feasible solutions;

[0113] Step 3: Decode and calculate the fitness value of each solution according to the objective function of the problem, and determine the current individual extreme value pbest and global extreme value gbest;

[0114] Step 4: For each individual, perform the following operations:

[0115] Step 4.1: Update individual codes using selection, crossover, mutation, etc.

[0116] Step 4.2: Optimize the quality of the solution using the 2-opt local search strategy;

[0117] Step 4.3: Evaluate the objective function of the updated individual and calculate its fitness value;

[0118] Step 4.4: Calculate the fitness diversity index. When the diversity index is low, reduce the crossover rate and increase the mutation rate to escape the local optimum. When the diversity index of the population is high, maintain a high crossover rate and a low mutation rate to promote the generation of new solutions.

[0119] Step 4.5: The number of iterations does not reach t=t max / 2, the new solution replaces the original solution; the number of iterations reaches t=t max When the value of the new solution is greater than the original solution, the new solution can replace the original solution.

[0120] Step 4.6: Update pbest and gbest;

[0121] Step 5: According to the preset termination conditions, determine whether the algorithm has ended. If the termination conditions are met, output the current global optimal solution and the algorithm ends; otherwise, return to step 4 and continue iteration.

[0122] 3.2.1. Dynamic adjustment of crossover rate and mutation rate;

[0123] The relative diversity index D is obtained by dividing the standard deviation of the fitness value by the mean value. It can measure the degree of dispersion of the fitness value relative to the fitness mean and reflect the relative diversity of the fitness value. The relative diversity index is shown in formula (26):

[0124] (26)

[0125] Where D is the fitness diversity index, is the standard deviation of fitness value, is the average fitness value, and the piecewise linear relationship between the crossover rate and mutation rate and the diversity index D is shown in formula (27) and formula (28):

[0126]

[0127]

[0128] Among them, p c is the crossover rate, p m is the mutation rate, D this the diversity index threshold, and formula (27) and formula (28) are respectively the following: th When the crossover rate p c and mutation rate p m They change linearly within the specified interval, and when the diversity index D is higher than the threshold, it is always a high crossover rate and a low mutation rate; by dynamically adjusting the crossover rate p c and mutation rate p m , the parameters can be flexibly adjusted according to the fitness diversity of the current population; if the population diversity index is low, the crossover rate is reduced and the mutation rate is increased to avoid local optimality; if the diversity index is high, it means that the population has greater uncertainty, then a high crossover rate and a low mutation rate are maintained to promote the generation of new solutions.

[0129] 3.2.2, 2-opt local search strategy;

[0130] For each pair of nodes, first calculate the sum of the fitness of the original path segments, and then calculate the sum of the fitness of the new path segments after the exchange. If the latter is better, perform the exchange operation; Figure 7 As shown in the figure, if the path generated by the algorithm is A-B-D-E-F-C, and after the local search of the 2-opt algorithm, the path segment between D-E and F-C is reversed to D-F and E-C, it is obvious that the path lengths between D-F and E-C are shorter than the path length between D-E and F-C, and the fitness value after reversal is better. Therefore, the paths at the local positions are exchanged. This method realizes local exchange by continuously comparing the fitness values ​​of different node pairs, thereby enhancing the exploration ability of the solution space and the convergence of the algorithm.

[0131] 3.2.3, Evolutionary comparison operation;

[0132] When the number of iterations of the algorithm reaches t=t max When the value of the algorithm is / 2, an evolutionary comparison operation is introduced to compare the newly generated solution with the original solution. Only when the fitness of the new solution is significantly better than the original solution is the new solution accepted; otherwise, the original solution is retained. This improvement ensures the stability of the algorithm's evolution and improves global search capabilities.

[0133] 3.3, lawnmower algorithm determines the order of path points;

[0134] The lawnmower algorithm is used to traverse the optimal open point set obtained by IMA. Starting from the first column, a scanning operation is performed on each row and each layer. The drone will sequentially pass through each point in the open point set determined by IMA, thereby obtaining the access path of the drone to traverse all open points. The schematic diagram of the lawnmower algorithm is shown in the figure below. Figure 8 As shown, the specific traversal steps are as follows:

[0135] Step 1: Set the path index initial value d = 0, start traversing from the first layer grid;

[0136] Step 2: For each layer grid (from x = 1 to the maximum horizontal grid number X max / l), perform the following operations:

[0137] Step 2.1: If the current layer is the first layer, traverse each column in turn (z = 1 to the maximum height grid number Z max / l) according to the row index y = 1 to the maximum vertical grid Y max / l), check whether each grid point is an open point, and if the condition is met, store the three-dimensional coordinates of the point in the path array S and increment the path index d;

[0138] Step 2.2: If the current layer is a subsequent layer (x > 2), adjust the traversal direction according to the parity of the layer number:

[0139] When the layer number is even, traverse each row in turn according to the row index Y max / l to y = 1;

[0140] When the layer number is odd, traverse each row in turn according to the row index y = 1 to Y max / l;

[0141] When traversing each row, check the state of the grid points of each column and height in turn, record all open points in the path array S, and increment the path index d;

[0142] Step 3: After traversal, check whether the path array S covers all path points, if there are uncovered points, the traversal order needs to be adjusted and step 2 is repeated until the path meets the full coverage requirement;

[0143] Step 4: The generated path array S is used as the access order of the unmanned aerial vehicle traversing the open points, and the full coverage path planning in the three-dimensional space is completed.

[0144] The unmanned aerial vehicle has a coverage range, and when performing a coverage task, the unmanned aerial vehicle will cover all open points (points that the unmanned aerial vehicle itself will pass through) and all closed points (points covered by the coverage range of the unmanned aerial vehicle) in the process of passing through all open points. The open points and closed points are collectively referred to as path points, and all open points are traversed by the unmanned aerial vehicle once;

[0145] The optimization goal of the embodiment is to reduce the number of open points under the constraint of ensuring full coverage of the path points, because many path points have been covered by the coverage range, so closing this point will make the open points of the unmanned aerial vehicle fewer, thereby the path is shorter and the task time is shorter.

[0146] 3.4, dynamic adjustment strategy of charging points;

[0147] After the UAV gets the full coverage path of the mower, in order to reduce the UAV charging waiting time, the dynamic charging point position is updated based on the obstacle condition between the UAV and the unmanned vehicle, specifically:

[0148] Step 1: Determine whether there is an obstacle between the UAV and the unmanned vehicle in a straight line distance;

[0149] Step 2: If there is an obstacle, the UAV moves the farthest distance to the unmanned vehicle under low power protection, and if the UAV can fly over the obstacle at this time, the position after flying over the obstacle is taken as the new charging point position;

[0150] Step 3: If the remaining power of the UAV is insufficient to fly over the obstacle, the UAV flies to the position of the low power hovering protection as the new charging point position

[0151] Step 4: If there is no obstacle between the UAV and the unmanned vehicle, the UAV flies to the position of the low power hovering protection as the new charging point position;

[0152] The schematic diagram of the UAV charging point update scenario is shown in Figure 9 In this scenario, the unmanned vehicle needs to serve the charging points 1, 2, and 3 in turn, and the dark grid is the obstacle blocking the post-disaster road. When the unmanned vehicle goes to the charging point 2 for charging service, it needs to detour. At this time, the UAV can fly over the obstacle and update the position after crossing the obstacle as the charging point 2', which effectively reduces the UAV charging waiting time and further reduces the overall task completion time by allowing the UAV and the unmanned vehicle to meet each other.

[0153] The embodiment also provides a computer device, which comprises a memory, a processor to store a computer program on the memory and run the computer program on the processor, and the processor executes the computer program to realize the steps of the above method.

[0154] The embodiment also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the above method.

[0155] This embodiment studies the problem of collaborative full-coverage path planning between drones and unmanned vehicles in a post-disaster urban environment. By constructing a post-disaster urban scenario in a grid network environment, the full-coverage path planning problem is modeled as a discrete optimization problem with the minimum task time. Then, an improved meme algorithm is used to solve the above optimization problem, which mainly includes dynamically adjusting the crossover rate and mutation rate, introducing a 2-opt local search and evolutionary comparison mechanism, effectively improving the global search capability and convergence speed of the meme algorithm. In addition, this embodiment also designs a dynamic adjustment strategy for charging points with two-way rendezvous, which effectively reduces the charging waiting time. Experimental results show that the results obtained by the improved meme algorithm in test scenarios of different scales are better than those of the comparison algorithm, and ablation experiments also verify the impact of various improvement strategies on the performance improvement of the algorithm.

[0156] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A post-disaster city-wide coverage detection method based on heterogeneous unmanned systems, characterized by: The following steps are involved: Considering drones and unmanned vehicles as heterogeneous unmanned systems, a full-coverage path planning model for heterogeneous unmanned systems is constructed; The objective function of the heterogeneous unmanned system full coverage path planning model is: ； Where, The time consumed by the i-th UAV to perform the coverage mission, is the charging time of the i-th drone, is the hovering waiting time of the i-th UAV; Discretizing the heterogeneous unmanned system full coverage path planning model by coding to obtain a discrete optimization model for the heterogeneous unmanned system full coverage path planning; The low-altitude airspace in the post-disaster city is divided into countless discrete cubic points, forming a three-dimensional position array, where each element in the array represents a discrete path point. These discrete path points are encoded to discretize the path planning problem for drones and unmanned vehicles, resulting in a discrete optimization model for full-coverage path planning for heterogeneous unmanned systems. An improved memetic algorithm is used to optimize the switch states of path points in the discrete optimization model to obtain the optimal open point set, and a lawnmower algorithm is used to traverse the optimal open point set to obtain the shortest full coverage detection path; The process of optimizing the switch states of path points in a discrete optimization model using an improved memetic algorithm includes: randomly generating an initial population and repairing it, evaluating the fitness of individuals in the population, selecting individuals with satisfactory fitness as parents through selection operations, generating offspring through crossover and mutation operations using dynamic adaptive crossover and mutation rates, and performing 2-opt local search optimization on the offspring; when the number of iterations reaches half, an evolutionary comparison operation is introduced, and the population is iteratively updated repeatedly until the termination condition is met to obtain the optimal set of open points; The process of repairing the initial population includes: assuming that all path points are open points, randomly select an open point; change the state of the selected open point to closed, check whether the remaining open points can achieve complete coverage, and if so, close the selected open point; otherwise, keep the selected open point open; randomly select the next open point that has not been closed and repeat the above steps until all open points have been closed, and output a feasible solution; Constructing a dynamic adjustment mechanism for charging points, and performing full coverage detection based on the shortest full coverage detection path and the dynamic adjustment mechanism for charging points; When the drone's battery level drops to a threshold, the dynamic charging point adjustment mechanism is activated. The drone dynamically adjusts the charging point position based on the current battery level and the obstacles between it and the unmanned vehicle. When there is an obstacle on the straight path between the drone and the unmanned vehicle, if the drone has enough remaining battery to fly over the obstacle, the position after flying over the obstacle will be used as the new charging point; if the remaining battery level is insufficient or there are no obstacles on the path, the drone will fly toward the unmanned vehicle to the low-battery hovering protection position and set the corresponding position as the new charging point.

2. The method for full coverage detection of cities after disasters based on heterogeneous unmanned systems according to claim 1 is characterized in that: The constraints of the heterogeneous unmanned system full coverage path planning model include power constraints, kinematic constraints, communication constraints, mission constraints, safety constraints, environmental constraints, and collaborative constraints.

3. The post-disaster city full coverage detection method based on heterogeneous unmanned systems according to claim 1 is characterized in that: The relative diversity index is obtained by calculating the ratio of the standard deviation of the fitness value to the mean value. If the diversity index is lower than the threshold, the crossover rate is reduced and the mutation rate is increased. If the diversity index is higher than the threshold, the high crossover rate and low mutation rate are maintained.

4. A computer device 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 method according to any one of claims 1 to 3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images