Low-altitude unmanned aerial vehicle cluster inspection path planning method based on group cooperation

Abstract

The application relates to the technical field of unmanned aerial vehicle path planning, in particular to a low-altitude unmanned aerial vehicle cluster inspection path planning method based on group cooperation, which comprises the following steps: constructing a group cooperation index by analyzing the inspection task progress and the body energy consumption state cooperation characteristics of each unmanned aerial vehicle in the unmanned aerial vehicle cluster in the inspection process, taking the group cooperation index as the cost constraint weight when the unmanned aerial vehicle cluster performs path planning, and avoiding the problem that the low efficiency of the unmanned aerial vehicle cluster inspection caused by the unreasonable path planning due to the load imbalance problem of the unmanned aerial vehicle cluster; then constructing a single machine stability index by analyzing the change characteristics of the flight attitude of a single unmanned aerial vehicle during the inspection, avoiding the problem that the energy consumption is intensified due to the frequent change of the flight attitude of the unmanned aerial vehicle, and thus improving the path planning optimization capability of the unmanned aerial vehicle inspection.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) path planning technology, specifically to a method for low-altitude UAV swarm inspection path planning based on swarm collaboration. Background Technology

[0002] With the development of UAV technology, it has been widely used in patrol and inspection surveying fields such as power line inspection and terrain mapping, enabling it to complete target inspection tasks in complex terrain environments. However, the inspection areas are characterized by large terrain spatial scale and dispersed inspection targets, which makes single-UAV inspections suffer from limited energy consumption and low inspection efficiency, making it difficult to quickly complete large-scale inspection tasks. Therefore, using UAV swarms for collaborative inspection as a key technology to improve inspection efficiency has gradually become a research hotspot in the field of inspection.

[0003] While collaborative inspection by drone swarms offers significant operational advantages, traditional swarm intelligence optimization algorithms, when planning inspection paths for drone swarms in complex environments, rely on the sum of constraints such as the shortest inspection path or shortest inspection time for each drone as the optimization objective. Achieving a reasonable path plan for the drone swarm to minimize energy consumption and maximize inspection efficiency remains a challenge. These methods fail to consider the collaborative operational characteristics of drone swarms in complex terrain conditions. For example, complex terrain causes significant differences in energy consumption among drones during actual flight. Inspections based on pre-planned fixed paths can lead to uneven load distribution among drones in the swarm. Some drones with abnormally high loads may experience unstable flight states at the end of their inspection tasks due to limited power, affecting inspection efficiency. This "bottleneck effect" caused by unbalanced drone workloads can negatively impact the overall efficiency of the drone swarm inspection. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides a method for planning low-altitude UAV swarm inspection paths based on swarm collaboration, thereby resolving the existing problems.

[0005] The low-altitude UAV swarm inspection path planning method based on swarm collaboration in this application adopts the following technical solution: One embodiment of this application provides a method for path planning of low-altitude UAV swarm inspection based on swarm collaboration, the method comprising the following steps: Obtain the drone flight map of the area to be inspected; The flight map is divided into grids, and all grids containing the inspection target areas are used as target node grids. The inspection path of the UAV swarm is planned by combining the multi-ant colony algorithm, and the path obtained by each ant colony after each iteration is obtained. On the path, when the UAV moves to the current evaluation node, the first index when moving from the current evaluation node to the pre-selected node is calculated based on the length of the path already traveled and the remaining path length, combined with the estimated power consumption and the estimated remaining power. The average of the first indexes obtained by all UAVs at the same time when they evaluate the next node on their most recent iteration path at the current evaluation node is obtained. Combined with the first index, the group coordination index of UAVs when moving from the current evaluation node to the pre-selected node is determined, so as to calculate the corresponding load penalty weight and determine the corresponding weighted cost function. During the process of the UAV moving from the previous node to the current evaluation node and then to the pre-selected node, the change in the angle of the UAV's forward direction is obtained in order to calculate the corresponding single-machine stability index and determine the corresponding adaptive pheromone volatility. The heuristic function and pheromone update strategy for the next iteration are determined based on the weighted cost function and the adaptive pheromone volatility, respectively, and the next iteration is performed; the UAV cluster inspection path obtained at the end of the iteration cycle is obtained.

[0006] In one embodiment, the process of obtaining the first index is as follows: Calculate the difference between the estimated remaining battery power of each drone when it reaches the current evaluation node and the preset minimum battery power threshold, and record it as the first difference; calculate the difference between the battery power of each drone when it is at the first node and the estimated remaining battery power of the drone when it reaches the current evaluation node, and record it as the second difference; calculate the ratio of the first difference to the second difference, and record it as the first ratio. Obtain the path length already traveled and the remaining path length of each drone after reaching the pre-selected node; calculate the ratio of the remaining path length to the path length already traveled, and record it as the second ratio. The first index of each drone is positively correlated with the first ratio and negatively correlated with the second ratio.

[0007] In one embodiment, the group coordination index is the ratio of the first index to the mean.

[0008] In one embodiment, the process of obtaining the load penalty weight is as follows: When the values ​​of the first index and the group coordination index are both greater than a constant 1, the load penalty weight is set to the product of the first index and the group coordination index. Otherwise, the load penalty weight is set to a constant 1.

[0009] In one embodiment, the weighted cost function is positively correlated with the load penalty weight.

[0010] In one embodiment, the weighted cost function is specifically the product of the original cost function and the load penalty weight; wherein the value of the original cost function is equal to the current evaluation node. With pre-selected nodes The result is obtained by dividing the path length between them by the drone's cruising speed.

[0011] In one embodiment, the process of obtaining the angle change is as follows: The direction from the centroid of the previous node to the centroid of the current evaluation node is obtained and denoted as the first direction; the direction from the centroid of the current evaluation node to the centroid of the preselected node is denoted as the second direction, and the angle change is the angle that changes when turning from the first direction to the second direction.

[0012] In one embodiment, the single-machine stability index is positively correlated with the normalized value of the angle change.

[0013] In one embodiment, the single-machine stability index is the sum of the angle change and the constant 1.

[0014] In one embodiment, the adaptive pheromone volatility is: obtaining the product of the single-machine stability index and the preset initial value of the pheromone volatility coefficient, and taking the minimum value between the product and the preset maximum pheromone volatility coefficient as the adaptive pheromone volatility.

[0015] This application has at least the following beneficial effects: This application obtains the drone flight map of the area to be inspected, divides the flight map into a grid, obtains the target node grid, and plans the inspection path of the drone swarm using a multi-ant colony algorithm, obtaining the path obtained by each ant colony after each iteration. During the inspection process of the drone swarm according to the path obtained in each iteration, the collaborative characteristics of the inspection task progress and airframe energy consumption status of each drone in the drone swarm are analyzed to construct a group collaboration index, which serves as the cost constraint weight when the drone swarm is planning the path, avoiding the problem of low inspection efficiency caused by unreasonable path planning due to the unbalanced load of the drone swarm. Then, the application analyzes the performance of individual drones in the inspection process. The flight attitude change characteristics during inspection are used to construct a single-machine stability index to adaptively adjust pheromone volatility, avoiding the problem of increased energy consumption caused by frequent changes in UAV flight attitude, thereby improving the UAV's ability to optimize inspection path planning. The heuristic function and pheromone update strategy for the next iteration are determined based on the weighted cost function and the adaptive pheromone volatility, respectively, and the next iteration is performed. The UAV swarm inspection path obtained at the end of the iteration cycle is obtained. This solves the problem that traditional multi-ant colony algorithms do not consider the group collaboration characteristics between UAVs in the inspection path planning of UAV swarms, resulting in unbalanced load in the UAV swarm and poor overall inspection efficiency of the obtained inspection path. Attached Figure Description

[0016] Figure 1 A flowchart of the low-altitude UAV swarm inspection path planning method based on swarm collaboration provided in this application; Figure 2 This is a schematic diagram illustrating the process of obtaining the change in angle. Detailed Implementation

[0017] The following description, in conjunction with the accompanying drawings, details the specific scheme of the low-altitude UAV swarm inspection path planning method based on group collaboration provided in this application.

[0018] One embodiment of this application provides a method for planning inspection paths for low-altitude unmanned aerial vehicle (UAV) swarms based on group collaboration.

[0019] Specifically, the following method for low-altitude UAV swarm inspection path planning based on swarm collaboration is provided. Please refer to [link / reference]. Figure 1 The method includes the following steps: Step S1: Obtain the drone flight map of the area to be inspected.

[0020] Remote sensing data of the area to be inspected is acquired using satellite remote sensing technology, and topographic data features of the area are obtained using GIS technology. A flight map of the UAV is constructed by combining the remote sensing image data and the topographic data acquired from GIS. Both satellite remote sensing and GIS technologies are well-known, and the specific process will not be elaborated further.

[0021] The method for constructing the flight map specifically involves using a GIS platform, such as QGIS software platform, to align remote sensing image data and GIS terrain data in the WGS84 coordinate system, performing spatial registration to obtain an electronic map of the area to be inspected, which includes terrain data such as mountains, vegetation, and buildings in the area. The GIS modeling method mentioned is existing technology, and its specific process will not be elaborated further. Step S2: The flight map is divided into grids, with all grids containing the inspection target areas as target node grids. The inspection path of the UAV swarm is planned using a multi-ant colony algorithm, and the path obtained by each ant colony after each iteration is obtained. On the path, when the UAV moves to the current evaluation node, based on the length of the path already traveled and the remaining path length, combined with the estimated power consumption and the estimated remaining power, the first index for moving from the current evaluation node to the pre-selected node is calculated.

[0022] In the process of path planning for drone swarms, the complexity of the terrain and the collaborative characteristics among the drones play a crucial role in improving the overall inspection efficiency. Traditional swarm intelligence optimization algorithms, such as multi-ant colony optimization, optimize the inspection path of a drone swarm by summing the constraints of each drone as the optimization objective to find the minimum constraint cost. However, they do not consider the collaborative nature of the drones, leading to a non-optimal path. Drones are constrained by complex terrain during inspections, causing increased energy consumption when adjusting their flight attitude. If the status and energy consumption of drones in the swarm are not analyzed in a timely and coordinated manner to optimize path planning, drones may experience reduced inspection speed due to energy constraints and susceptibility to environmental interference, resulting in unstable flight attitudes. Ultimately, this leads to an unbalanced load on the drone swarm, and this "weakest link" effect results in low overall inspection efficiency, failing to meet the requirements for efficient drone swarm inspections.

[0023] To address the aforementioned issues, this application constructs a group collaboration index by analyzing the task progress and airframe energy consumption status of each UAV in the UAV swarm during the inspection process. This index serves as a cost constraint weight for the UAV swarm when performing path planning, thus avoiding the problem of low inspection efficiency caused by unreasonable path planning due to unbalanced load in the UAV swarm. Then, it analyzes the flight attitude change characteristics of individual UAVs during inspection to construct a single-aircraft stability index, avoiding the problem of increased energy consumption caused by frequent changes in UAV flight attitude, thereby improving the UAV inspection path planning optimization capability.

[0024] Specifically, the flight map of the area to be inspected is processed into a grid, with each grid having a side length of [missing information]. rice, Its size can be set by the implementer according to the inspection scenario and the accuracy requirements of the inspection task. This application does not impose any special restrictions. In this embodiment, The size is set to 10. The inspection target area in the flight map after being divided into grids is used as the grid of each target node. When the UAV swarm is planning the inspection path, it needs to completely traverse all the target node grids to complete the inspection task.

[0025] Furthermore, the first node in the target node mesh The target node grid is defined as follows: ,in, , This indicates the total number of target node meshes. The topographic elevation data is recorded as ;Will Spatial coordinate information is defined as This includes the two-dimensional coordinates of the target node's mesh center. The first... The drone is at the first node. The amount of electricity at that time is recorded as the initial reference amount of electricity. , The data type is a dimensionless percentage. For the number of drones, the first The real-time location information of each drone is recorded as follows: The real-time battery level and location information of the drone are obtained through the drone control system.

[0026] This application adopts the multi-ant colony algorithm as the basic intelligent optimization algorithm for UAV swarm inspection path planning. Based on the group cooperation index and the single-machine stability index, the multi-ant colony algorithm is improved to enhance its ability to find the inspection path.

[0027] First, we construct the basic cost function for inspection path planning. ,in express The cost set of drone inspection This indicates finding the maximum value of the set. Indicates the first The set of target node grids selected by a drone during the inspection process. express The spatial distance between two target node grids should be noted. It should be noted that the UAV selects its path from the node... Move to node hour, Must be The reachable nodes, the The reachable node is defined as... Nodes that are directly adjacent in space. The spatial distance between two nodes can be calculated given their coordinates and elevation data; the specific calculation process will not be elaborated further. Indicates the first The cruising speed of each drone. The basic cost function aims to maximize the inspection time in the drone swarm, and minimizes the basic cost function. The final optimization result is that the entire drone swarm can complete the inspection task evenly. The principle behind constructing the basic cost function is that the completion of the drone swarm inspection task refers to the maximum task time taken by each drone in its respective inspection task. Only when all drones have completed their inspection tasks can the inspection of the entire area be completed.

[0028] Furthermore, with The reciprocal of the multi-ant colony algorithm is the first... An ant in a colony is at a node and Heuristic function for moving between , No. An ant in a colony is at a node Pheromones concentration function during movement Initial value of pheromone evaporation coefficient and initial value of pheromone concentration We take values ​​of 0.5 and 0.2 respectively. It should be noted that... and The implementer can set the value according to the actual situation; this application does not impose specific restrictions. The number of ants in the ant colony is the same as the number of drones, and the number of ants in one ant colony is... indivual, The reference number of nodes is set, typically 1 to 1.5 times the number of nodes. In this embodiment, it is set to a positive integer closest to 1.5 times the number of nodes. The iteration cycle of the multi-ant colony algorithm... In this embodiment, the value is set to 200. The optimization termination condition of the multi-ant colony algorithm is the end of the iteration cycle, and the output is the inspection path node sequence of each UAV. The multi-ant colony algorithm is existing technology, and its specific optimization process will not be described in detail.

[0029] Then, the first iteration is started using the multi-ant colony algorithm defined above. After the first iteration, the inspection paths of each drone in the drone cluster are obtained, denoted as the first path. , For the first iteration The optimal paths found by the ant colony have high pheromone concentrations. In the multi-ant colony algorithm, the ant colony continues to optimize the path based on the pheromone state of the path in the previous round during each iteration. Therefore, the pheromone state of the path after each iteration can be used by the ant colony to explore the path in subsequent iterations.

[0030] Based on the above analysis, then taking the first... Taking the optimization process of the first ant colony as an example, let's take the first ant colony as an example. The path obtained by the ant colony in the first iteration Based on this, we construct ant colony coordination and individual machine stability indices to optimize subsequent iterative optimization processes. The path obtained by the ant colony in the first iteration is denoted as the first path.

[0031] Furthermore, in the second iteration process, the... An ant colony in Expand and optimize accordingly. The set of nodes contained therein is denoted as .

[0032] The first path obtained by planning the inspection path of the drones in the basic cost function does not consider the swarm collaboration characteristics among drones. This leads to significant differences in the inspection task load of each drone. For example, some drones may have inspection paths with more complex terrain, requiring frequent take-offs and landings for inspection, and consuming more time and power to adjust their flight attitude. Conversely, some drones may have inspection paths with smooth terrain, allowing them to complete the task by flying at a single horizontal altitude. This results in an imbalance in the task load of each individual drone in the swarm, affecting the overall inspection efficiency of the swarm.

[0033] No. A drone from the node Move gradually to the current evaluation node At that time, the power consumption per unit distance is dynamically calculated based on the already traveled grid distance and the preset average power consumption per unit distance. A drone moved to the node. The estimated power consumption is calculated by multiplying the ratio of this value to the total battery capacity of the drone by 100%. The result is recorded as the percentage power consumption estimate. The total battery capacity and the average power consumption per unit distance are obtained based on the drone model's factory parameters. In this embodiment, the total battery capacity is taken as 1.5 kWh, and the average power consumption per unit distance is taken as 0.1 kWh / km. Based on the above initial baseline battery capacity of the drone... Subtracting the percentage power consumption estimate from the first value will give the second value. A drone moves to the node. Estimated remaining power at that time .

[0034] Based on the above analysis, then when the first An ant colony in Nodes in and The first exponent when moving between can be expressed as: In the formula, For the k-th drone from the current evaluation node To the pre-selected node The first index during movement; For the k-th drone to reach the node The estimated remaining power at that time; The k-th drone is at the first node Battery life at that time; This indicates the minimum battery threshold for drone path optimization. Its size can be set by the implementer according to the drone parameters, without special restrictions; in this embodiment, it is taken as 25%. The purpose of this function is to avoid... Negative numbers affect the calculation of the first exponent. express From the node To the destination node The set of nodes visited Find the path length over a set of nodes, i.e. arrive The sum of the spatial distances between all nodes represents the node currently selected by the drone. The remaining inspection path length. Similarly, This indicates the length of the inspection path that the drone has traveled so far. Specifically, when u=1, since both the used battery power and the distance traveled are 0, then... The value is directly set to 1; when the node When it is the endpoint node, because If it is 0, then The value is directly set to 1.

[0035] Explanation of the calculation principle: Indicates that the drone has reached Stable available power at that time, when This indicates that the drone's battery is too low, resulting in reduced flight speed and unstable flight status. This indicates that no additional optimization or expansion of the drone's inspection path will be performed, and the inspection task will be completed according to the original path. The meaning is that the drone has reached The amount of electricity already used. Overall. The meaning expressed is the proportional relationship between the remaining battery power of the drone and the remaining inspection tasks. At that time, it indicated that the drone still had more than [a certain number of] battery cells after completing its expected inspection mission. If so, then the drone has the additional capability to undertake inspection tasks; conversely, if not, If the drone is just able to complete the expected inspection task and its current battery level is too low, it is not suitable for extended inspections.

[0036] Specifically, when the ant colony deviates during the node selection process... After that, i.e., the current node. Not here In the middle, then at this time Indicated as from arrive The sum of spatial distances of the grid set traversed by the line connecting the centroid coordinates.

[0037] Step S3: Obtain the average of the first indices obtained by all UAVs at the same time when they evaluate the next node on their respective paths obtained in the most recent iteration at the current evaluation node. Combine the first index to determine the group coordination index when the UAV moves from the current evaluation node to the pre-selected node, so as to calculate the corresponding load penalty weight and determine the corresponding weighted cost function.

[0038] Furthermore, calculate the current time. The first index of each ant colony when making path selection, and then the mean of the first index of the current drone swarm is calculated and denoted as . , then the first An ant colony in On the node and The group coordination index during movement between them can be expressed as: Explanation of the calculation principle: When When, it indicates the first The ability of one drone to conduct extended inspections during the inspection process is higher than the average extended inspection capability of all drones, that is, the [missing information]. Compared to other drones, this drone has more battery power remaining after completing its inspection mission, allowing it to undertake more inspection tasks; conversely, when... Time indicates the first The extended inspection capability of this drone is lower than the standard of other drones' extended inspection capabilities. Specifically, when... When the value is 0, meaning the estimated remaining battery power of all drones in the current drone cluster is lower than the preset minimum battery threshold, then... The value is set to 0.

[0039] Based on the above analysis, then the first An ant colony in On the node Load penalty weight when moving between and weighted cost function It can be represented as: Explanation of the calculation principle: When and When, it indicates the current number. A drone in selection If a node serves as an inspection path and has additional remaining power to support extended inspections after completing its inspection task, and its extended inspection capability far exceeds the average extended inspection capability of a drone swarm, then the selection will be... Nodes used as inspection paths are subject to additional load penalty weights to prevent them from being selected. Instead of using nodes as inspection paths, other nodes are selected for extended inspections, thereby taking on additional inspection tasks and balancing the overall load of the drone swarm. The penalty weight is related to the ability to perform extended inspections; the more remaining battery power and the higher the extended inspection capability, the greater the weight, thus encouraging the first node to perform the extended inspection. One drone is used for extended inspections. Conversely, when this condition is not met... If so, no penalty will be imposed when the drone selects a node.

[0040] Step S4: During the process of the UAV moving from the previous node to the current evaluation node and then to the pre-selected node, obtain the angle change of the UAV's forward direction to calculate the corresponding single-machine stability index and determine the corresponding adaptive pheromone volatility.

[0041] Furthermore, the flight attitude adjustment of drones has a strong impact on their power consumption and inspection time. For example, when the terrain is complex or the direction needs to be changed, the drone cannot perform continuous and smooth inspections along the horizontal contour lines. It needs to frequently ascend, descend, and turn to adjust its flight attitude, which will lead to a significant increase in the drone's energy consumption. Moreover, the process of adjusting the flight attitude is time-consuming, affecting the efficiency of the inspection.

[0042] Based on the above analysis, a single-machine stability index is constructed based on the angle difference between the pre-selected node and the preceding node of the UAV to evaluate the impact of the currently selected node on the flight stability of the UAV.

[0043] Given that the three-dimensional spatial coordinates of a node can be obtained from its two-dimensional spatial coordinates and elevation information, the geometric relationships between nodes in space can then be calculated. Therefore, the third... An ant colony in On the node and Single-machine stability index during movement and adaptive pheromone volatility It can be represented as: In the formula, express The preceding node; Indicates from Center of mass direction The direction of the center of mass is denoted as the first direction; Indicates from Center of mass direction The direction of the center of mass is denoted as the second direction; It is 180° in radians, used for normalizing molecules; The initial value for the preset pheromone evaporation coefficient; The preset maximum pheromone evaporation coefficient; This indicates the method for calculating the angle between two directed lines in space. The method for calculating the angle given the coordinates of three points is existing technology and will not be elaborated further. Specifically, if the current evaluation node is the first node... At that time, the angle change is not calculated, and its single-machine stability index is directly assigned a constant value of 1. Among these, greater than Furthermore, values ​​less than 1 can be set by the implementer according to the actual situation; this application does not impose specific restrictions. In this embodiment, Set it to 0.9.

[0044] Explanation of the calculation principle: When and When they are completely in the same direction, that is If the line connecting the points in space is a smooth straight line, it means that the path chosen by the drone is smooth enough that no readjustment of flight attitude is required after selecting this path, and thus the path is considered good. It should be noted that even... The elevations vary, but as long as and If they are completely in the same direction, then along During flight, the drone's flight attitude is either upward or downward, continuing along... If no flight attitude adjustment is needed during flight, then the angle between the two directed lines is 0. This indicates that the pheromone concentration will not be changed, and the pheromone concentration along the path will be calculated according to the preset pheromone concentration evaporation coefficient. Conversely, when... and When they are not in the same direction, It gradually increases as the angle between the two lines increases. The larger the value, the more likely the drone is to be selected. When serving as the next path node, the drone needs to quickly adjust its flight attitude for ascent, descent, or sharp turns. This requires additional power and time for the drone to make these adjustments. Increase to reduce nodes The pheromone concentration between these points helps avoid choosing this less desirable path later.

[0045] Step S5: Determine the heuristic function and pheromone update strategy for the next iteration based on the weighted cost function and the adaptive pheromone volatility, and perform the next iteration; obtain the UAV cluster inspection path obtained at the end of the iteration cycle.

[0046] Based on the above analysis, in the second iteration, The reciprocal of is used as the heuristic function for the multi-ant colony algorithm, and the adaptive pheromone volatility is used. Substitute it into the pheromone concentration calculation formula as the pheromone update strategy of the multi-ant colony algorithm.

[0047] Using the above method, the group coordination index and individual stability index of each UAV can be calculated after the first iteration, and their respective heuristic functions and pheromone update strategies can be constructed for the second iteration. After the second iteration is completed, the second path of each UAV is obtained. Then, in the second iteration, the above steps are repeated based on the second path of each UAV to continue iterating and updating until the iteration cycle ends and the planning of the UAV cluster inspection path is completed.

[0048] A schematic diagram of the process of obtaining the angle change is shown below. Figure 2 As shown.

[0049] In summary, this application embodiment obtains the drone flight map of the area to be inspected, divides the flight map into grids, obtains the target node grids, and plans the inspection path of the drone swarm using a multi-ant colony algorithm, obtaining the path obtained by each ant colony after each iteration. During the inspection process of the drone swarm according to the path obtained in each iteration, the collaborative characteristics of the inspection task progress and airframe energy consumption status of each drone in the drone swarm are analyzed to construct a group collaboration index, which serves as the cost constraint weight for the drone swarm when planning the path, avoiding the problem of low inspection efficiency caused by unreasonable path planning due to the unbalanced load of the drone swarm. Then, the individual drones are analyzed. The system constructs a single-machine stability index based on the flight attitude change characteristics of the drone during inspection, adaptively adjusting the pheromone volatility to avoid increased energy consumption caused by frequent changes in drone flight attitude, thereby improving the drone's ability to optimize inspection path planning. Based on the weighted cost function and the adaptive pheromone volatility, the system determines the heuristic function and pheromone update strategy for the next iteration, and performs the next iteration. The system obtains the drone swarm inspection path at the end of the iteration cycle. This solves the problem that traditional multi-ant colony algorithms do not consider the group collaboration characteristics between drones in drone swarm inspection path planning, leading to unbalanced loads in the drone swarm and resulting in poor overall inspection efficiency.

Claims

1. A method for low-altitude UAV swarm inspection path planning based on swarm collaboration, characterized in that, The method includes the following steps: Obtain the drone flight map of the area to be inspected; The flight map is divided into grids, and all grids containing the inspection target areas are used as target node grids. The inspection path of the UAV swarm is planned by combining the multi-ant colony algorithm, and the path obtained by each ant colony after each iteration is obtained. On the path, when the UAV moves to the current evaluation node, the first index when moving from the current evaluation node to the pre-selected node is calculated based on the length of the path already traveled and the remaining path length, combined with the estimated power consumption and the estimated remaining power. The average of the first indexes obtained by all UAVs at the same time when they evaluate the next node on their most recent iteration path at the current evaluation node is obtained. Combined with the first index, the group coordination index of UAVs when moving from the current evaluation node to the pre-selected node is determined, so as to calculate the corresponding load penalty weight and determine the corresponding weighted cost function. During the process of the UAV moving from the previous node to the current evaluation node and then to the pre-selected node, the change in the angle of the UAV's forward direction is obtained in order to calculate the corresponding single-machine stability index and determine the corresponding adaptive pheromone volatility. The heuristic function and pheromone update strategy for the next iteration are determined based on the weighted cost function and the adaptive pheromone volatility, respectively, and the next iteration is performed; the UAV cluster inspection path obtained at the end of the iteration cycle is obtained.

2. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The process of obtaining the first index is as follows: Calculate the difference between the estimated remaining battery power of each drone when it reaches the current evaluation node and the preset minimum battery power threshold, and record it as the first difference. Calculate the difference between the battery level of each drone when it is at the first node and the estimated remaining battery level when the drone reaches the current evaluation node, and record it as the second difference. Calculate the ratio of the first difference to the second difference, and denot it as the first ratio. Obtain the path length and remaining path length of each drone after reaching the pre-selected node; Calculate the ratio of the remaining path length to the already traveled path length, and denot it as the second ratio; The first index of each drone is positively correlated with the first ratio and negatively correlated with the second ratio.

3. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The group coordination index is the ratio of the first index to the mean.

4. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The process of obtaining the load penalty weight is as follows: When the values ​​of the first index and the group coordination index are both greater than a constant 1, the load penalty weight is set to the product of the first index and the group coordination index. Otherwise, the load penalty weight is set to a constant 1.

5. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The weighted cost function is positively correlated with the load penalty weight.

6. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 5, characterized in that, The weighted cost function is specifically the product of the original cost function and the load penalty weight; wherein the value of the original cost function is equal to the current evaluation node. With pre-selected nodes The result is obtained by dividing the path length between them by the drone's cruising speed.

7. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The process for obtaining the angle change is as follows: The direction from the centroid of the previous node to the centroid of the current evaluation node is obtained and denoted as the first direction; the direction from the centroid of the current evaluation node to the centroid of the preselected node is denoted as the second direction, and the angle change is the angle that changes when turning from the first direction to the second direction.

8. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The single-machine stability index is positively correlated with the normalized value of the angle change.

9. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 8, characterized in that, The single-machine stability index is the sum of the angle change and the constant 1.

10. The method for low-altitude UAV swarm inspection path planning based on swarm collaboration as described in claim 1, characterized in that, The adaptive pheromone volatility is defined as follows: the product of the single-machine stability index and the preset initial value of the pheromone volatility coefficient is obtained, and the minimum value between the product and the preset maximum pheromone volatility coefficient is taken as the adaptive pheromone volatility.

Images