Unmanned aerial vehicle cluster collaborative disaster area three-dimensional point cloud modeling method

Abstract

The invention relates to a disaster area three-dimensional point cloud modeling method based on unmanned aerial vehicle cluster cooperation. The method comprises the following steps: carrying out regional division on a disaster area based on different terrain types; based on the divided areas, performing task allocation on the unmanned aerial vehicle cluster by adopting a task allocation algorithm to obtain a flight path of each unmanned aerial vehicle; based on the flight path, sensor data of the unmanned aerial vehicle in the flight process is obtained, the integrity of the collected data is detected in real time, and a disaster area data set is obtained; and processing the disaster area data set by adopting a three-dimensional reconstruction algorithm to obtain a disaster area three-dimensional point cloud model. The system can effectively cope with a complex terrain environment, achieves the precise analysis and comprehensive data collection of a disaster area, and provides important support for disaster assessment and rescue decision making.

Description

Technical Field

[0001] This invention relates to the field of disaster relief and environmental reconstruction technology, and in particular to a method for three-dimensional point cloud modeling of disaster areas using a swarm of unmanned aerial vehicles (UAVs). Background Technology

[0002] Unmanned aerial vehicle (UAV) swarm collaboration technology is of vital value in disaster relief and environmental reconstruction, enabling rapid acquisition of disaster area information and providing a reliable basis for rescue decisions. With the increasing frequency of disasters, the efficient and accurate construction of 3D models of disaster areas has become an indispensable part of emergency response, especially in complex terrain and emergency environments, where the importance of this technology is becoming increasingly prominent.

[0003] However, current methods often reveal shortcomings when facing the complex environments of disaster areas. Many solutions lack adaptability to diverse terrains and dynamic changes in disasters, especially in areas with varied terrain and different levels of disaster severity, where the comprehensiveness and consistency of data collection are difficult to guarantee. Furthermore, existing technologies often suffer from uneven task allocation or unreasonable regional division when using multi-machine collaboration, leading to the neglect of some key areas and affecting the overall modeling results.

[0004] Against this backdrop, the core challenges facing this field are becoming increasingly apparent. The complexity of disaster-stricken terrain makes it difficult for a single scanning strategy to adapt to the different terrain features and disaster impact levels of various areas, easily leading to blind spots or redundancy in data collection. This complexity further exacerbates the difficulty of task allocation and path planning for UAV swarms. Without reasonably dividing the scanning area and optimizing flight paths, it is difficult to achieve full coverage and efficient data acquisition. These two factors are closely related; insufficient terrain adaptability directly affects the rationality of task allocation, thereby limiting the accuracy and efficiency of overall modeling.

[0005] Therefore, how to design a reasonable regional division and task allocation strategy based on the terrain characteristics of the disaster area, and how to achieve full coverage and seamless connection of data collection by optimizing flight paths, has become a key issue that this study urgently needs to solve. Summary of the Invention

[0006] The purpose of this invention is to propose a method for three-dimensional point cloud modeling of disaster areas using a swarm of unmanned aerial vehicles (UAVs) to solve the problems existing in the prior art.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] A method for collaborative 3D point cloud modeling of disaster areas using drone swarms includes:

[0009] The disaster area was divided into regions based on different terrain types;

[0010] Based on the divided regions, a task allocation algorithm is used to allocate tasks to the drone swarm and obtain the flight trajectory of each drone.

[0011] Based on the flight trajectory, sensor data of the UAV during flight is acquired, the integrity of the collected data is detected in real time, and a disaster area data set is obtained.

[0012] A 3D reconstruction algorithm was used to process the disaster area dataset to obtain a 3D point cloud model of the disaster area.

[0013] Optionally, the disaster area can be divided into regions based on different terrain types, including:

[0014] Topographic data of the disaster area was obtained by acquiring high-resolution satellite imagery, and the imagery was processed using a pre-built classification model to extract topographic feature information.

[0015] Based on the extracted terrain feature information, complexity analysis is performed for different types of terrain. If the terrain features of a certain area show complexity characteristics of a preset threshold, priority labeling is performed to obtain labeled terrain distribution data.

[0016] Using labeled terrain distribution data and the actual range of the target area, the support vector machine algorithm is used to perform fine-grained classification of terrain categories and obtain refined terrain classification results.

[0017] Based on the refined terrain classification results, we formulate division strategies for different types of terrain and determine regional division schemes.

[0018] Optionally, the flight trajectory of each drone may be obtained, including:

[0019] A task allocation algorithm is used to assign tasks to the drone swarm. The performance parameters of each drone are matched with the terrain adaptability of the current area to obtain the task list for each drone.

[0020] For the task list, obtain the flight path planning requirements for each drone, combine the area division scheme and terrain feature data, generate the initial flight path through the path optimization algorithm, and obtain the flight trajectory of each drone.

[0021] Based on the initial flight trajectory, real-time environmental data and dynamic environmental change information are acquired. If a data acquisition blind spot is detected in a certain area, the flight path in that area is locally adjusted to obtain an optimized flight trajectory.

[0022] Optionally, obtaining the task list for each drone includes:

[0023] For the divided areas, the performance parameter data of each drone is obtained. Logical judgment rules are applied. If the performance parameters of a certain drone meet the terrain adaptation requirements of the current area, it is added to the candidate matching list to determine the preliminary matching relationship.

[0024] Using a candidate matching list, a genetic algorithm is employed to optimize task allocation for the drone swarm, addressing multi-objective constraints to obtain a task allocation scheme for each drone.

[0025] Based on the task allocation plan, the degree of terrain matching between each drone and the corresponding area is analyzed. If the degree of matching is lower than the preset threshold, the allocation plan is readjusted to determine the final task allocation result.

[0026] Optionally, the flight trajectory of each drone may be obtained, including:

[0027] Based on the task list, obtain the task allocation data corresponding to each drone, and determine the priority sequence of tasks to be performed by each drone by combining the correspondence between the task list and the drone swarm.

[0028] Based on the priority sequence, spatial grid data in the regional division scheme is obtained, terrain feature information is fused, and a regional distribution map containing terrain constraints is generated to obtain preliminary spatial constraints.

[0029] Using regional distribution results and combined with planning requirements, the start and end information of the flight path is extracted. An initial path that conforms to terrain constraints is generated through a path optimization algorithm to determine the preliminary flight trajectory.

[0030] For the initial flight trajectory, real-time wind speed and obstacle distribution data are acquired. If the wind speed exceeds the preset threshold or the obstacle is located within the path range, the path offset is recalculated through the dynamic adjustment module to obtain the corrected flight trajectory.

[0031] Based on the corrected flight trajectory, the communication range data of the drone swarm is integrated. If the communication signal strength is lower than the preset threshold, the optimized flight formation structure is determined by adjusting the cooperation distance between the drones.

[0032] By optimizing the flight formation structure, the flight trajectory execution parameters of each UAV are obtained, and the final flight path data is generated.

[0033] Optionally, real-time detection of the integrity of the collected data includes:

[0034] The data coverage analysis method is used to evaluate the distribution and coverage of the collected data. If the coverage is lower than the preset threshold, a supplementary collection instruction is triggered to determine the under-covered area.

[0035] Based on areas with insufficient coverage, the drone's flight path is adjusted, and the data update status is monitored in real time. By comparing the differences between the newly collected data and the initial dataset, it is determined whether the coverage meets the requirements.

[0036] Optionally, obtaining the disaster area dataset includes:

[0037] The detected data undergoes data consistency fusion processing, and data registration technology is used to spatially calibrate data from different sources to obtain a unified disaster area data set.

[0038] Optionally, obtaining the three-dimensional point cloud model of the disaster area includes:

[0039] A 3D reconstruction algorithm is used to process the disaster area dataset, and combined with terrain features and dynamic environmental change information, to generate a 3D point cloud model of the disaster area.

[0040] The beneficial effects of this invention are as follows:

[0041] This invention proposes a method for collaborative 3D point cloud modeling of disaster areas using UAV swarms. Addressing the complex terrain of disaster areas, it employs a pre-established terrain classification model for initial analysis, dynamically adjusts the region division scheme based on real-time UAV data, and allocates tasks according to UAV performance and terrain adaptability. Building upon this, the invention generates an initial flight trajectory using a path optimization algorithm, making local adjustments based on real-time environmental data. Simultaneously, it monitors the integrity of data acquisition in real time and dynamically adjusts the acquisition frequency. Finally, the invention fuses and spatially calibrates multi-source data to generate a high-precision 3D model of the disaster area. This method effectively handles complex terrain environments, enabling accurate analysis and comprehensive data acquisition of disaster areas, providing crucial support for disaster assessment and rescue decision-making. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of a method for three-dimensional point cloud modeling of disaster areas using a drone swarm collaboration, according to an embodiment of the present invention. Detailed Implementation

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

[0045] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] like Figure 1 As shown in the figure, this embodiment proposes a method for 3D point cloud modeling of disaster areas using UAV swarm collaboration, including:

[0047] The disaster area was divided into regions based on different terrain types;

[0048] Based on the divided regions, a task allocation algorithm is used to allocate tasks to the drone swarm and obtain the flight trajectory of each drone.

[0049] Based on the flight trajectory, sensor data of the drone during flight is acquired, the integrity of the collected data is detected in real time, and a disaster area data set is obtained.

[0050] A 3D reconstruction algorithm was used to process the disaster area dataset to obtain a 3D point cloud model of the disaster area.

[0051] Furthermore, the disaster area is divided into regions based on different terrain types, including:

[0052] Topographic data of the disaster area was obtained by acquiring high-resolution satellite imagery, and the imagery was processed using a pre-built classification model to extract topographic feature information.

[0053] Based on the extracted terrain feature information, complexity analysis is performed for different types of terrain. If the terrain features of a certain area show complexity characteristics of a preset threshold, priority labeling is performed to obtain labeled terrain distribution data.

[0054] Using labeled terrain distribution data and the actual range of the target area, the support vector machine algorithm is used to perform fine-grained classification of terrain categories and obtain refined terrain classification results.

[0055] Based on the refined terrain classification results, we formulate division strategies for different types of terrain and determine regional division schemes.

[0056] Specifically, in this embodiment, for the analysis and classification of the complex terrain features of the disaster area, a preliminary analysis of the target area is first conducted using a pre-established terrain classification model. A convolutional neural network (CNN) algorithm based on deep learning is employed. Through training on historical terrain data, the model can identify terrain types such as mountains, plains, and valleys. The specific training dataset contains a large number of labeled terrain images. The geographic coordinate range of the target area is input, and a terrain classification probability distribution map is automatically output, laying the foundation for subsequent analysis. Next, terrain feature data is extracted from high-resolution satellite imagery. A digital elevation model (DEM) combined with image processing algorithms (such as Sobel edge detection) is used to extract slope, aspect, and elevation features. The average slope, aspect, and elevation range are calculated, and a feature dataset is automatically generated and stored in raster format to ensure data accuracy meets analytical requirements. Subsequently, an initial region division scheme is generated for different terrain types. The K-means clustering algorithm is used to classify the feature data, dividing the target area into mountains, valleys, plains, and other types. Spatial proximity constraints are incorporated during the division process to ensure the continuity of the divided regions. The system outputs a vector-format boundary map of the division and the preliminary division results. Based on the initial division results, the real-time location and sensor data of the drone swarm are obtained, and dynamic adjustments are made in combination with terrain feature data. If the terrain complexity of a certain area exceeds a preset threshold, the area is further subdivided to determine the adjusted area division scheme.

[0057] This embodiment, based on the complex terrain features, uses a pre-established terrain classification model to perform preliminary analysis of the target area. This model accurately identifies different terrain types and generates an initial region division scheme accordingly. Furthermore, it dynamically adjusts the region division by combining the real-time location of the UAV swarm with sensor data, resulting in more refined and accurate division. This provides a high-precision foundation for subsequent task allocation and flight path planning, effectively avoiding data blind spots and redundancy caused by unreasonable region division and improving the accuracy of the modeling results.

[0058] Furthermore, obtaining the flight trajectory of each drone includes:

[0059] A task allocation algorithm is used to assign tasks to the drone swarm. The performance parameters of each drone are matched with the terrain adaptability of the current area to obtain the task list for each drone.

[0060] For the task list, obtain the flight path planning requirements for each drone, combine the area division scheme and terrain feature data, generate the initial flight path through the path optimization algorithm, and obtain the flight trajectory of each drone.

[0061] Based on the initial flight trajectory, real-time environmental data and dynamic environmental change information are acquired. If a data acquisition blind spot is detected in a certain area, the flight path in that area is locally adjusted to obtain an optimized flight trajectory.

[0062] Furthermore, obtaining the mission list for each drone includes:

[0063] For the divided areas, the performance parameter data of each drone is obtained. Logical judgment rules are applied. If the performance parameters of a certain drone meet the terrain adaptation requirements of the current area, it is added to the candidate matching list to determine the preliminary matching relationship.

[0064] Using a candidate matching list, a genetic algorithm is employed to optimize task allocation for the drone swarm, addressing multi-objective constraints to obtain a task allocation scheme for each drone.

[0065] Based on the task allocation plan, the degree of terrain matching between each drone and the corresponding area is analyzed. If the degree of matching is lower than the preset threshold, the allocation plan is readjusted to determine the final task allocation result.

[0066] Specifically, in this embodiment, obtaining the task list for each drone includes:

[0067] Based on the divided sub-regions, a task allocation algorithm is adopted, such as an optimization method based on genetic algorithms. The initial population size is set to 100 and the number of iterations is 200. The optimal solution is selected by calculating the fitness function (combining UAV performance and terrain matching degree). It is assumed that the upper limit of UAV 1 is 1000 meters and the climb rate is 2 meters / second, which is suitable for mountain missions, and its fitness score is 0.85; while the upper limit of UAV 2 is 500 meters and the climb rate is 1 meter / second, with a fitness score of 0.6, which is more suitable for plains. Furthermore, the system matches the performance parameters and terrain adaptability of each drone. Drone 1's battery life is automatically calculated to be 2 hours, with a maximum payload of 5 kg. This is compared to the task requirements of sub-region A (3 hours of inspection, 4 kg payload). The system finds insufficient battery life, but through algorithm optimization, its task time is adjusted to 1.5 hours. By having another drone share the remaining tasks, a successful match is achieved, and a task list is generated. For example, Drone 1 is assigned to inspect the first half of sub-region A, specifically at coordinates 120.5–120.7 E, 30.1–30.3 N, with a task duration of 1.5 hours; Drone 2 is assigned to inspect the entire sub-region B, at coordinates 120.8–121.0 E, 30.2–30.4 N, with a task duration of 2 hours. Through this process, the algorithm automatically completes the entire process from region division to task allocation, ensuring a high degree of fit between each drone's task, performance, and terrain, resulting in a logically rigorous allocation.

[0068] This embodiment uses a task allocation algorithm to assign tasks to the drone swarm based on the adjusted regional division scheme. It comprehensively considers the performance parameters of each drone and its adaptability to the terrain of the current area, achieving reasonable task matching and efficient allocation. This avoids situations where some drones are overloaded with tasks while others are idle, fully leveraging the advantages of the drone swarm, improving overall operational efficiency, and enabling the completion of data collection tasks in large disaster areas in a shorter time.

[0069] Furthermore, obtaining the flight trajectory of each drone includes:

[0070] Based on the task list, obtain the task allocation data corresponding to each drone, and combine the correspondence between the task list and the drone swarm to determine the priority sequence of tasks to be performed by each drone.

[0071] Based on the priority sequence, spatial grid data in the regional division scheme is obtained, terrain feature information is fused, and a regional distribution map containing terrain constraints is generated to obtain preliminary spatial constraints.

[0072] Using regional distribution results and combined with planning requirements, the start and end information of the flight path is extracted. An initial path that conforms to terrain constraints is generated through a path optimization algorithm to determine the preliminary flight trajectory.

[0073] For the initial flight trajectory, real-time wind speed and obstacle distribution data are acquired. If the wind speed exceeds the preset threshold or the obstacle is located within the path range, the path offset is recalculated through the dynamic adjustment module to obtain the corrected flight trajectory.

[0074] Based on the corrected flight trajectory, the communication range data of the drone swarm is integrated. If the communication signal strength is lower than the preset threshold, the optimized flight formation structure is determined by adjusting the cooperation distance between the drones.

[0075] By optimizing the flight formation structure, the flight trajectory execution parameters of each UAV are obtained, and the final flight path data is generated.

[0076] Specifically, in this embodiment, for the task list, the flight path planning requirements of each UAV are obtained. Combining the area division scheme and terrain feature data, an initial flight path is generated through a path optimization algorithm, and the flight trajectory of each UAV is determined, including:

[0077] To address the flight path planning requirements for each drone in the task list, the system first automatically analyzes the task data. Assuming there are three drones covering a 10 square kilometer area, the task list specifies that each drone must inspect a particular area: Drone 1 is responsible for area A (3 square kilometers), Drone 2 for area B (4 square kilometers), and Drone 3 for area C (3 square kilometers). Based on the area division scheme, the system divides area A into three sub-grids (each 1 square kilometer), area B into four sub-grids (each 1 square kilometer), and area C into three sub-grids (each 1 square kilometer). Terrain feature data is then loaded, assuming area A has two high points (200 meters above sea level), area B has one body of water (which needs to be bypassed), and area C has flat terrain. The system generates an initial flight path using path optimization algorithms, such as the A* algorithm. The specific process is as follows: Starting from the UAV's takeoff point (coordinates 0,0) and the target point as the center point of the sub-grid, the system calculates the distance between nodes and the cost of terrain obstacles. For example, in region A, the straight-line distance from sub-grid 1 to sub-grid 2 is 1.5 kilometers, but due to the need to detour around high ground, the actual path cost increases by 20%. The optimized path length is 1.8 kilometers, and the flight altitude is adjusted to 250 meters to avoid obstacles. Similarly, in region B, due to the need to detour around water, the path cost increases by 15%, and the single sub-grid inspection path for UAV 2 is adjusted from 2 kilometers to 2.3 kilometers. Region C has flat terrain, and the path directly uses a straight-line distance of 1.2 kilometers. The system comprehensively calculates the total flight trajectory of each drone. The total path of drone 1 is 5.4 kilometers, drone 2 is 9.2 kilometers, and drone 3 is 3.6 kilometers. It generates three-dimensional trajectory data (including latitude, longitude, and altitude) to ensure that the paths do not intersect and the flight time is controlled within the mission window (e.g., 2 hours). If the time limit is exceeded, the inspection sequence is automatically adjusted to optimize time costs and form a complete flight trajectory plan.

[0078] In this embodiment, real-time environmental data and dynamic environmental change information are also obtained based on the initial flight trajectory. If a data acquisition blind spot is detected in a certain area, the flight path in that area is locally adjusted to obtain an optimized flight trajectory.

[0079] Furthermore, real-time detection of the integrity of the collected data includes:

[0080] The data coverage analysis method is used to evaluate the distribution and coverage of the collected data. If the coverage is lower than the preset threshold, a supplementary collection instruction is triggered to determine the under-covered area.

[0081] Based on areas with insufficient coverage, the drone's flight path is adjusted, and the data update status is monitored in real time. By comparing the differences between the newly collected data and the initial dataset, it is determined whether the coverage meets the requirements.

[0082] Specifically, in this embodiment, in accordance with the requirement of comprehensive data coverage, a preset coverage algorithm is used for real-time detection. The algorithm is based on spatial grid division, for example, dividing the flight area into 10m × 10m grid units, and calculating the number of data collections for each unit. If the number of collections for a unit exceeds a preset threshold (e.g., 5 times), it is determined to be data redundancy; if it is below the threshold, it is marked as a data missing area, and a supplementary collection path is automatically generated. The path planning uses the A* algorithm to ensure that the shortest path covers the missing area, while avoiding repeated flights to already covered areas. Subsequently, to address the data collection redundancy problem, the collection frequency is dynamically adjusted. For example, in redundant areas, the frequency is reduced from 10 times per second to 5 times per second. The adjustment is based on the redundancy percentage calculation formula: Redundancy = (Actual collection count - Threshold) / Threshold × 100%. If the redundancy exceeds 20%, the frequency is reduced. The adjusted frequency is smoothly transitioned using linear interpolation to avoid data abrupt changes. The final data acquisition plan is generated through comprehensive analysis. Based on the adjusted frequency and supplementary paths, the total data volume is calculated to ensure coverage of over 98%. A final plan report is generated, including a frequency distribution map and path optimization results, and stored in a cloud database for subsequent analysis. This method forms a complete logical chain from data acquisition to plan optimization, ensuring a balance between data integrity and efficiency.

[0083] Furthermore, obtaining the disaster area dataset includes:

[0084] The detected data undergoes data consistency fusion processing, and data registration technology is used to spatially calibrate data from different sources to obtain a unified disaster area data set.

[0085] Specifically, in this embodiment, a preset format conversion tool is used to perform preliminary standardization processing on various types of data to obtain a standardized initial dataset. Based on the standardized initial dataset, data consistency checks are performed. If deviations in data format or content are detected, a preset correction tool is used to adjust the deviations, determining the dataset after consistency verification. For the consistency-verified dataset, data registration technology is used to perform spatial calibration processing on data from different sources, obtaining a spatially aligned dataset. Using the spatially aligned dataset, preset fusion rules are applied to integrate the multi-source data. If data conflicts are found during the fusion process, they are filtered according to priority rules to obtain the fused disaster area dataset.

[0086] Furthermore, obtaining the 3D point cloud model of the disaster area includes:

[0087] A 3D reconstruction algorithm was used to process the disaster area dataset, and combined with terrain features and dynamic environmental change information to generate a 3D point cloud model of the disaster area.

[0088] Specifically, in this embodiment, a 3D reconstruction algorithm is used to perform structured analysis on the data, and combined with terrain features, an initial 3D framework structure is generated. Using this initial 3D framework structure, information on changes in the dynamic environment is acquired, environmental data is fused, and an intermediate model incorporating the time dimension is constructed. If a deviation is detected between the terrain features and the dynamic environment data in the intermediate model, the model parameters are adjusted using an iterative optimization method to determine the corrected intermediate model. Based on the corrected intermediate model, precision enhancement techniques are used to refine the model details, resulting in a high-precision 3D model. Using this high-precision 3D model, all information data is integrated to generate the final modeling result, completing the full presentation of the 3D scene of the disaster area.

[0089] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for 3D point cloud modeling of disaster areas using UAV swarm collaboration, characterized in that, include: The disaster area was divided into regions based on different terrain types; Based on the divided regions, a task allocation algorithm is used to allocate tasks to the drone swarm and obtain the flight trajectory of each drone. Based on the flight trajectory, sensor data of the UAV during flight is acquired, the integrity of the collected data is detected in real time, and a disaster area data set is obtained. A 3D reconstruction algorithm was used to process the disaster area dataset to obtain a 3D point cloud model of the disaster area.

2. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 1, characterized in that, The disaster area is divided into regions based on different terrain types, including: Topographic data of the disaster area was obtained by acquiring high-resolution satellite imagery, and the imagery was processed using a pre-built classification model to extract topographic feature information. Based on the extracted terrain feature information, complexity analysis is performed for different types of terrain. If the terrain features of a certain area show complexity characteristics of a preset threshold, priority labeling is performed to obtain labeled terrain distribution data. Using labeled terrain distribution data and the actual range of the target area, the support vector machine algorithm is used to perform fine-grained classification of terrain categories and obtain refined terrain classification results. Based on the refined terrain classification results, we formulate division strategies for different types of terrain and determine regional division schemes.

3. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 1, characterized in that, The flight path of each drone was obtained, including: A task allocation algorithm is used to assign tasks to the drone swarm. The performance parameters of each drone are matched with the terrain adaptability of the current area to obtain the task list for each drone. For the task list, obtain the flight path planning requirements for each drone, combine the area division scheme and terrain feature data, generate the initial flight path through the path optimization algorithm, and obtain the flight trajectory of each drone. Based on the initial flight trajectory, real-time environmental data and dynamic environmental change information are acquired. If a data acquisition blind spot is detected in a certain area, the flight path in that area is locally adjusted to obtain an optimized flight trajectory.

4. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 3, characterized in that, The task list for each drone includes: For the divided areas, the performance parameter data of each drone is obtained. Logical judgment rules are applied. If the performance parameters of a certain drone meet the terrain adaptation requirements of the current area, it is added to the candidate matching list to determine the preliminary matching relationship. Using a candidate matching list, a genetic algorithm is employed to optimize task allocation for the drone swarm, addressing multi-objective constraints to obtain a task allocation scheme for each drone. Based on the task allocation plan, the degree of terrain matching between each drone and the corresponding area is analyzed. If the degree of matching is lower than the preset threshold, the allocation plan is readjusted to determine the final task allocation result.

5. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 3, characterized in that, The flight path of each drone was obtained, including: Based on the task list, obtain the task allocation data corresponding to each drone, and determine the priority sequence of tasks to be performed by each drone by combining the correspondence between the task list and the drone swarm. Based on the priority sequence, spatial grid data in the regional division scheme is obtained, terrain feature information is fused, and a regional distribution map containing terrain constraints is generated to obtain preliminary spatial constraints. Using regional distribution results and combined with planning requirements, the start and end information of the flight path is extracted. An initial path that conforms to terrain constraints is generated through a path optimization algorithm to determine the preliminary flight trajectory. For the initial flight trajectory, real-time wind speed and obstacle distribution data are acquired. If the wind speed exceeds the preset threshold or the obstacle is located within the path range, the path offset is recalculated through the dynamic adjustment module to obtain the corrected flight trajectory. Based on the corrected flight trajectory, the communication range data of the drone swarm is integrated. If the communication signal strength is lower than the preset threshold, the optimized flight formation structure is determined by adjusting the cooperation distance between the drones. By optimizing the flight formation structure, the flight trajectory execution parameters of each UAV are obtained, and the final flight path data is generated.

6. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 1, characterized in that, Real-time detection of the integrity of the collected data includes: The data coverage analysis method is used to evaluate the distribution and coverage of the collected data. If the coverage is lower than the preset threshold, a supplementary collection instruction is triggered to determine the under-covered area. Based on areas with insufficient coverage, the drone's flight path is adjusted, and the data update status is monitored in real time. By comparing the differences between the newly collected data and the initial dataset, it is determined whether the coverage meets the requirements.

7. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 1, characterized in that, Obtaining the disaster area data set includes: The detected data undergoes data consistency fusion processing, and data registration technology is used to spatially calibrate data from different sources to obtain a unified disaster area data set.

8. The method for three-dimensional point cloud modeling of disaster areas using UAV swarm collaboration according to claim 1, characterized in that, Obtaining the 3D point cloud model of the disaster area includes: A 3D reconstruction algorithm is used to process the disaster area dataset, and combined with terrain features and dynamic environmental change information, to generate a 3D point cloud model of the disaster area.

Images