Cooperative dynamic obstacle avoidance method and system for unmanned aerial vehicle group

Abstract

The invention relates to a collaborative dynamic obstacle avoidance method and system for an unmanned aerial vehicle group, and relates to the technical field of unmanned aerial vehicle obstacle avoidance, and the method comprises the steps: obtaining unmanned aerial vehicle group information and delivery cargo information, determining the application data of each unmanned aerial vehicle and an unmanned aerial vehicle application number according to the unmanned aerial vehicle group information, and constructing an unmanned aerial vehicle matrix based on the unmanned aerial vehicle application number, and binding the delivery cargo information with the unmanned aerial vehicle matrix to obtain unmanned aerial vehicle matrix data. According to the unmanned aerial vehicle matrix, determining a reconnaissance unmanned aerial vehicle layer, collecting a real-time view image, performing obstacle analysis on the real-time view image and unmanned aerial vehicle application data, if an air obstacle interfering the unmanned aerial vehicle matrix exists, performing static and dynamic analysis on the air obstacle, and performing obstacle avoidance flight optimization adjustment on the unmanned aerial vehicle matrix. And the optimized and adjusted unmanned aerial vehicle matrix is obtained. And controlling the unmanned aerial vehicle matrix to perform air obstacle avoidance evolution based on the optimized and adjusted unmanned aerial vehicle matrix. According to the invention, the efficiency and stability of cargo distribution of the unmanned aerial vehicle group are improved.

Description

Technical Field

[0001] This application relates to the technical field of obstacle avoidance for unmanned aerial vehicles (UAVs), and in particular to a cooperative dynamic obstacle avoidance method and system for UAV swarms. Background Technology

[0002] In the logistics and distribution sector, drone swarms (multi-drone collaborative systems) are gradually becoming a key force in solving the "last mile" delivery problem due to their high efficiency and flexibility. Especially in scenarios such as urban low-altitude delivery, remote area material transportation, and medical emergency services, drone swarms can overcome obstacles such as traffic congestion and terrain limitations to achieve fast and accurate cargo delivery.

[0003] However, during logistics delivery, drone swarms encounter dynamic obstacles in the air. Currently, drone swarms employ dynamic path replanning strategies based on real-time perception and collaborative decision-making to achieve dynamic obstacle avoidance, ensuring the safe delivery of goods to designated locations. However, this obstacle avoidance process requires individual drones to adjust their posture, causing the goods on board to sway and vibrate. This not only directly threatens the physical integrity of the goods, especially fragile items, where even slight vibrations can lead to damage or performance degradation, increasing logistics costs and impacting customer satisfaction. Furthermore, the frequent shaking of the goods interferes with the drones' flight stability, further increasing obstacle avoidance difficulty and energy consumption. In extreme cases, it can even trigger a chain reaction, affecting the overall collaborative efficiency and safety of the drone swarm. Summary of the Invention

[0004] To address at least one of the aforementioned technical problems, this application provides a cooperative dynamic obstacle avoidance method and system for unmanned aerial vehicle (UAV) swarms.

[0005] Firstly, this application provides a cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms, employing the following technical solution: A cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms includes: Obtain information about the drone swarm and the delivery cargo carried by each drone in the swarm; Based on the drone swarm information, determine the application data and drone application number for each drone; A drone matrix is ​​constructed based on the drone application ID, and the delivery cargo information is bound to the drone matrix to obtain drone matrix data; The reconnaissance drone layer is determined based on the drone matrix, and real-time field-of-view images of the drone matrix during the cargo delivery process are collected based on the reconnaissance drone layer. Obstacle analysis is performed on the real-time field-of-view image and the UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix. If there are, static and dynamic analysis is performed on the aerial obstacles, and obstacle avoidance flight optimization is performed on the UAV matrix based on the analysis results and the UAV matrix data to obtain the optimized UAV matrix. Based on the optimized and adjusted UAV matrix control, the UAV matrix evolves to avoid aerial obstacles.

[0006] By employing the aforementioned technical solution, information on the drone swarm and the delivery cargo carried by each drone is obtained. Based on this information, the overall situation of the drone swarm and the specific tasks of each drone can be accurately understood, ensuring the orderly conduct of the entire delivery process and avoiding chaos caused by missing or incorrect information, thus improving the initial accuracy and reliability of the delivery task. The application data and application number of each drone are determined based on the acquired drone swarm information. By clearly defining the application data and number, precise positioning and parameter settings for individual drones can be achieved, laying the foundation for building an orderly drone matrix and making subsequent matrix operations more accurate and efficient. A drone matrix is ​​constructed based on the drone application numbers, and the delivery cargo information is bound to the drone matrix to obtain drone matrix data. Constructing a drone matrix integrates scattered drones into an organic whole. Through reasonable matrix arrangement, collaborative operations between drones can be achieved, improving delivery efficiency. Binding the delivery cargo information to the matrix makes the position and task of each drone within the matrix clearer, avoiding confusion in task allocation. The obtained drone matrix data provides comprehensive information for subsequent real-time monitoring and adjustments, ensuring that drones can be accurately scheduled according to actual conditions during the delivery process, improving the flexibility and responsiveness of the entire delivery system. Based on the drone matrix, a reconnaissance drone layer is determined, and real-time field-of-view images of the drone matrix during cargo delivery are collected from this layer. By acquiring these real-time images, the aerial situation surrounding the drone matrix is ​​promptly obtained, including information on potential obstacles and weather changes. These real-time images provide intuitive data for subsequent obstacle analysis and obstacle avoidance decisions, enabling drones to accurately perceive their surroundings in complex environments, detect potential hazards in advance, and provide strong support for safe delivery, effectively reducing risks during the delivery process. Obstacle analysis is performed on the real-time field-of-view images and drone application data to determine if there are any aerial obstacles interfering with the drone matrix. If so, static and dynamic analysis of the aerial obstacles is conducted, and the drone matrix is ​​optimized for obstacle avoidance flight based on the analysis results and drone matrix data, resulting in an optimized drone matrix. Obstacle analysis can quickly and accurately identify threats, while static and dynamic analysis further reveals the movement state of the obstacles. By combining obstacle avoidance optimization with UAV matrix data, the system can plan the optimal obstacle avoidance path for each UAV based on actual conditions. This allows the UAV matrix to efficiently adjust its flight status when encountering obstacles, ensuring the safety of delivered goods while avoiding collisions. This guarantees the smooth progress of delivery tasks and improves delivery safety and reliability. The optimized UAV matrix control system evolves to handle aerial obstacle avoidance, ensuring that the UAVs accurately execute obstacle avoidance maneuvers according to the predetermined plan. This not only successfully avoids obstacles but also minimizes the impact on delivered goods, maintaining the efficiency and stability of delivery.

[0007] In a preferred embodiment, this application can be further configured as follows: the obstacle analysis of the real-time field-of-view image and the UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix includes: A 3D model of the drones in flight is constructed based on the drone application data. Based on the delivery cargo information, the delivery cargo is subjected to three-dimensional simulation processing, and the processed cargo three-dimensional model is fitted with the drone three-dimensional model according to the loading correspondence and preset ratio to obtain the delivery three-dimensional model. The real-time field-of-view image is subjected to image fusion processing to obtain a processed real-time field-of-view image; The processed real-time field-of-view image is input into a preset obstacle model for detection to determine whether there is an obstacle in the real-time field-of-view image. If there is an obstacle, the relative positional relationship between the obstacle in the real-time field-of-view image and the reconnaissance drone layer is determined, and the relative positional relationship is used to determine whether the obstacle has reached the preset flight range of the reconnaissance drone layer. If the obstacle does not reach the preset flight range of the reconnaissance drone layer, the obstacle evolution at different time points is determined, the obstacle evolution is predicted and extrapolated, the future obstacle evolution between the obstacle and the reconnaissance drone layer is determined in the future time period, a three-dimensional model of the obstacle evolution is built based on the future obstacle evolution, and the three-dimensional model of the evolution is fitted to the delivery three-dimensional model according to the relative position relationship to obtain the monitoring three-dimensional model; The presence of aerial obstacles that could interfere with the UAV matrix is ​​determined based on the monitoring 3D model.

[0008] In a preferred embodiment, this application can be further configured such that: the image fusion processing of the real-time field-of-view image to obtain a processed real-time field-of-view image includes: Edge filtering is performed on the real-time field-view image to obtain a first-level layer and a second-level layer. The first-level layer is the basic outline layer of the scene in the real-time field-view image, and the second-level layer is the detailed outline layer of the scene in the real-time field-view image. The first-level layer is divided according to a preset division rule to obtain multiple layer sub-regions; Collect the sub-region histogram and sub-region flatness corresponding to each layer sub-region, as well as the global histogram and global flatness corresponding to the first-level layer, and determine the first-level tone curve and brightness enhancement curve corresponding to the first-level layer based on the global histogram and global flatness. Based on the sub-region histogram and sub-region flatness corresponding to each sub-region of the layer, determine the secondary tone curve and curve blending weight corresponding to each sub-region, and then blend the primary tone curve and the secondary tone curve to obtain the tone blending curve. Based on the curve blending weight, the tone blending curve and the brightness enhancement curve are blended, and each sub-region in the first-level layer is adjusted according to the blended field of view adjustment curve to obtain the adjusted first-level layer. The adjusted first-level layer and the second-level layer are then fused to obtain the processed real-time field-of-view image.

[0009] In a preferred embodiment, this application can be further configured as follows: performing static and dynamic analysis on the aerial obstacles, and optimizing the drone matrix for obstacle avoidance flight based on the analysis results and the drone matrix data to obtain an optimized drone matrix, including: Based on the monitoring 3D model, the collision point area between the aerial obstacle and the UAV matrix in the future time period is determined, and the dynamically adjusted evolving UAVs involved in the UAV matrix are determined based on the collision point area. Based on the drone matrix data, the first cargo information corresponding to each evolved drone is determined, and the first cargo information is input into a preset cargo delivery standard model for identification to obtain the first cargo sensitivity value corresponding to each cargo in the first cargo information. Determine whether the first cargo sensitivity value is greater than a preset sensitivity value. If so, take the cargo corresponding to the first cargo sensitivity value as a marker point, measure the boundary distance value of each area boundary of the impact point area, and determine the minimum boundary distance value and the target area boundary corresponding to the minimum boundary distance value based on the boundary distance value. Based on the UAV matrix data, determine the second cargo information corresponding to the non-evolving UAV located at the boundary of the target area, and input the second cargo information into a preset cargo delivery standard model for identification to obtain the second cargo sensitivity value corresponding to each cargo in the second cargo information. Determine whether the second cargo sensitivity value is greater than a preset sensitivity value. If not, calculate the relative distance between each non-evolved UAV in the target area boundary and the marker point. Based on the relative distance, determine the target non-evolved UAV that is closest to the marker point. Generate a position replacement command based on the target non-evolved UAV and the UAV corresponding to the marker point. Control the target non-evolved UAV and the UAV corresponding to the marker point to perform position replacement, and obtain the UAV matrix after position replacement. The evolution of drones in the drone matrix after position replacement is analyzed by interleaving them with the drones in the drone matrix that are at the boundary of the non-target area, and the matrix evolution path is obtained. Based on the matrix evolution path, the drone matrix after position replacement is optimized for obstacle avoidance flight, resulting in an optimized drone matrix.

[0010] In a preferred embodiment, this application can be further configured such that determining whether the sensitivity value of the second goods is greater than a preset sensitivity value includes: If the second cargo sensitivity value is greater than the preset sensitivity value, then the third cargo information corresponding to the non-evolving drone at each region boundary is determined one by one according to the boundary distance value. The third cargo information is then input into the preset cargo delivery standard model for identification to obtain the third cargo sensitivity value corresponding to each cargo in the third cargo information. Determine whether the sensitivity value of the third cargo is greater than the preset sensitivity value. If it is still greater than the preset sensitivity value, determine the center point position of the UAV matrix and the impact center point position of the impact point area, and connect the center point position and the impact center point position by extending the position line to obtain the matrix movement vector line. The obstacle avoidance route is generated based on the movement vector line, and the UAV matrix is ​​controlled to move along the flight path.

[0011] In a preferred embodiment, this application can be further configured such that: the process of controlling the UAV matrix to perform aerial obstacle avoidance evolution based on the optimized and adjusted UAV matrix further includes: When the drone matrix is ​​detected to have passed through an aerial obstacle, the optimized drone matrix is ​​backtracked according to the matrix evolution path or the flight path of the drone matrix is ​​backtracked according to the movement vector line to obtain the backtracked drone matrix.

[0012] In a preferred embodiment, this application can be further configured such that: the step of controlling the UAV matrix to perform aerial obstacle avoidance evolution based on the optimized and adjusted UAV matrix further includes: The system collects network communication information for controlling the UAV matrix and performs network status detection on the network communication information to determine whether the network status in the network communication information meets the preset network status. If it does not meet the preset network status, the system obtains regional network information within a preset range centered on the UAV matrix, determines network channel information based on the regional network information, and determines network channel groups other than the current application channel in the network communication information based on the network channel information. The signal transmission rate of each network channel group was obtained by performing signal transmission tests on each network channel. The signal transmission rates are filtered and combined for analysis, and the network channels are reorganized according to the transmission order corresponding to each signal transmission rate in the filtered and combined network channel group information to obtain the spliced ​​network channel for sending control commands to the UAV matrix.

[0013] Secondly, this application provides a cooperative dynamic obstacle avoidance system for unmanned aerial vehicle (UAV) swarms, employing the following technical solution: A cooperative dynamic obstacle avoidance system for unmanned aerial vehicle (UAV) swarms includes: The information acquisition module is used to acquire information about the drone swarm and the delivery cargo carried by each drone in the swarm. The application determination module is used to determine the application data and application number of each drone based on the drone swarm information. The matrix generation module is used to construct a drone matrix based on the drone application number and bind the delivery cargo information to the drone matrix to obtain drone matrix data; The image acquisition module is used to determine the reconnaissance drone layer based on the drone matrix, and to acquire real-time field-of-view images of the drone matrix during the cargo delivery process based on the reconnaissance drone layer; The obstacle analysis module is used to perform obstacle analysis on the real-time field-of-view image and the UAV application data, determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix, and if so, perform static and dynamic analysis on the aerial obstacles, and optimize the UAV matrix for obstacle avoidance flight based on the analysis results and the UAV matrix data to obtain the optimized UAV matrix. The obstacle avoidance optimization module is used to control the drone matrix to perform aerial obstacle avoidance evolution based on the optimized and adjusted drone matrix.

[0014] In one possible implementation, when the obstacle analysis module performs obstacle analysis on the real-time field-of-view image and the UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix, it is specifically used for: A 3D model of the drones in flight is constructed based on the drone application data. Based on the delivery cargo information, the delivery cargo is subjected to three-dimensional simulation processing, and the processed cargo three-dimensional model is fitted with the drone three-dimensional model according to the loading correspondence and preset ratio to obtain the delivery three-dimensional model. The real-time field-of-view image is subjected to image fusion processing to obtain a processed real-time field-of-view image; The processed real-time field-of-view image is input into a preset obstacle model for detection to determine whether there is an obstacle in the real-time field-of-view image. If there is an obstacle, the relative positional relationship between the obstacle in the real-time field-of-view image and the reconnaissance drone layer is determined, and the relative positional relationship is used to determine whether the obstacle has reached the preset flight range of the reconnaissance drone layer. If the obstacle does not reach the preset flight range of the reconnaissance drone layer, the obstacle evolution at different time points is determined, the obstacle evolution is predicted and extrapolated, the future obstacle evolution between the obstacle and the reconnaissance drone layer is determined in the future time period, a three-dimensional model of the obstacle evolution is built based on the future obstacle evolution, and the three-dimensional model of the evolution is fitted to the delivery three-dimensional model according to the relative position relationship to obtain the monitoring three-dimensional model; The presence of aerial obstacles that could interfere with the UAV matrix is ​​determined based on the monitoring 3D model.

[0015] In another possible implementation, when the obstacle analysis module performs image fusion processing on the real-time field-of-view image to obtain the processed real-time field-of-view image, it is specifically used for: Edge filtering is performed on the real-time field-view image to obtain a first-level layer and a second-level layer. The first-level layer is the basic outline layer of the scene in the real-time field-view image, and the second-level layer is the detailed outline layer of the scene in the real-time field-view image. The first-level layer is divided according to a preset division rule to obtain multiple layer sub-regions; Collect the sub-region histogram and sub-region flatness corresponding to each layer sub-region, as well as the global histogram and global flatness corresponding to the first-level layer, and determine the first-level tone curve and brightness enhancement curve corresponding to the first-level layer based on the global histogram and global flatness. Based on the sub-region histogram and sub-region flatness corresponding to each sub-region of the layer, determine the secondary tone curve and curve blending weight corresponding to each sub-region, and then blend the primary tone curve and the secondary tone curve to obtain the tone blending curve. Based on the curve blending weight, the tone blending curve and the brightness enhancement curve are blended, and each sub-region in the first-level layer is adjusted according to the blended field of view adjustment curve to obtain the adjusted first-level layer. The adjusted first-level layer and the second-level layer are then fused to obtain the processed real-time field-of-view image.

[0016] In another possible implementation, when the obstacle analysis module performs static and dynamic analysis on the aerial obstacles and optimizes the drone matrix for obstacle avoidance based on the analysis results and the drone matrix data to obtain an optimized drone matrix, it is specifically used for: Based on the monitoring 3D model, the collision point area between the aerial obstacle and the UAV matrix in the future time period is determined, and the dynamically adjusted evolving UAVs involved in the UAV matrix are determined based on the collision point area. Based on the drone matrix data, the first cargo information corresponding to each evolved drone is determined, and the first cargo information is input into a preset cargo delivery standard model for identification to obtain the first cargo sensitivity value corresponding to each cargo in the first cargo information. Determine whether the first cargo sensitivity value is greater than a preset sensitivity value. If so, take the cargo corresponding to the first cargo sensitivity value as a marker point, measure the boundary distance value of each area boundary of the impact point area, and determine the minimum boundary distance value and the target area boundary corresponding to the minimum boundary distance value based on the boundary distance value. Based on the UAV matrix data, determine the second cargo information corresponding to the non-evolving UAV located at the boundary of the target area, and input the second cargo information into a preset cargo delivery standard model for identification to obtain the second cargo sensitivity value corresponding to each cargo in the second cargo information. Determine whether the second cargo sensitivity value is greater than a preset sensitivity value. If not, calculate the relative distance between each non-evolved UAV in the target area boundary and the marker point. Based on the relative distance, determine the target non-evolved UAV that is closest to the marker point. Generate a position replacement command based on the target non-evolved UAV and the UAV corresponding to the marker point. Control the target non-evolved UAV and the UAV corresponding to the marker point to perform position replacement, and obtain the UAV matrix after position replacement. The evolution of drones in the replaced drone matrix is ​​analyzed by interleaving them with the drones in the replaced drone matrix that are at the boundary of the non-target area, and the evolution path of the matrix is ​​obtained. Based on the matrix evolution path, the drone matrix after position replacement is optimized for obstacle avoidance flight, resulting in an optimized drone matrix.

[0017] In another possible implementation, when the obstacle analysis module determines whether the second cargo sensitivity value is greater than a preset sensitivity value, it is specifically used for: If the second cargo sensitivity value is greater than the preset sensitivity value, then the third cargo information corresponding to the non-evolving drone at each region boundary is determined one by one according to the boundary distance value. The third cargo information is then input into the preset cargo delivery standard model for identification to obtain the third cargo sensitivity value corresponding to each cargo in the third cargo information. Determine whether the sensitivity value of the third cargo is greater than the preset sensitivity value. If it is still greater than the preset sensitivity value, determine the center point position of the UAV matrix and the impact center point position of the impact point area, and connect the center point position and the impact center point position by extending the position line to obtain the matrix movement vector line. The obstacle avoidance route is generated based on the movement vector line, and the UAV matrix is ​​controlled to move along the flight path.

[0018] In another possible implementation, the system further includes a matrix backtracking module, wherein, The matrix backtracking module is used to perform interleaved backtracking on the optimized and adjusted drone matrix according to the matrix evolution route or backtracking the flight route of the drone matrix according to the movement vector line when the drone matrix is ​​detected to have flown through an aerial obstacle, so as to obtain the backtracked drone matrix.

[0019] In another possible implementation, the system further includes: a channel combining module, a channel testing module, and a channel reassembly module, wherein, The channel combination module is used to collect network communication information for controlling the UAV matrix, and to perform network status detection on the network communication information to determine whether the network status in the network communication information meets the preset network status. If it does not meet the preset network status, the module obtains regional network information within a preset range centered on the UAV matrix, determines network channel information based on the regional network information, and determines a network channel group other than the current application channel in the network communication information based on the network channel information. The channel testing module is used to perform signal transmission tests on the network channel group one by one to obtain the signal transmission rate corresponding to each network channel; The channel reassembly module is used to perform filtering and combination analysis on the signal transmission rates, and to reassemble the network channels according to the transmission order corresponding to each signal transmission rate in the filtered and combined network channel group information, so as to obtain a spliced ​​network channel for sending control commands to the UAV matrix.

[0020] Thirdly, this application provides an electronic device that adopts the following technical solution: An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the aforementioned cooperative dynamic obstacle avoidance method for unmanned aerial vehicle swarms.

[0021] Fourthly, this application provides a computer storage medium, as follows: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned cooperative dynamic obstacle avoidance method for unmanned aerial vehicle swarms.

[0022] In summary, this application has the following beneficial technical effects: The process involves acquiring information about the drone swarm and the delivery cargo carried by each drone. This information allows for an accurate understanding of the overall drone swarm situation and the specific tasks of each drone, ensuring the orderly operation of the entire delivery process and avoiding chaos caused by missing or incorrect information. This improves the initial accuracy and reliability of the delivery task. Based on the acquired drone swarm information, the application data and application number of each drone are determined. Clearly defining the application data and number enables precise positioning and parameter settings for individual drones, laying the foundation for building an orderly drone matrix and making subsequent matrix operations more accurate and efficient. A drone matrix is ​​constructed based on the drone application numbers, and the delivery cargo information is bound to the drone matrix to obtain drone matrix data. Constructing a drone matrix integrates dispersed drones into an organic whole. Through a reasonable matrix arrangement, collaborative operations between drones can be achieved, improving delivery efficiency. Binding the delivery cargo information to the matrix makes the position and task of each drone within the matrix clearer, avoiding confusion in task allocation. The obtained drone matrix data provides comprehensive information for subsequent real-time monitoring and adjustments, ensuring precise scheduling of drones based on actual conditions during the delivery process, improving the flexibility and responsiveness of the entire delivery system. Based on the drone matrix, a reconnaissance drone layer is determined, and real-time field-of-view images of the drone matrix during cargo delivery are collected from this layer. By acquiring these real-time images, the aerial situation surrounding the drone matrix is ​​promptly obtained, including information on potential obstacles and weather changes. These real-time images provide intuitive data for subsequent obstacle analysis and obstacle avoidance decisions, enabling drones to accurately perceive their surroundings in complex environments, detect potential hazards in advance, and provide strong support for safe delivery, effectively reducing risks during the delivery process. Obstacle analysis is performed on the real-time field-of-view images and drone application data to determine if there are any aerial obstacles interfering with the drone matrix. If so, static and dynamic analysis of the aerial obstacles is conducted, and the drone matrix is ​​optimized for obstacle avoidance flight based on the analysis results and drone matrix data, resulting in an optimized drone matrix. Obstacle analysis can quickly and accurately identify threats, while static and dynamic analysis further reveals the movement state of the obstacles. By combining obstacle avoidance optimization with UAV matrix data, the system can plan the optimal obstacle avoidance path for each UAV based on actual conditions. This allows the UAV matrix to efficiently adjust its flight status when encountering obstacles, ensuring the safety of delivered goods while avoiding collisions. This guarantees the smooth progress of delivery tasks and improves delivery safety and reliability. The optimized UAV matrix control system evolves to handle aerial obstacle avoidance, ensuring that the UAVs accurately execute obstacle avoidance maneuvers according to the predetermined plan. This not only successfully avoids obstacles but also minimizes the impact on delivered goods, maintaining the efficiency and stability of delivery. Attached Figure Description

[0023] Figure 1 This is a flowchart of a cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms in one embodiment of this application.

[0024] Figure 2 This is a schematic diagram of a cooperative dynamic obstacle avoidance system for a drone swarm, according to one embodiment of this application.

[0025] Figure 3 This is a schematic block diagram of an electronic device according to one embodiment of this application. Detailed Implementation

[0026] The following is in conjunction with the appendix Figure 1 To be continued Figure 3 This application will be described in further detail.

[0027] This specific embodiment is merely an explanation of this application and is not intended to limit it. After reading this specification, those skilled in the art can make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they fall within the scope of the claims of this application.

[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0030] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0031] This application provides a cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms, executed by an electronic device. This electronic device can be a server or a terminal device. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smartphone, tablet, laptop, desktop computer, etc., but is not limited to these. The terminal device and the server can be directly or indirectly connected via wired or wireless communication. This application does not impose any limitations on this connection. Figure 1 As shown, the method includes: Step S10: Obtain information about the drone swarm and the delivery cargo carried by each drone in the swarm.

[0032] In this application, a drone swarm refers to a collection of multiple drones that can work together to complete specific tasks, such as logistics delivery. Drone swarm information is used to represent the overall data of the drone swarm, including but not limited to the number of drones, their location, flight status, task type, drone attribute parameters, etc. The delivery cargo information refers to the detailed information of the cargo loaded on each drone for delivery, such as the type, weight, fragility, volume, destination, etc.

[0033] In this embodiment, various types of sensors are installed on the drone, such as a GPS module for obtaining location information and an accelerometer for obtaining flight acceleration data. Each cargo item is also equipped with a unique RFID tag. A ground control station is established, equipped with corresponding data receiving and processing equipment, and a communication link is established with the drones. During mission execution, the drones transmit information collected by their own sensors and the RFID tag information read from the cargo items to the ground control station in real time via the communication link. The ground control station stores and analyzes the received information. It is understood that other methods can also be used to obtain information about the drone swarm and the cargo information carried by each drone in the swarm. For example, satellite communication technology can be used to allow drones to communicate directly with satellites and transmit information to the ground receiving station; or blockchain technology can be used to ensure the authenticity and immutability of the information, with each drone uploading information to the blockchain network, from which relevant parties can obtain the required information. This is not limited to these methods.

[0034] Step S11: Determine the application data and application number of each drone based on the drone swarm information.

[0035] In this embodiment, UAV application data represents specific data related to the application of a single UAV within a UAV swarm. This data encompasses the UAV's geometry (e.g., fuselage length, width, height, wing shape and size), appearance features (e.g., color, material texture), internal structure (e.g., engine location, cargo hold layout), and flight attitude parameters (e.g., angles and motion data during takeoff, cruising, and landing). This data allows for the accurate reconstruction of the UAV's three-dimensional form in a real-world scenario. The UAV application number is a unique identifier assigned to each UAV. This number is used to accurately distinguish and identify different UAVs within the system, facilitating UAV management, scheduling, and the recording and tracking of related data. For example, in a UAV swarm of 10 UAVs, each UAV has its own independent application number, such as UAV-001, UAV-002, etc. The three-dimensional model constructed based on each UAV's application data will vary depending on the UAV's model and function. For instance, the three-dimensional models of cargo UAVs and reconnaissance UAVs will differ significantly in their cargo hold structure and reconnaissance equipment layout.

[0036] Step S12: Construct a drone matrix based on the drone application ID, and bind the delivery cargo information to the drone matrix to obtain drone matrix data.

[0037] In this embodiment, a drone matrix is ​​a logical structure for the orderly arrangement and organization of drones. It is constructed according to different dimensions (such as spatial location, functional type, task priority, etc.). For example, it can be constructed as a three-dimensional grid matrix based on spatial location, with each cell corresponding to a drone, facilitating overall control and collaborative operation of the drone swarm. The drone matrix data is a comprehensive data set formed after binding operations. It not only contains the structural information of the drone matrix but also the delivery information bound to the drone corresponding to each matrix position. This data clearly shows the position of each drone in the matrix and the goods it needs to deliver. For example, in an e-commerce drone delivery system, there are 10 drones, each with a unique application number. By constructing a 2x5 drone matrix, these 10 drones are arranged in different positions within the matrix, and the delivery information of different customers is bound to the corresponding drones in the matrix, thus forming the drone matrix data.

[0038] In this embodiment, a layout diagram of a drone matrix is ​​drawn on a computer using a graphical user interface (GUI). In the layout diagram, each cell represents a drone position. Cells are filled by inputting drone application numbers, thus visually constructing the drone matrix. Simultaneously, the drones in the real world are controlled according to the constructed drone matrix, allowing them to be arranged according to the drone matrix in the layout diagram. Then, a cargo information management platform is established, which can import delivery cargo information and provide query and filtering functions. Cargo information is filtered by setting filtering conditions, such as delivery time range and cargo weight range. An automatic binding script is developed. This script reads the numbering information from the drone matrix layout diagram and the filtering results from the cargo information management platform, and automatically binds the cargo information to the drones in the drone matrix according to a certain algorithm (such as the shortest delivery path algorithm). After binding, the binding results are displayed in the graphical interface, and the relevant data is exported as a drone matrix data table, containing detailed information such as drone number, cargo name, and destination.

[0039] Step S13: Determine the reconnaissance drone layer based on the drone matrix, and collect real-time field-of-view images of the drone matrix during the cargo delivery process based on the reconnaissance drone layer.

[0040] In this embodiment, during the construction of the drone matrix, a location identifier is assigned to each drone, containing its row and column information within the matrix. For example, a combination of numbers and letters, such as "A1" and "B2," is used to represent the drone's location. A location recognition program is written to read the location identifier of each drone, filtering out the drones located in the first row (or column, depending on the matrix's orientation) of the matrix and identifying them as the reconnaissance drone layer. Each drone in the reconnaissance drone layer is equipped with a high-definition camera, and the communication module between the camera and the drone is ensured to be properly connected. An image acquisition and transmission system is developed, capable of receiving image data captured by the camera in real time and transmitting the images to the control center via wireless communication technologies (such as Wi-Fi, 4G / 5G, etc.). During cargo delivery, the image acquisition and transmission system is activated, and the drones in the reconnaissance drone layer continuously acquire field-of-view images and transmit them back to the control center.

[0041] Step S14: Perform obstacle analysis on the real-time field-of-view image and UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix. If there are, perform static and dynamic analysis on the aerial obstacles, and optimize the UAV matrix for obstacle avoidance flight based on the analysis results and UAV matrix data to obtain the optimized UAV matrix.

[0042] Specifically, based on UAV application data, a 3D model of the UAV matrix during flight is constructed. The delivered goods are simulated in 3D based on delivery information. The processed 3D model of the goods is then fitted to the UAV 3D model according to the loading correspondence and a preset ratio to obtain the delivery 3D model. Real-time field-of-view images are fused to obtain processed real-time field-of-view images. These processed images are then input into a preset obstacle model for detection. If obstacles exist in the real-time field-of-view image, the relative positional relationship between the obstacle and the reconnaissance UAV layer is determined. Based on this relative positional relationship, it is determined whether the obstacle has entered the preset flight range of the reconnaissance UAV layer. If the obstacle has not entered the preset flight range, the evolution of the obstacle's relative positional relationship at different time points is determined. This evolution is predicted and extrapolated to determine the future obstacle evolution between the obstacle and the reconnaissance UAV layer in the future time period. Based on this future obstacle evolution, a 3D model of obstacle evolution is constructed. This evolution 3D model is then fitted to the delivery 3D model according to the relative positional relationship to obtain the monitoring 3D model. The monitoring 3D model is used to determine whether there are aerial obstacles in the real-time field-of-view image that could interfere with the UAV matrix.

[0043] In this embodiment, the loading correspondence clarifies the specific placement location and method of the cargo on the drone, such as whether the cargo is placed inside the drone's cargo hold or fixed externally by a mounting device. The preset scale is a proportional scale set based on the actual size relationship between the drone and the cargo to ensure the accuracy and rationality of the model fitting. By fitting the model according to the loading correspondence and the preset scale, the 3D model of the cargo and the 3D model of the drone are integrated to form a complete virtual model reflecting the actual loading situation. The preset obstacle model is a model pre-trained using machine learning or image recognition technology, capable of identifying various possible obstacles in the image, such as buildings, trees, and other flying objects. The preset flight range is a pre-defined safe space area required for the safe flight of the reconnaissance drone layer. Based on the relative positional relationship between the obstacle and the reconnaissance drone layer, it is determined whether the obstacle has entered this preset range, thereby assessing the potential threat to the drone's flight.

[0044] In this embodiment, the image fusion process for the real-time field-of-view image includes: edge filtering of the real-time field-of-view image to obtain a first-level layer and a second-level layer. The first-level layer is the basic outline layer of the scene in the real-time field-of-view image, and the second-level layer is the detailed outline layer of the scene in the real-time field-of-view image. The first-level layer is divided according to a preset division rule to obtain multiple sub-regions. The histogram and flatness of each sub-region are collected, as well as the global histogram and global flatness of the first-level layer. Based on the global histogram and global flatness, the first-level tone curve and brightness enhancement curve of the first-level layer are determined. Based on the histogram and flatness of each sub-region, the second-level tone curve and curve fusion weight of each sub-region are determined. The first-level tone curve and the second-level tone curve are fused to obtain a tone fusion curve. Based on the curve fusion weight, the tone fusion curve and the brightness enhancement curve are fused. Each sub-region in the first-level layer is adjusted according to the fused field-of-view adjustment curve to obtain an adjusted first-level layer. The adjusted first-level layer and the second-level layer are fused to obtain the processed real-time field-of-view image.

[0045] In this embodiment of the application, the process of optimizing the obstacle avoidance flight of the UAV matrix includes: determining the impact point area of ​​aerial obstacles with the UAV matrix in the future time period based on the monitoring three-dimensional model, and determining the dynamically adjusted evolved UAVs involved in the UAV matrix based on the impact point area. Determining the first cargo information corresponding to each evolved UAV based on the UAV matrix data, and inputting the first cargo information into a preset cargo delivery standard model for identification to obtain a first cargo sensitivity value corresponding to each cargo in the first cargo information. Determining whether the first cargo sensitivity value is greater than a preset sensitivity value; if so, using the cargo corresponding to the first cargo sensitivity value as a marker point, measuring the boundary distance value between the cargo and the boundary of each region of the impact point area, and determining the minimum boundary distance value and the target region boundary corresponding to the minimum boundary distance value based on the boundary distance value. Determining the second cargo information corresponding to the non-evolved UAVs located at the boundary of the target region based on the UAV matrix data, and inputting the second cargo information into the preset cargo delivery standard model for identification to obtain a second cargo sensitivity value corresponding to each cargo in the second cargo information. If the sensitivity value of the second cargo is greater than a preset sensitivity value, then the relative distance between each non-evolving UAV and the marker point in the target area boundary is calculated. Based on the relative distance, the target non-evolving UAV with the closest relative distance to the marker point is determined. A position replacement command is generated based on the target non-evolving UAV and the UAV corresponding to the marker point, controlling the target non-evolving UAV and the UAV corresponding to the marker point to perform position replacement, resulting in a position-replaced UAV matrix. The evolved UAVs in the position-replaced UAV matrix are then subjected to interleaved evolution analysis with the UAVs in the position-replaced UAV matrix that are not at the target area boundary, resulting in a matrix evolution path. Based on the matrix evolution path, obstacle avoidance flight optimization is performed on the position-replaced UAV matrix, resulting in an optimized UAV matrix.

[0046] In addition, when the sensitivity values ​​of the second cargo are all greater than the preset sensitivity values, the third cargo information corresponding to the non-evolving UAVs at each region boundary is determined one by one based on the boundary distance values. The third cargo information is then input into the preset cargo delivery standard model for identification, obtaining the third cargo sensitivity value corresponding to each cargo in the third cargo information. It is then determined whether the third cargo sensitivity value is greater than the preset sensitivity value. If it is, the center point position of the UAV matrix and the impact center point position of the impact point area are determined. The center point position and the impact center point position are then connected by a position line extension to obtain the matrix movement vector line. Based on the movement vector line, a matrix obstacle avoidance route is generated, and the UAV matrix is ​​controlled to move along the flight path.

[0047] In this embodiment, the preset cargo delivery standard model is a pre-established evaluation model based on various attributes of the cargo (such as fragility, value, etc.). After inputting the first cargo information, second cargo information, and third cargo information into the model, the model assigns a sensitivity value to each cargo according to preset rules and algorithms. These three values—the first cargo sensitivity value, the second cargo sensitivity value, and the third cargo sensitivity value—are used to measure the cargo's sensitivity to collisions or changes in flight status. The preset sensitivity value is a pre-set threshold used to determine whether a cargo is highly sensitive. By comparing the first cargo sensitivity value, the second cargo sensitivity value, and the third cargo sensitivity value with the preset sensitivity value, it is possible to determine which cargoes are more sensitive to collisions or flight adjustments.

[0048] Step S15: Based on the optimized and adjusted UAV matrix, control the UAV matrix to perform aerial obstacle avoidance evolution.

[0049] It is worth noting that when the drone matrix is ​​detected to have passed through an aerial obstacle, the optimized drone matrix is ​​backtracked according to the matrix evolution path or the flight path of the drone matrix is ​​backtracked according to the movement vector line to obtain the backtracked drone matrix.

[0050] In this embodiment, information about the drone swarm and the delivery cargo information carried by each drone is acquired. Based on this information, the overall situation of the drone swarm and the specific tasks of each drone can be accurately understood, ensuring the orderly conduct of the entire delivery process and avoiding chaos caused by missing or incorrect information, thus improving the initial accuracy and reliability of the delivery task. The application data and application number of each drone are determined based on the acquired drone swarm information. By clearly defining the application data and number, precise positioning and parameter setting of individual drones can be achieved, laying the foundation for building an orderly drone matrix and making subsequent matrix operations more accurate and efficient. A drone matrix is ​​constructed based on the drone application numbers, and the delivery cargo information is bound to the drone matrix to obtain drone matrix data. Constructing a drone matrix integrates dispersed drones into an organic whole. Through reasonable matrix arrangement, collaborative operations between drones can be achieved, improving delivery efficiency. Binding the delivery cargo information to the matrix makes the position and task of each drone in the matrix clearer, avoiding confusion in task allocation. The obtained drone matrix data provides comprehensive information for subsequent real-time monitoring and adjustment, ensuring that drones can be accurately scheduled according to the actual situation during the delivery process, improving the flexibility and responsiveness of the entire delivery system. Based on the drone matrix, a reconnaissance drone layer is determined, and real-time field-of-view images of the drone matrix during cargo delivery are collected from this layer. By acquiring these real-time images, the aerial situation surrounding the drone matrix is ​​promptly obtained, including information on potential obstacles and weather changes. These real-time images provide intuitive data for subsequent obstacle analysis and obstacle avoidance decisions, enabling drones to accurately perceive their surroundings in complex environments, detect potential hazards in advance, and provide strong support for safe delivery, effectively reducing risks during the delivery process. Obstacle analysis is performed on the real-time field-of-view images and drone application data to determine if there are any aerial obstacles interfering with the drone matrix. If so, static and dynamic analysis of the aerial obstacles is conducted, and the drone matrix is ​​optimized for obstacle avoidance flight based on the analysis results and drone matrix data, resulting in an optimized drone matrix. Obstacle analysis can quickly and accurately identify threats, while static and dynamic analysis further reveals the movement state of the obstacles. By combining obstacle avoidance optimization with UAV matrix data, the system can plan the optimal obstacle avoidance path for each UAV based on actual conditions. This allows the UAV matrix to efficiently adjust its flight status when encountering obstacles, ensuring the safety of delivered goods while avoiding collisions. This guarantees the smooth progress of delivery tasks and improves delivery safety and reliability. The optimized UAV matrix control system evolves to handle aerial obstacle avoidance, ensuring that the UAVs accurately execute obstacle avoidance maneuvers according to the predetermined plan. This not only successfully avoids obstacles but also minimizes the impact on delivered goods, maintaining the efficiency and stability of delivery.

[0051] Furthermore, to ensure the efficiency of signal reception by the UAVs, this application collects network communication information used to control the UAV matrix in real time and performs network status detection on the network communication information to determine whether the network status in the network communication information meets the preset network status. If not, it acquires regional network information within a preset range centered on the UAV matrix, determines network channel information based on the regional network information, and determines network channel groups other than the current application channel in the network communication information based on the network channel information. It then performs signal transmission tests on each network channel group to obtain the signal transmission rate corresponding to each network channel. The signal transmission rates are then filtered and combined, and the network channels are reassembled according to the transmission order corresponding to each signal transmission rate in the filtered and combined network channel group information to obtain a spliced ​​network channel for sending control commands to the UAV matrix.

[0052] The above embodiments describe a cooperative dynamic obstacle avoidance method for UAV swarms from the perspective of method flow. The following embodiments describe a cooperative dynamic obstacle avoidance system for UAV swarms from the perspective of virtual modules or virtual units. For details, please refer to the following embodiments.

[0053] This application provides a cooperative dynamic obstacle avoidance system 20 for unmanned aerial vehicle (UAV) swarms, such as... Figure 2 As shown, Figure 2 This is a schematic diagram of a cooperative dynamic obstacle avoidance system for a drone swarm, provided as an embodiment of this application. The system 20 specifically may include: Information acquisition module 21 is used to acquire information about the drone swarm and the delivery information carried by each drone in the drone swarm; Application determination module 22 is used to determine the application data and application number of each drone based on the drone swarm information; The matrix generation module 23 is used to construct a drone matrix based on the drone application number and bind the delivery cargo information to the drone matrix to obtain drone matrix data; The image acquisition module 24 is used to determine the reconnaissance drone layer based on the drone matrix, and to acquire real-time field-of-view images of the drone matrix during the cargo delivery process based on the reconnaissance drone layer; The obstacle analysis module 25 is used to perform obstacle analysis on real-time field-of-view images and UAV application data to determine whether there are aerial obstacles in the real-time field-of-view images that interfere with the UAV matrix. If there are, static and dynamic analysis of the aerial obstacles is performed, and the UAV matrix is ​​optimized and adjusted for obstacle avoidance flight based on the analysis results and UAV matrix data to obtain the optimized UAV matrix. The obstacle avoidance optimization module 26 is used to control the UAV matrix to perform aerial obstacle avoidance evolution based on the optimized and adjusted UAV matrix.

[0054] In one possible implementation of this application embodiment, when the obstacle analysis module 25 performs obstacle analysis on the real-time field-of-view image and UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix, it is specifically used for: A 3D model of the drone matrix during flight is built based on drone application data. Based on the delivery cargo information, the delivery cargo is processed in three dimensions. The resulting three-dimensional cargo model is then fitted to the three-dimensional drone model according to the loading correspondence and preset proportions to obtain the delivery three-dimensional model. Image fusion processing is performed on the real-time field-of-view image to obtain the processed real-time field-of-view image; The processed real-time field-of-view image is input into a preset obstacle model for detection to determine whether there are obstacles in the real-time field-of-view image. If there are obstacles, the relative positional relationship between the obstacle and the reconnaissance drone layer in the real-time field-of-view image is determined, and whether the obstacle has reached the preset flight range of the reconnaissance drone layer is determined based on the relative positional relationship. If the obstacle does not reach the preset flight range of the reconnaissance drone layer, the obstacle evolution at different time points is determined, the obstacle evolution is predicted and extrapolated, the future obstacle evolution between the obstacle and the reconnaissance drone layer is determined in the future time period, a three-dimensional model of obstacle evolution is built based on the future obstacle evolution, and the three-dimensional model of evolution is fitted with the delivery three-dimensional model according to the relative position relationship to obtain the monitoring three-dimensional model. The presence of aerial obstacles that could interfere with the UAV matrix is ​​determined based on the monitoring 3D model.

[0055] In another possible implementation of this application embodiment, when the obstacle analysis module 25 performs image fusion processing on the real-time field-of-view image to obtain the processed real-time field-of-view image, it is specifically used for: Edge filtering is performed on the real-time field-view image to obtain a first-level layer and a second-level layer. The first-level layer is the basic outline layer of the scene in the real-time field-view image, and the second-level layer is the detailed outline layer of the scene in the real-time field-view image. The first-level layer is divided according to a preset division rule to obtain multiple layer sub-regions; Collect the sub-region histogram and sub-region flatness corresponding to each layer's sub-region, as well as the global histogram and global flatness corresponding to the first-level layer. Based on the global histogram and global flatness, determine the first-level tone curve and brightness enhancement curve corresponding to the first-level layer. Based on the sub-region histogram and sub-region flatness corresponding to each sub-region of the layer, determine the secondary tone curve and curve blending weight corresponding to each sub-region, and then blend the primary tone curve and the secondary tone curve to obtain the tone blending curve. The color blending curve and the brightness enhancement curve are blended based on the curve blending weight, and the view adjustment curve after blending is used to adjust each sub-region in the first-level layer to obtain the adjusted first-level layer. The adjusted first-level layer and the second-level layer are then merged to obtain the processed real-time view image.

[0056] In another possible implementation of this application embodiment, when the obstacle analysis module 25 performs static and dynamic analysis on aerial obstacles and optimizes the drone matrix for obstacle avoidance flight based on the analysis results and drone matrix data to obtain an optimized drone matrix, it is specifically used for: Based on the monitoring 3D model, the collision point area between aerial obstacles and the UAV matrix in the future time period is determined, and the dynamically adjusted evolution UAVs involved in the UAV matrix are determined based on the collision point area. Based on the drone matrix data, the first cargo information corresponding to each evolved drone is determined, and the first cargo information is input into the preset cargo delivery standard model for identification, thereby obtaining the first cargo sensitivity value corresponding to each cargo in the first cargo information. Determine whether the sensitivity value of the first cargo is greater than the preset sensitivity value. If so, take the cargo corresponding to the sensitivity value of the first cargo as the marker point, measure the boundary distance value between the cargo and the boundary of each area of ​​the impact point, and determine the minimum boundary distance value and the target area boundary corresponding to the minimum boundary distance value based on the boundary distance value. Based on the UAV matrix data, the second cargo information corresponding to the non-evolving UAV located at the boundary of the target area is determined, and the second cargo information is input into the preset cargo delivery standard model for identification to obtain the second cargo sensitivity value corresponding to each cargo in the second cargo information. Determine whether the sensitivity value of the second cargo is greater than the preset sensitivity value. If not, calculate the relative distance between each non-evolved UAV and the marker point in the boundary of the target area. Based on the relative distance, determine the target non-evolved UAV that is closest to the marker point. Generate a position replacement command based on the target non-evolved UAV and the UAV corresponding to the marker point. Control the target non-evolved UAV and the UAV corresponding to the marker point to perform position replacement, and obtain the UAV matrix after position replacement. By performing an interleaved evolution analysis on the evolved drones in the position-replaced drone matrix and the drones in the position-replaced drone matrix that are at the boundary of the non-target area, the evolution path of the matrix is ​​obtained. Based on the matrix evolution path, the drone matrix after position replacement is optimized for obstacle avoidance flight, resulting in the optimized drone matrix.

[0057] In another possible implementation of this application embodiment, when the obstacle analysis module 25 determines whether the second cargo sensitivity value is greater than a preset sensitivity value, it is specifically used for: If the sensitivity values ​​of the second cargo are all greater than the preset sensitivity values, the third cargo information corresponding to the non-evolving drones at each region boundary is determined one by one according to the boundary distance value. The third cargo information is then input into the preset cargo delivery standard model for identification to obtain the third cargo sensitivity value corresponding to each cargo in the third cargo information. Determine whether the sensitivity value of the third cargo is greater than the preset sensitivity value. If it is still greater than the preset sensitivity value, determine the center point position of the UAV matrix and the impact center point position of the impact point area, and connect the center point position and the impact center point position by extending the position line to obtain the matrix movement vector line. The obstacle avoidance route is generated based on the moving vector line matrix, and the UAV matrix is ​​controlled to move along the flight path.

[0058] In another possible implementation of this application embodiment, system 20 further includes: a matrix backtracking module, wherein, The matrix backtracking module is used to backtrack the optimized drone matrix based on the matrix evolution path or the flight path of the drone matrix based on the movement vector line when the drone matrix is ​​detected to have passed through an aerial obstacle, so as to obtain the backtracked drone matrix.

[0059] In another possible implementation of this application embodiment, system 20 further includes: a channel combining module, a channel testing module, and a channel reassembly module, wherein... The channel combination module is used to collect network communication information for controlling the UAV matrix, and to perform network status detection on the network communication information to determine whether the network status in the network communication information meets the preset network status. If it does not meet the preset network status, the module acquires the regional network information within the preset range centered on the UAV matrix, determines the network channel information based on the regional network information, and determines the network channel group other than the current application channel in the network communication information based on the network channel information. The channel test module is used to perform signal transmission tests on each network channel group to obtain the signal transmission rate corresponding to each network channel. The channel reassembly module is used to filter and combine signal transmission rates, and reassemble the network channels according to the transmission order corresponding to each signal transmission rate in the filtered and combined network channel group information to obtain the spliced ​​network channel for sending control commands to the UAV matrix.

[0060] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described cooperative dynamic obstacle avoidance system 20 for UAV swarms can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0061] This application provides an electronic device, such as... Figure 3 As shown, Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 The illustrated electronic device 300 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of this electronic device 300 does not constitute a limitation on the embodiments of this application.

[0062] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0063] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The symbol is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0064] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0065] The memory 303 is used to store application code that executes the solution of this application, and its execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0066] Among them, electronic devices include, but are not limited to: ruggedized computers, monitoring and management systems and other terminals, and mobile phones, laptops, PADs and other devices can also be used in situations where mobile devices are not restricted. Figure 3 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0067] This application provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the aforementioned method embodiments.

[0068] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0069] The above are only some embodiments of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms, characterized in that, include: Obtain information about the drone swarm and the delivery cargo carried by each drone in the swarm; Based on the drone swarm information, determine the application data and drone application number for each drone; A drone matrix is ​​constructed based on the drone application ID, and the delivery cargo information is bound to the drone matrix to obtain drone matrix data; The reconnaissance drone layer is determined based on the drone matrix, and real-time field-of-view images of the drone matrix during the cargo delivery process are collected based on the reconnaissance drone layer. Obstacle analysis is performed on the real-time field-of-view image and the UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix. If there are, static and dynamic analysis is performed on the aerial obstacles, and obstacle avoidance flight optimization is performed on the UAV matrix based on the analysis results and the UAV matrix data to obtain the optimized UAV matrix. Based on the optimized and adjusted UAV matrix control, the UAV matrix evolves to avoid aerial obstacles.

2. The cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The step of performing obstacle analysis on the real-time field-of-view image and the UAV application data to determine whether there are aerial obstacles in the real-time field-of-view image that could interfere with the UAV matrix includes: A 3D model of the drones in flight is constructed based on the drone application data. Based on the delivery cargo information, the delivery cargo is subjected to three-dimensional simulation processing, and the processed cargo three-dimensional model is fitted with the drone three-dimensional model according to the loading correspondence and preset ratio to obtain the delivery three-dimensional model. The real-time field-of-view image is subjected to image fusion processing to obtain a processed real-time field-of-view image; The processed real-time field-of-view image is input into a preset obstacle model for detection to determine whether there is an obstacle in the real-time field-of-view image. If there is an obstacle, the relative positional relationship between the obstacle in the real-time field-of-view image and the reconnaissance drone layer is determined, and the relative positional relationship is used to determine whether the obstacle has reached the preset flight range of the reconnaissance drone layer. If the obstacle does not reach the preset flight range of the reconnaissance drone layer, the obstacle evolution at different time points is determined, the obstacle evolution is predicted and extrapolated, the future obstacle evolution between the obstacle and the reconnaissance drone layer is determined in the future time period, a three-dimensional model of the obstacle evolution is built based on the future obstacle evolution, and the three-dimensional model of the evolution is fitted to the delivery three-dimensional model according to the relative position relationship to obtain the monitoring three-dimensional model; The presence of aerial obstacles that could interfere with the UAV matrix is ​​determined based on the monitoring 3D model.

3. The cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms according to claim 2, characterized in that, The step of performing image fusion processing on the real-time field-of-view image to obtain the processed real-time field-of-view image includes: Edge filtering is performed on the real-time field-view image to obtain a first-level layer and a second-level layer. The first-level layer is the basic outline layer of the scene in the real-time field-view image, and the second-level layer is the detailed outline layer of the scene in the real-time field-view image. The first-level layer is divided according to a preset division rule to obtain multiple layer sub-regions; Collect the sub-region histogram and sub-region flatness corresponding to each layer sub-region, as well as the global histogram and global flatness corresponding to the first-level layer, and determine the first-level tone curve and brightness enhancement curve corresponding to the first-level layer based on the global histogram and global flatness. Based on the sub-region histogram and sub-region flatness corresponding to each sub-region of the layer, determine the secondary tone curve and curve blending weight corresponding to each sub-region, and then blend the primary tone curve and the secondary tone curve to obtain the tone blending curve. Based on the curve blending weight, the tone blending curve and the brightness enhancement curve are blended, and each sub-region in the first-level layer is adjusted according to the blended field of view adjustment curve to obtain the adjusted first-level layer. The adjusted first-level layer and the second-level layer are then fused to obtain the processed real-time field-of-view image.

4. The cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms according to claim 2, characterized in that, The process involves performing static and dynamic analysis on the aerial obstacles, and then optimizing the drone matrix for obstacle avoidance based on the analysis results and the drone matrix data, resulting in an optimized drone matrix, including: Based on the monitoring 3D model, the collision point area between the aerial obstacle and the UAV matrix in the future time period is determined, and the dynamically adjusted evolving UAVs involved in the UAV matrix are determined based on the collision point area. Based on the drone matrix data, the first cargo information corresponding to each evolved drone is determined, and the first cargo information is input into a preset cargo delivery standard model for identification to obtain the first cargo sensitivity value corresponding to each cargo in the first cargo information. Determine whether the first cargo sensitivity value is greater than a preset sensitivity value. If so, take the cargo corresponding to the first cargo sensitivity value as a marker point, measure the boundary distance value of each area boundary of the impact point area, and determine the minimum boundary distance value and the target area boundary corresponding to the minimum boundary distance value based on the boundary distance value. Based on the UAV matrix data, determine the second cargo information corresponding to the non-evolving UAV located at the boundary of the target area, and input the second cargo information into a preset cargo delivery standard model for identification to obtain the second cargo sensitivity value corresponding to each cargo in the second cargo information. Determine whether the second cargo sensitivity value is greater than a preset sensitivity value. If not, calculate the relative distance between each non-evolved UAV in the target area boundary and the marker point. Based on the relative distance, determine the target non-evolved UAV that is closest to the marker point. Generate a position replacement command based on the target non-evolved UAV and the UAV corresponding to the marker point. Control the target non-evolved UAV and the UAV corresponding to the marker point to perform position replacement, and obtain the UAV matrix after position replacement. The evolution of drones in the replaced drone matrix is ​​analyzed by interleaving them with the drones in the replaced drone matrix that are at the boundary of the non-target area, and the evolution path of the matrix is ​​obtained. Based on the matrix evolution path, the drone matrix after position replacement is optimized for obstacle avoidance flight, resulting in an optimized drone matrix.

5. A cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The step of determining whether the sensitivity value of the second cargo is greater than the preset sensitivity value includes: If the second cargo sensitivity value is greater than the preset sensitivity value, then the third cargo information corresponding to the non-evolving drone at each region boundary is determined one by one according to the boundary distance value. The third cargo information is then input into the preset cargo delivery standard model for identification to obtain the third cargo sensitivity value corresponding to each cargo in the third cargo information. Determine whether the sensitivity value of the third cargo is greater than the preset sensitivity value. If it is still greater than the preset sensitivity value, determine the center point position of the UAV matrix and the impact center point position of the impact point area, and connect the center point position and the impact center point position by extending the position line to obtain the matrix movement vector line. The obstacle avoidance route is generated based on the movement vector line, and the UAV matrix is ​​controlled to move along the flight path.

6. A cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms according to claim 5, characterized in that, The process of controlling the drone matrix to perform aerial obstacle avoidance based on the optimized and adjusted drone matrix also includes: When the drone matrix is ​​detected to have passed through an aerial obstacle, the optimized drone matrix is ​​backtracked according to the matrix evolution path or the flight path of the drone matrix is ​​backtracked according to the movement vector line to obtain the backtracked drone matrix.

7. A cooperative dynamic obstacle avoidance method for unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The process of controlling the drone matrix to perform aerial obstacle avoidance evolution based on the optimized and adjusted drone matrix also includes: The system collects network communication information for controlling the UAV matrix and performs network status detection on the network communication information to determine whether the network status in the network communication information meets the preset network status. If it does not meet the preset network status, the system obtains regional network information within a preset range centered on the UAV matrix, determines network channel information based on the regional network information, and determines network channel groups other than the current application channel in the network communication information based on the network channel information. The signal transmission rate of each network channel group was obtained by performing signal transmission tests on each network channel. The signal transmission rates are filtered and combined for analysis, and the network channels are reorganized according to the transmission order corresponding to each signal transmission rate in the filtered and combined network channel group information to obtain the spliced ​​network channel for sending control commands to the UAV matrix.

8. A cooperative dynamic obstacle avoidance system for unmanned aerial vehicle (UAV) swarms, characterized in that, include: The information acquisition module is used to acquire information about the drone swarm and the delivery cargo carried by each drone in the swarm. The application determination module is used to determine the application data and application number of each drone based on the drone swarm information. The matrix generation module is used to construct a drone matrix based on the drone application number and bind the delivery cargo information to the drone matrix to obtain drone matrix data; The image acquisition module is used to determine the reconnaissance drone layer based on the drone matrix, and to acquire real-time field-of-view images of the drone matrix during the cargo delivery process based on the reconnaissance drone layer; The obstacle analysis module is used to perform obstacle analysis on the real-time field-of-view image and the UAV application data, determine whether there are aerial obstacles in the real-time field-of-view image that interfere with the UAV matrix, and if so, perform static and dynamic analysis on the aerial obstacles, and optimize the UAV matrix for obstacle avoidance flight based on the analysis results and the UAV matrix data to obtain the optimized UAV matrix. The obstacle avoidance optimization module is used to control the drone matrix to perform aerial obstacle avoidance evolution based on the optimized and adjusted drone matrix.

9. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as any one of claims 1 to 7 for a cooperative dynamic obstacle avoidance method for a drone swarm.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed as any one of the cooperative dynamic obstacle avoidance methods for unmanned aerial vehicle swarms as described in claims 1 to 7.

Images