A method and system for cooperative search of a UAV cluster

Abstract

The application provides a UAV cluster cooperative search method and system, and relates to the technical field of UAV control. In the application, environment state information and real-time position of the UAV cluster are acquired, an coverage evaluation strategy is generated by using interactive historical decision data, and then a heterogeneous search sub-region is dynamically allocated. The environment perception segment is shared in real time, and the search trajectory confidence is reconstructed accordingly, so that the cooperative decision constraint in the strategy is updated asynchronously, the cluster cooperative decision rule is formed, and finally the distributed flight path is adjusted according to the rule, the adaptive cluster motion mode is formed, and the multi-round efficient iterative search of the target area is realized. The technical scheme provided by the application not only overcomes the defect of poor adaptability of the traditional pre-defined rule system, but also endows the UAV cluster with the ability to autonomously adjust the cooperative strategy when facing unknown threats and dynamic environments, and improves the intelligence of the search process and the robustness to environmental changes.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a method and system for collaborative search of UAV swarms. Background Technology

[0002] In applications such as reconnaissance in complex areas and disaster search and rescue, the use of drone swarms for collaborative search has become a key means to improve efficiency. This usually requires drone swarms to autonomously achieve efficient coverage search of a wide area in an unknown environment with potential dynamic threats. This places significant technical demands on the real-time environmental adaptability and intelligent collaborative decision-making capabilities of drone swarms.

[0003] Existing technical solutions use a collaborative control method based on predefined rules and virtual potential fields to pre-set rules for obstacle avoidance, swarm aggregation, and target area attraction for UAV swarms. By calculating the resultant force of virtual forces in real time, instantaneous motion control commands are generated for each UAV, enabling the swarm to maintain formation and avoid known obstacles during the search process.

[0004] However, this scheme has inherent flaws. Its collaborative search performance is highly dependent on the completeness and accuracy of the preset rules, making it difficult to cope with sudden threats or complex environmental changes that are not pre-modeled. Due to the lack of the ability to learn from the interaction history and optimize the collaborative strategy, the system's behavior pattern is relatively fixed, and it cannot improve its collaborative efficiency and adaptability in multiple search iterations, resulting in insufficient robustness of search in real dynamic environments. Summary of the Invention

[0005] This application provides a method and system for collaborative search of unmanned aerial vehicle (UAV) swarms, which solves the problems in the prior art that result in poor system adaptability, inability to learn and optimize collaborative strategies from historical interactions, and insufficient robustness in dynamic and unknown environments due to reliance on fixed predefined rules.

[0006] Firstly, this application provides a method for collaborative search in a swarm of unmanned aerial vehicles (UAVs), including:

[0007] Obtain environmental status information of the area to be searched and real-time location information of the drone swarm;

[0008] Based on the environmental state information and the real-time location information, a coverage assessment strategy is generated by interacting with historical decision data.

[0009] Based on the coverage evaluation strategy, heterogeneous search sub-regions are dynamically allocated;

[0010] In the heterogeneous search sub-region, environmental perception segments are shared in real time, and the confidence level of the search trajectory is reconstructed using the environmental perception segments;

[0011] Based on the confidence of the reconstructed search trajectory, the collaborative decision constraints in the coverage evaluation strategy are updated asynchronously to obtain the cluster collaborative decision rules;

[0012] Distributed trajectory adjustments are performed according to the cluster collaborative decision-making rules to form a cluster motion pattern to complete multiple rounds of iterative search of the area to be searched.

[0013] Optionally, environmental status information of the area to be searched and real-time location information of the drone swarm can be obtained, including:

[0014] Raw sensing data of the area to be searched is collected through an airborne sensing unit.

[0015] The raw sensing data is combined with the navigation data of the UAV swarm to obtain a local environment description;

[0016] The local environment description is exchanged through a wireless communication network to construct a global environment map;

[0017] Based on the global environment map, passable areas and potential threat areas in the area to be searched are identified to obtain environmental status information;

[0018] The real-time location information of the drone swarm is obtained by aggregating the real-time locations of the drone swarm through a wireless communication network.

[0019] Optionally, based on the environmental state information and the real-time location information, a coverage assessment strategy is generated by interacting with historical decision data, including:

[0020] Extract historical decision fragments that are similar to the environmental state information from the interactive historical decision data;

[0021] Based on the real-time location information, analyze the coverage of the executed areas of the historical decision segments in the area to be searched;

[0022] Based on the coverage of the executed regions, calculate the strategy preference vector for the region to be searched in the historical decision segments;

[0023] The strategy preference vector is fused with the environmental state information to generate a preliminary evaluation strategy;

[0024] The spatial distribution of the preliminary evaluation strategy is adjusted based on the real-time location information to generate a coverage evaluation strategy.

[0025] Optionally, heterogeneous search sub-regions are dynamically allocated according to the coverage evaluation strategy, including:

[0026] Based on the coverage evaluation strategy, the area to be searched is divided into multiple sub-regions with different coverage characteristics;

[0027] Based on the potential threat areas in the environmental status information, the sub-region is labeled with a dynamic threat level;

[0028] Based on the real-time location information, calculate the movement cost of the drone swarm to different sub-regions;

[0029] By fusing the coverage characteristics of different sub-regions, the dynamic threat level, and the mobility cost, the search adaptability of the drone swarm to different sub-regions can be obtained.

[0030] Based on the search fitness, multiple heterogeneous search sub-regions are dynamically allocated to the UAV cluster.

[0031] Optionally, environmental awareness segments are shared in real time within the heterogeneous search sub-regions, and the search trajectory confidence is reconstructed using these environmental awareness segments, including:

[0032] Within the heterogeneous search sub-region, the drone swarm continuously acquires environmental perception fragments;

[0033] The environmental perception segments are shared within the UAV cluster via a wireless communication network to form a shared perception data pool.

[0034] The environmental perception segments in the shared perception data pool are integrated with the environmental perception segments collected by the UAV cluster itself to generate comprehensive environmental perception information.

[0035] Based on the comprehensive environmental perception information, the actual search trajectory of the UAV cluster in the heterogeneous search sub-region is analyzed.

[0036] The actual search trajectory is compared with the expected trajectory in the coverage evaluation strategy, and the confidence level of the search trajectory is reconstructed based on the comparison results.

[0037] Optionally, based on the confidence level of the reconstructed search trajectory, the collaborative decision constraints in the coverage evaluation strategy are updated asynchronously to obtain cluster collaborative decision rules, including:

[0038] Based on the confidence level of the reconstructed search trajectory, the execution deviation of the coverage evaluation strategy in the heterogeneous search sub-region is analyzed;

[0039] Based on the execution deviation, calculate the amount of coordination strategy adjustment required by the drone swarm;

[0040] The adjustment amount of the collaborative strategy is combined with the real-time location information and the environmental state information to generate correction parameters;

[0041] The constraint terms corresponding to the collaborative decision-making constraints are asynchronously adjusted according to the correction parameters to form cluster collaborative decision-making rules.

[0042] Optionally, distributed trajectory adjustment is performed according to the cluster collaborative decision-making rules to form a cluster motion pattern to complete multiple rounds of iterative search of the area to be searched, including:

[0043] Based on the cluster collaborative decision-making rules and the real-time location information, a preliminary flight path is generated for the UAV cluster;

[0044] Based on the potential threat areas in the environmental status information, the initial flight path is modified to avoid threats, generating an individual modified flight path;

[0045] Based on the real-time status information of the individual corrected trajectory and neighboring UAVs, the interaction between the individual corrected trajectory is evaluated;

[0046] Based on the interaction effects and the cluster collaborative decision-making rules, the individual correction trajectory is adjusted to generate the final trajectory;

[0047] Based on the final flight path, the drone swarm is controlled to fly along the final flight path to form a swarm movement pattern;

[0048] Based on the cluster movement pattern, a new round of search is performed on the area to be searched to complete multiple rounds of iterative search.

[0049] Secondly, this application provides a drone swarm cooperative search system, comprising:

[0050] The acquisition module is used to acquire environmental status information of the area to be searched and the real-time location information of the drone cluster;

[0051] The generation module is used to generate a coverage assessment strategy based on the environmental state information and the real-time location information by interacting with historical decision data;

[0052] The allocation module is used to dynamically allocate heterogeneous search sub-regions according to the coverage evaluation strategy;

[0053] The reconstruction module is used to share environmental perception segments in real time in the heterogeneous search sub-regions and reconstruct the search trajectory confidence using the environmental perception segments;

[0054] The update module is used to asynchronously update the collaborative decision constraints in the coverage evaluation strategy based on the confidence of the reconstructed search trajectory, so as to obtain the cluster collaborative decision rules.

[0055] The adjustment module is used to perform distributed trajectory adjustment according to the cluster collaborative decision-making rules to form a cluster motion pattern to complete multiple rounds of iterative search of the area to be searched.

[0056] Thirdly, this application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement a drone swarm cooperative search method as described in the first aspect above.

[0057] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a computer, implements a drone swarm cooperative search method as described in the first aspect.

[0058] This application utilizes interactive historical decision data to generate coverage assessment strategies and dynamically reconstructs search trajectory confidence based on real-time environmental perception fragments. This enables UAV swarms to learn from past collaborative search experiences and continuously optimize their decision-making logic, overcoming the poor adaptability of traditional predefined rule systems. It empowers UAV swarms to autonomously adjust their collaborative strategies when facing unknown threats and dynamic environments, thereby significantly improving the intelligence of the search process and its robustness to environmental changes.

[0059] Furthermore, by asynchronously updating collaborative decision constraints to form cluster collaborative decision rules, and executing distributed trajectory adjustments accordingly, each UAV can maintain a high degree of coordination with the overall UAV cluster while avoiding threats. This ultimately leads to an efficient and adaptive cluster movement mode, ensuring that the UAV cluster can continuously optimize its coverage path in multiple rounds of iterative search, thereby achieving comprehensive and efficient area search in complex environments and effectively improving the quality and efficiency of mission completion.

[0060] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 A flowchart of a UAV swarm cooperative search method provided in this application is shown;

[0063] Figure 2 This paper presents a schematic diagram of the structure of a UAV swarm cooperative search system provided in this application;

[0064] Figure 3 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0065] To enable those skilled in the art to better understand the present application, the technical solution of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0066] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0067] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and 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.

[0068] Figure 1 A flowchart of a UAV swarm cooperative search method is provided in this application, such as... Figure 1 As shown, the method includes:

[0069] Step 101: Obtain environmental status information of the area to be searched and real-time location information of the drone cluster.

[0070] Optionally, step 101 may specifically include the following steps:

[0071] Step 1011: Collect raw sensing data of the area to be searched through the airborne sensing unit;

[0072] Step 1012: Combine the original perception data with the navigation data of the UAV swarm to obtain a local environment description;

[0073] Step 1013: Exchange the local environment descriptions through a wireless communication network to construct a global environment map;

[0074] Step 1014: Based on the global environment map, identify passable areas and potential threat areas in the area to be searched to obtain environmental status information;

[0075] Step 1015: The real-time location of the drone cluster is aggregated through the wireless communication network to obtain the real-time location information of the drone cluster.

[0076] In the above steps, the area to be searched refers to the pre-defined geographical area that needs to be explored and surveyed by the drone swarm;

[0077] Environmental status information refers to a digital description of the terrain, obstacles, and safety conditions within the search area, specifically including the location and extent of passable and potentially threatening areas, which is used to guide the safe path planning of drones;

[0078] Real-time location information refers to the precise geographic coordinates of each drone in a drone swarm at a certain moment, including data such as latitude, longitude, and altitude, which is used to determine the instantaneous distribution and movement status of the drone swarm.

[0079] Airborne sensing units refer to the combination of sensors installed on a drone, such as visible light cameras, infrared sensors, and lidar, used to actively detect and collect physical feature data of its surrounding environment.

[0080] Raw sensing data refers to the unprocessed raw sensor signals directly collected by the airborne sensing unit, such as image pixel matrix, laser point cloud, infrared thermal map, etc., which are the most direct reflection of the environment;

[0081] Navigation data for drone swarms refers to the motion state data of drones obtained through devices such as global satellite navigation systems and inertial measurement units, including position, speed, heading angle, and attitude angle, which are used for positioning and navigation;

[0082] Local environment description refers to the structured information fragments about the surrounding environment formed by a single UAV after it correlates and fuses the raw perception data it acquires with its own navigation data in time and space.

[0083] Wireless communication networks refer to the data links that connect the members of a drone swarm, enabling them to exchange information with each other. They typically employ self-organizing network technology.

[0084] A global environment map refers to a unified environment model that covers the entire search area by integrating the local environment descriptions provided by all drones in a drone swarm, under a central node or distributed consensus mechanism.

[0085] Passable areas refer to the spaces marked as unobstructed and safe for drones to fly in the global environment map;

[0086] Potential threat areas refer to areas identified in the global environment map that contain uncertain factors that may affect the safety of drones, such as dynamic obstacles, signal interference sources, or severe weather.

[0087] The real-time location of a drone swarm refers to the set of real-time location information of all drones within the swarm, which is collected and updated through a wireless communication network.

[0088] In this embodiment, the surrounding environment is first scanned synchronously by multiple sensors carried by each drone in the drone swarm to collect raw perception data containing various information such as images and distances.

[0089] Secondly, each drone runs a data fusion algorithm to perform spatiotemporal alignment and fusion processing on the raw perception data collected at this moment and the precise position, attitude and other navigation data provided by its own navigation system, thereby generating a local environmental description that describes the surrounding environment.

[0090] Next, the drone swarm uses the wireless communication network established between them to send out the local environment descriptions they generate and receive local environment descriptions from other drones through broadcasting or designated routing. Through a distributed consensus algorithm, all the received local environment descriptions are stitched together, denoised, and fused to finally construct a global environment map.

[0091] Then, a threat identification model is applied to the global environment map to automatically analyze the features in the map. Areas without obstacles and with stable structures are marked as passable areas, while areas with features such as moving objects, signal blind spots, or sudden terrain changes are marked as potential threat areas. The passable areas and potential threat areas together constitute the environmental status information.

[0092] Finally, each drone continuously reports its real-time location via the wireless communication network. The location reports of all drones are collected and integrated through a distributed protocol to obtain the real-time location information of the drone swarm.

[0093] For example, suppose there is a cluster of five drones, B1 to B5, that are sent to a mountainous area C to search for missing persons. First, the lidar and visual sensors on each drone are activated to scan the mountainous terrain below and obtain raw point cloud and image data.

[0094] Secondly, the drone swarm uses its own navigation data to convert the original point cloud and image data into terrain segments with coordinate labels that are referenced to its own position;

[0095] Next, UAV B1, acting as a temporary communication hub, received terrain fragments from all its companions via a wireless communication network. B1 then ran a map-building program, piecing these fragments together like a mosaic. Figure 1 These elements, when combined, form a complete three-dimensional map covering most of the mountainous area C.

[0096] Then, this 3D map is analyzed, and flat forest clearings are marked as passable areas, while canyons with violently fluctuating radar echoes and deep forests with weak signals are marked as potential threat areas. The passable areas and potential threat areas together constitute the environmental status information.

[0097] Finally, each drone reports its own coordinates once per second. B1 then summarizes the drones' position information to obtain the precise real-time positions of the five drones.

[0098] This step enables real-time, collaborative perception and digital modeling of the mission area environment, accurately distinguishing between safe and risky areas. Simultaneously, it achieves synchronous and accurate mastery of the positions of all members of the UAV swarm, providing an accurate and reliable environmental cognition foundation and global situational awareness data support for subsequent intelligent decision-making and collaborative control. This is a prerequisite for the entire search mission to be carried out autonomously, safely, and efficiently.

[0099] Step 102: Based on the environmental state information and the real-time location information, generate a coverage assessment strategy by interacting with historical decision data.

[0100] Optionally, step 102 may specifically include the following steps:

[0101] Step 1021: Extract historical decision fragments similar to the environmental state information from the interactive historical decision data;

[0102] Step 1022: Based on the real-time location information, analyze the coverage of the executed areas of the historical decision segments in the area to be searched;

[0103] Step 1023: Calculate the strategy preference vector for the search region based on the coverage of the executed region;

[0104] Step 1024: Fuse the strategy preference vector with the environmental state information to generate a preliminary evaluation strategy;

[0105] Step 1025: Adjust the spatial distribution of the preliminary evaluation strategy based on the real-time location information to generate a coverage evaluation strategy.

[0106] In the above steps, interactive historical decision data refers to the complete work logs recorded by the drone swarm during similar search tasks in the past. The content includes the environmental information at the time, the action instructions taken by the drone swarm, and the results of the actions. It is used to provide experience reference for current decision-making and is obtained by recording the entire process of historical tasks.

[0107] Coverage assessment strategy refers to the behavioral guidelines that guide drone swarms on how to efficiently allocate search resources. It clearly defines under what circumstances which areas should be searched first in order to maximize the search range, and is generated through a complete processing procedure.

[0108] Historical decision fragments refer to historical task records that are highly similar to the current environmental state information and are selected from a large amount of interactive historical decision data. They are used as analog templates for the current decision and obtained through similarity matching.

[0109] The executed area coverage refers to the proportion of the area actually flown by the drone swarm in the historical decision segment analysis relative to the total mission area at that time. It is used to measure the effectiveness of the historical decision segment and is calculated by the area coverage analysis algorithm.

[0110] The area to be searched refers to the part of the current area to be searched that has not yet been explored by the drone swarm, which is obtained by comparing the global environment map with the area already flown.

[0111] The strategy preference vector refers to a tendency index expressed in mathematical vector form, which quantifies the degree of preference of the strategy represented by historical decision segments for exploring different search areas. It is generated through a preference quantification model.

[0112] The preliminary evaluation strategy refers to a draft search strategy that has not yet been finely adjusted, formed by initially combining historical experience with the current situation, and is obtained through strategy fusion operations.

[0113] In this embodiment, the environmental state information is used as a query condition by comparing the similarity of environmental features. A comprehensive search is performed in the interactive historical decision database to find the task scenario with the most similar environmental features encountered in history and its corresponding decision record as a historical decision fragment.

[0114] Secondly, by calculating the coverage of historical flight trajectories and combining the real-time location information of the drone swarm, a detailed analysis of historical decision segments is conducted. By overlaying and comparing the drone flight trajectories guided by historical decisions with the map of the current search area, the actual completed area coverage is obtained.

[0115] Next, by establishing a preference intensity mapping relationship, the coverage value is transformed into the exploration preference intensity of different search areas. Based on the coverage of the executed areas, a strategy preference vector is generated for the historical decision fragments. The preference vector corresponding to the historical decision with higher coverage is usually considered to be more worthy of reference.

[0116] Then, the strategy preference vector and environmental state information are fused by a weighted integration method, taking into account the validity of each historical strategy and the specific constraints of the current environment, to generate a preliminary evaluation strategy.

[0117] Finally, by using a spatial distribution optimization method, the initial evaluation strategy is modified based on the real-time location information of the drone swarm, generating a spatially executable coverage evaluation strategy.

[0118] Following the implementation case of the previous step, we first searched the historical task database for successful cases similar to the current mountainous environment C, and found historical task records that efficiently completed the search in another similar mountainous terrain.

[0119] Secondly, historical mission records were used as a reference template to analyze and evaluate the effectiveness of the flight strategies used at that time in ultimately covering most of the mission area.

[0120] Next, based on the effectiveness of covering most of the mission area, it was calculated that there is a strong tendency to explore key terrain such as ridgelines;

[0121] Then, the calculated strong tendency was combined with the specific three-dimensional map of mountain area C, and in particular, considering the identified potential threat areas, a preliminary search plan was formulated that prioritized patrolling open ridgelines and avoiding dangerous canyons.

[0122] Finally, based on the actual distribution of UAVs B1 to B5 in mountainous area C at this moment, the search plan was spatially assigned specific tasks. UAVs B1 and B2, located near the valley on the north side, were instructed to work together to search the ridgeline on the north slope, while UAVs B3 and B4, located on the south side, were responsible for the south slope. UAV B5, located in the center, served as a mobile force, ready to provide reinforcement scanning for key areas or emergencies.

[0123] This step leverages historical mission experience to generate an intelligent search strategy that is highly adaptable to the current specific environment and the real-time status of the swarm. This enables the swarm's search actions to move beyond blind or fixed approaches, allowing it to learn from experience, dynamically assess the search value of different areas, and rationally allocate swarm resources. This lays the foundation for efficient and collaborative regional coverage in the future.

[0124] Step 103: Dynamically allocate heterogeneous search sub-regions according to the coverage evaluation strategy.

[0125] Optionally, step 103 may specifically include the following steps:

[0126] Step 1031: Based on the coverage evaluation strategy, the area to be searched is divided into multiple sub-regions with different coverage characteristics;

[0127] Step 1032: Based on the potential threat areas in the environmental status information, mark the dynamic threat level of the sub-region;

[0128] Step 1033: Calculate the movement cost of the drone cluster to different sub-regions based on the real-time location information;

[0129] Step 1034: The coverage characteristics of different sub-regions, the dynamic threat level, and the mobility cost are fused to obtain the search adaptability of the drone swarm to different sub-regions;

[0130] Step 1035: Based on the search fitness, dynamically allocate multiple heterogeneous search sub-regions to the UAV cluster.

[0131] In the above steps, heterogeneous search sub-regions refer to different blocks to be explored, divided according to differences in search value and task difficulty, used to achieve a fine division of search tasks, and are obtained through the allocation process;

[0132] Different coverage characteristics refer to the differences in the priority of search assigned to each sub-region in the coverage assessment strategy, which are used to identify the search value of the region and are obtained through strategy analysis;

[0133] A sub-region refers to a smaller management unit formed by dividing the entire search area according to certain rules. It is used as the basic unit for task allocation and is obtained through region division operations.

[0134] Dynamic threat level refers to a quantitative indicator of the degree of danger assigned to each sub-region in real time based on potential threat areas in environmental status information. It is used to assess the risk of performing tasks and is obtained through a threat labeling process.

[0135] Mobility cost refers to the estimated cost required for a drone swarm to fly from its current location to a designated sub-area. It is used to measure the spatial accessibility of mission execution and is obtained through a cost calculation model.

[0136] Search fitness is a comprehensive evaluation index that quantifies the suitability of a drone swarm to perform a search task in a certain sub-region. It serves as the core basis for dynamic allocation and is obtained through multi-feature fusion calculation.

[0137] In this embodiment, firstly, using natural boundary recognition technology, and based on the guidance regarding the priority of different regions in the coverage assessment strategy, the entire area to be searched is divided into multiple sub-regions with different areas and shapes.

[0138] Secondly, by using threat proximity analysis and risk propagation models, combined with the potential threat areas marked in the environmental status information, the dynamic threat level is calculated for each divided sub-region. The relative position and threat intensity of the sub-region and the threat area are considered. The closer the sub-region is to the threat or the higher the threat intensity, the higher its dynamic threat level.

[0139] Next, using a path planning and energy consumption estimation model, based on the real-time location information of the drone swarm, the movement cost for each drone to reach the preset entry point of each sub-region is calculated. This movement cost typically takes into account factors such as the expected flight distance, time consumption, and energy loss.

[0140] Then, through multi-objective decision-making and weighted fusion methods, the coverage characteristics, dynamic threat level and the movement cost of the drone swarm to reach the sub-region of each sub-region are integrated and calculated. The calculation process balances search value, task risk and execution efficiency to obtain the search fitness of the drone swarm for each sub-region.

[0141] Finally, through task assignment and optimization algorithms, heterogeneous search sub-regions are dynamically allocated to each drone based on the search fitness of the drone swarm across all sub-regions, so that the overall search fitness of the entire drone swarm reaches the optimal level.

[0142] Following the implementation case of the previous step, the north and south slopes of mountainous area C are further subdivided into several more specific sub-regions, including the eastern section of the ridgeline on the north slope, the forest area on the north slope, and the cliff area on the south slope. Each sub-region is given different coverage characteristics according to its terrain and strategic priority.

[0143] Next, retrieve the 3D map of mountainous area C, read the potential threat areas marked therein, including the radar jamming canyon located on the north slope, calculate the dynamic threat level for each sub-region, increase the level of sub-regions closer to the canyon, and decrease the level of flat areas far from the threat.

[0144] Then, based on the precise location coordinates of UAVs B1 to B5 at this moment, the time and energy consumption required for them to fly to each sub-region were calculated. It was found that the movement cost of B1 to the eastern section of the north slope ridgeline was relatively low, while the movement cost to the south slope cliff area was very high.

[0145] Then, a decision-making model is established, and the coverage characteristics of the sub-region, the dynamic threat level, and the mobility cost of the drone swarm to reach the sub-region are substituted into the formula for comprehensive scoring to obtain the search fitness.

[0146] Finally, the task allocation algorithm was run, and global optimization was performed based on search fitness. As a result, UAV B1 was officially assigned to search the eastern section of the north slope ridgeline, B2 was responsible for the forest area on the north slope, B3 was responsible for the gentle slope area on the west side of the south slope, B4 was responsible for the cliff area on the east side of the south slope, and B5, as a mobile unit, showed good flexibility in each area according to search fitness, so it was temporarily suspended and ready to receive temporary instructions.

[0147] This step enables intelligent deconstruction of complex search areas and precise matching of task resources. It not only considers the search value of the area, but also comprehensively evaluates real-time environmental threats and the actual arrival cost of the drones, thereby generating a globally optimized dynamic task allocation scheme. This allows each drone to be dispatched to a high-value, low-risk area that best suits its current state, significantly improving the efficiency of drone swarm collaborative search and the safety of task execution.

[0148] Step 104: Share environmental awareness segments in real time in the heterogeneous search sub-regions, and reconstruct the search trajectory confidence using the environmental awareness segments.

[0149] Optionally, step 104 may specifically include the following steps:

[0150] Step 1041: Within the heterogeneous search sub-region, the UAV cluster continuously collects environmental perception segments;

[0151] Step 1042: The environmental perception segments are shared within the UAV cluster via a wireless communication network to form a shared perception data pool;

[0152] Step 1043: Integrate the environmental perception segments in the shared sensing data pool with the environmental perception segments collected by the UAV cluster itself to generate comprehensive environmental perception information.

[0153] Step 1044: Based on the comprehensive environmental perception information, analyze the actual search trajectory of the UAV cluster in the heterogeneous search sub-region;

[0154] Step 1045: Compare the actual search trajectory with the expected trajectory in the coverage evaluation strategy, and reconstruct the confidence of the search trajectory based on the comparison results.

[0155] In the above steps, the environmental perception segment refers to the raw data packet containing environmental features and its own pose collected by a single UAV through its own sensors within a specific time period. It is used to record the instantaneous state of the local environment and is continuously collected by airborne sensors.

[0156] Search trajectory confidence refers to the reliability assessment of the accuracy and effectiveness of the search paths executed by the drone swarm. It is used to measure the degree of consistency between the actual search actions and the expected plan, and is obtained through trajectory comparison analysis.

[0157] A shared sensing data pool refers to a shared data set formed by aggregating environmental sensing fragments uploaded by a cluster of drones through a wireless communication network, and established through a data sharing mechanism;

[0158] Integrated environmental perception information refers to the unified environmental cognition formed by fusing environmental perception segments from the shared perception data pool with environmental perception segments collected by the UAV itself, which is used to obtain a more comprehensive and accurate understanding of the environment. It is obtained through data integration and processing.

[0159] The actual search trajectory refers to the real flight path reconstructed by analyzing the continuous pose data of the UAV swarm in the heterogeneous search sub-region. It is used to reflect the actual search behavior of the UAV swarm and is obtained through trajectory reconstruction algorithms.

[0160] The expected trajectory refers to the idealized search path pre-planned in the coverage assessment strategy, which serves as a standard reference for evaluating the actual performance and is obtained through strategy analysis.

[0161] The comparison result refers to the quantitative deviation description obtained after analyzing the difference between the actual search trajectory and the expected trajectory. It is used to evaluate the execution effect and is calculated through trajectory comparison.

[0162] In this embodiment, when the drone swarm flies within the assigned heterogeneous search sub-area, it continuously scans the surrounding environment using its onboard sensor array and collects environmental perception fragments containing images, point clouds, and its own pose at fixed time intervals and distance intervals.

[0163] Secondly, the drone swarm broadcasts the collected environmental perception segments to the drone swarm communication network in real time through the wireless communication network. The environmental perception segments sent by the drone swarm are logically aggregated to form a shared perception data pool.

[0164] Next, each drone subscribes to and receives environmental perception segments from other drones from the shared perception data pool through a data integration algorithm. At the same time, it performs spatiotemporal registration and feature-level fusion of the environmental perception segments from other drones with the environmental perception segments it collects to generate comprehensive environmental perception information.

[0165] Then, the drone swarm uses a trajectory reconstruction algorithm to calculate the actual flight path of each drone in its respective sub-region based on its own and its companions' environmental perception segments under continuous timestamps in the comprehensive environmental perception information. These actual flight paths together constitute the actual search trajectory of the drone swarm.

[0166] Finally, the reconstructed actual search trajectory is compared with the predefined expected trajectory in the coverage evaluation strategy using a difference analysis algorithm. The degree of deviation between the two in terms of path shape, coverage, etc. is calculated. Based on this deviation analysis result, the confidence of the search trajectory is re-evaluated and adjusted.

[0167] Following the previous implementation case, the drone B1 first flew over the eastern section of the northern ridgeline it was responsible for. Its sensors continuously recorded the topographic point cloud data and optical images of the ridgeline. This data, along with its GPS coordinates, was packaged into environmental perception segments.

[0168] Secondly, B1 transmits environmental sensing segments via a wireless link and simultaneously receives environmental sensing segments from B2, B3, B4, and B5 from their respective areas. All environmental sensing segments are aggregated to form a shared sensing data pool.

[0169] Then, the B1's onboard computer immediately ran a data fusion program, integrating the segments about the south slope and forest area in the shared sensing data pool with the north slope ridgeline segments it had just collected, to generate comprehensive environmental sensing information covering the terrain of the north slope and part of the south slope.

[0170] Based on this comprehensive environmental perception information, the flight path of B1 on the ridgeline can be clearly depicted, and the detour route of B2 in the forest area can be roughly inferred. These paths together constitute the actual search trajectory of the drone swarm.

[0171] Finally, the actual search trajectory was compared with the ideal search route planned by the coverage assessment strategy. It was found that B1 strictly followed the ridgeline, while B2 deviated slightly from the planned path due to avoiding a dense forest. Based on the degree of deviation, the confidence index of the accuracy of the current search action was recalculated, reflecting that B1 had a high degree of confidence in its execution, while B2's path needed further evaluation due to the detour.

[0172] This step enables the UAV swarm to achieve global situational awareness and real-time self-evaluation of action effectiveness during mission execution. By sharing and fusing individual perception data, the UAV swarm gains environmental cognition beyond the perspective of a single UAV, thus accurately reconstructing the overall actual search trajectory. By intelligently comparing the actual trajectory with the expected plan, the quality and reliability of the current search action can be dynamically evaluated, providing accurate and quantitative decision-making basis for subsequent strategy adjustments, ensuring the effectiveness of the search mission and closed-loop optimization capabilities.

[0173] Step 105: Based on the confidence of the reconstructed search trajectory, asynchronously update the collaborative decision constraints in the coverage evaluation strategy to obtain the cluster collaborative decision rules.

[0174] Optionally, step 105 may specifically include the following steps:

[0175] Step 1051: Based on the confidence level of the reconstructed search trajectory, analyze the execution deviation of the coverage evaluation strategy in the heterogeneous search sub-region;

[0176] Step 1052: Calculate the amount of cooperative strategy adjustment required by the drone swarm based on the execution deviation;

[0177] Step 1053: Combine the adjustment amount of the cooperative strategy with the real-time location information and the environmental state information to generate correction parameters;

[0178] Step 1054: Asynchronously adjust the constraint terms corresponding to the collaborative decision constraint according to the correction parameters to form cluster collaborative decision rules.

[0179] In the above steps, collaborative decision constraints refer to the rules and restrictions defined in the coverage assessment strategy to regulate the collaborative behavior among drone swarms, and are used to ensure the consistency and security of swarm actions. These constraints are obtained through strategy initialization settings.

[0180] Cluster collaborative decision-making rules refer to the complete behavioral specifications that guide the collaborative search of drone clusters after dynamic updates, and are used to directly control the final actions of drone clusters.

[0181] Execution bias refers to the degree of difference between the expected effect of the coverage assessment strategy and the actual execution effect of the drone swarm in the heterogeneous search sub-region. It is used to quantify the execution error of the strategy and is obtained through comparative analysis.

[0182] The collaborative strategy adjustment amount refers to the quantitative value of the magnitude and direction of the modification to collaborative decision constraints required to correct execution deviations. It is used to guide strategy correction and is calculated through deviation analysis.

[0183] The correction parameter refers to a composite variable that integrates the adjustment amount of the cooperative strategy, real-time location information, and environmental state information. It is used to accurately calibrate the adjustment process of the constraint terms and is obtained through the fusion of multi-source information.

[0184] Constraints refer to specific rule entries in collaborative decision-making constraints that can be adjusted independently. They serve as the basic operational objects for strategy adjustment and are obtained through strategy parsing.

[0185] In this embodiment, a differential quantification algorithm is first used to quantitatively analyze the performance of the coverage evaluation strategy in the executed heterogeneous search sub-regions by taking the confidence of the reconstructed search trajectory as the main input. By comparing the expected search effect with the actual effect, the execution deviation of the strategy in each dimension is calculated.

[0186] Secondly, based on the calculated execution deviation, the magnitude and direction of the deviation are mapped to the degree of change required at the strategy level through the feedback adjustment model in control theory, and the amount of coordinated strategy adjustment required by the entire drone swarm to make up for the current deviation is calculated.

[0187] Next, by using a multi-factor fusion function, the adjustment amount of the cooperative strategy is combined with real-time location information and environmental state information. Taking into account the current distribution status of the UAV and the real-time constraints of the environment, more context-aware correction parameters are generated.

[0188] Then, using these corrected parameters, the various collaborative decision constraints in the coverage evaluation strategy are adjusted through an asynchronous update mechanism. The adjusted constraints are then reintegrated to form an updated set of cluster collaborative decision rules applicable to the current task phase.

[0189] Following the implementation case from the previous step, we first calculate the execution deviation of the current coverage assessment strategy in terms of the balance between obstacle avoidance and path following, based on the recalculated trust index.

[0190] Secondly, based on this execution deviation, the adjustment amount of the collaborative strategy is calculated as the amount that needs to be adjusted to appropriately relax the requirements for path tracking strictness and increase the weight of avoiding sudden obstacles.

[0191] Next, the adjustment amount of the collaborative strategy is combined with the real-time location information of B1 to B5 and the environmental status information of mountain area C to generate correction parameters;

[0192] Then, the correction parameters are applied asynchronously to adjust the constraint terms corresponding to the collaborative decision constraints, and all adjusted constraint terms are integrated into cluster collaborative decision rules.

[0193] This step enables online adaptation and optimization of the collaborative search strategy. Based on real feedback from task execution, it can intelligently diagnose mismatches in the strategy and comprehensively consider the real-time environment and cluster status to make precise and differentiated adjustments to the collaborative rules. This makes the collaborative strategy of the UAV cluster no longer static and rigid, but dynamically evolving, thereby continuously improving its robustness and efficiency in completing collaborative search tasks in complex and uncertain environments.

[0194] Step 106: Perform distributed trajectory adjustment according to the cluster collaborative decision-making rules to form a cluster motion pattern to complete multiple rounds of iterative search of the area to be searched.

[0195] Optionally, step 106 may specifically include the following steps:

[0196] Step 1061: Based on the cluster collaborative decision-making rules and the real-time location information, generate a preliminary flight path for the UAV cluster;

[0197] Step 1062: Based on the potential threat areas in the environmental state information, perform threat avoidance correction on the preliminary trajectory to generate an individual corrected trajectory;

[0198] Step 1063: Based on the real-time status information of the individual corrected trajectory and neighboring UAVs, evaluate the interaction between the individual corrected trajectory;

[0199] Step 1064: Adjust the individual correction trajectory according to the interaction influence and the cluster collaborative decision-making rules to generate the final trajectory;

[0200] Step 1065: Based on the final flight path, control the drone swarm to fly along the final flight path to form a swarm movement pattern;

[0201] Step 1066: Based on the cluster motion pattern, perform a new round of search on the area to be searched to complete multiple rounds of iterative search.

[0202] In the above steps, distributed trajectory adjustment refers to the process by which each UAV autonomously calculates and optimizes its own flight path based on shared cluster collaborative decision-making rules and local information. It is used to achieve individual path optimization under overall collaboration and is obtained through trajectory planning algorithms.

[0203] Cluster motion mode refers to the overall and emergent collaborative motion pattern exhibited by a cluster of drones after distributed trajectory adjustments, which is used to efficiently cover the search area and is formed through the execution of the final trajectory.

[0204] Multi-round iterative search refers to a task execution method that gradually covers the area to be searched by repeatedly executing a search evaluation and adjustment loop, in order to ensure the completeness of the search task;

[0205] The initial trajectory refers to the initial flight path that meets the basic mission requirements, generated based on cluster collaborative decision-making rules and real-time location information, and is obtained through a trajectory generation algorithm;

[0206] Threat avoidance correction is a path safety enhancement process that considers potential threat areas in environmental state information based on the initial flight path, in order to ensure flight safety. It is implemented through path optimization algorithms.

[0207] Individual corrected trajectory refers to the optimal path obtained by a single UAV after completing threat avoidance correction, considering only its own safety. It is used as input for cluster collaborative adjustment and obtained through local path optimization.

[0208] Interactive effects refer to the potential conflicts or collaborative effects in time and space that may arise from the individual modified trajectories of different UAVs. They are used to assess the mutual constraints between trajectories and are obtained through conflict detection algorithms.

[0209] The final flight path refers to the actual flight path of each drone determined after comprehensively considering individual safety and drone swarm collaboration. It is used to directly control the flight of the drone swarm and is generated through a distributed negotiation algorithm.

[0210] In this embodiment, a preliminary trajectory is first generated by running a local trajectory planning algorithm, guided by the cluster collaborative decision-making rules and combined with its own real-time location information.

[0211] Secondly, through the path optimization algorithm of threat perception, the potential threat areas in the environmental state information are used as constraints to perform safety checks and corrections on the initial trajectory, adjust the trajectory to move away from the threat area, and generate individual corrected trajectories.

[0212] Next, the individual corrected trajectories and real-time status information of neighboring UAVs are obtained through the wireless communication network. Through the conflict prediction and collaborative benefit evaluation model, the system analyzes whether there are spatiotemporal conflicts between its own individual corrected trajectories and the trajectories of neighboring UAVs and whether there are areas that can be collaboratively covered, and evaluates the interactive impact between trajectories.

[0213] Then, based on the assessed interaction effects, and in accordance with the rules of collision avoidance, formation maintenance, and information sharing in the cluster collaborative decision-making rules, the individual correction trajectories are fine-tuned through a distributed negotiation algorithm to eliminate conflicts, improve collaborative efficiency, and ultimately generate a final trajectory for each drone.

[0214] Afterwards, the flight control system of each UAV executes the flight mission according to the final trajectory. The execution of the UAV swarm trajectory shows an orderly, complementary, and efficient swarm movement mode on a macro level. It determines whether the current round of search has achieved the predetermined goal. If not, a new round of search cycle is triggered based on new environmental state information and real-time location information to complete multiple rounds of iterative search until the entire search area is completely covered.

[0215] Following the implementation case of the previous step, B1 first planned a preliminary track to scan westward along the ridgeline based on the cluster collaborative decision-making rules and its own location at the eastern starting point of the ridgeline on the north slope.

[0216] Secondly, B1 referenced the environmental map and found that the middle section of the flight path was close to an area marked with weak signal interference. Therefore, the flight path was modified to avoid threats, generating an individual modified flight path that was slightly off the highest point of the ridge to bypass the area.

[0217] Meanwhile, B1 receives individual corrected tracks and real-time location information from its neighbor B2 via communication. After running the analysis program, it finds that B1's own corrected track and B2's planned track to enter the north slope forest area have overlapping risks in airspace and time, and identifies negative interactive effects.

[0218] Next, according to the rules on collision avoidance and area handover coordination in the cluster collaborative decision-making rules, B1 and B2 conducted a simple negotiation through communication. B1 slightly delayed the start time of its track, while B2 slightly adjusted its entry angle. The two parties generated a final track that did not conflict and could be smoothly connected.

[0219] Then, B1 to B5 all began flying according to their respective final flight paths. B1 and B2 cooperated to cover the north slope, B3 and B4 cooperated to cover the south slope, and B5 maneuvered in the middle area. Their flights together constituted an efficient swarm movement pattern, completing the first round of search of the current key area of ​​mountainous area C. After evaluation, it was determined that it was not completely covered, so the drone swarm was instructed to start a new round of iterative search. Based on the new location and possibly updated environmental information, the above process was repeated until the entire search area was completely covered.

[0220] This step achieves a seamless transition from individual trajectory optimization to overall coordinated movement within the drone swarm. It ensures that the trajectory generated by each drone is not only safe and efficient, but also highly coordinated with the trajectories of other members in the swarm. Ultimately, this forms a macroscopic swarm movement pattern that can adapt to the environment, dynamically avoid threats, and efficiently cover the mission area. Through multiple iterations, this process can gradually cover the entire complex area, significantly improving the integrity, safety, and overall efficiency of the search mission.

[0221] Figure 2 This application provides a schematic diagram of the structure of a UAV swarm cooperative search system, such as... Figure 2 As shown, the system includes:

[0222] The acquisition module 21 is used to acquire environmental status information of the area to be searched and the real-time location information of the drone cluster;

[0223] The generation module 22 is used to generate a coverage assessment strategy based on the environmental state information and the real-time location information by interacting with historical decision data;

[0224] Allocation module 23 is used to dynamically allocate heterogeneous search sub-regions according to the coverage evaluation strategy;

[0225] The reconstruction module 24 is used to share environmental perception segments in real time in the heterogeneous search sub-region and reconstruct the search trajectory confidence using the environmental perception segments;

[0226] Update module 25 is used to asynchronously update the collaborative decision constraints in the coverage evaluation strategy based on the confidence of the reconstructed search trajectory, so as to obtain the cluster collaborative decision rules;

[0227] The adjustment module 26 is used to perform distributed trajectory adjustment according to the cluster collaborative decision-making rules to form a cluster motion mode to complete multiple rounds of iterative search of the area to be searched.

[0228] Figure 2 The aforementioned UAV swarm collaborative search system can perform Figure 1 The implementation principle and technical effects of the UAV swarm collaborative search method described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the UAV swarm collaborative search system in the above embodiments perform operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0229] In one possible design, Figure 2 The drone swarm collaborative search system of the embodiment shown can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0230] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0231] The processing component 32 is used for the above Figure 1 The embodiment describes a method for collaborative search by a swarm of unmanned aerial vehicles (UAVs).

[0232] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0233] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0234] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0235] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0236] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0237] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0238] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown illustrates a method for collaborative search by a drone swarm.

[0239] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0240] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0241] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0242] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for collaborative search by a swarm of unmanned aerial vehicles (UAVs), characterized in that, include: Obtain environmental status information of the area to be searched and real-time location information of the drone swarm; Based on the environmental state information and the real-time location information, a coverage assessment strategy is generated through interactive historical decision data, including: extracting historical decision fragments similar to the environmental state information from the interactive historical decision data; analyzing the executed area coverage of the historical decision fragments in the search area based on the real-time location information; calculating the strategy preference vector of the historical decision fragments in the search area based on the executed area coverage; fusing the strategy preference vector with the environmental state information to generate a preliminary assessment strategy; and adjusting the spatial distribution of the preliminary assessment strategy in conjunction with the real-time location information to generate a coverage assessment strategy. According to the coverage evaluation strategy, heterogeneous search sub-regions are dynamically allocated, where heterogeneous search sub-regions refer to different search areas divided according to differences in search value and task difficulty. In the heterogeneous search sub-region, the UAV cluster is controlled to share environmental perception segments in real time, and the search trajectory confidence is reconstructed using the environmental perception segments; Based on the confidence level of the reconstructed search trajectory, the collaborative decision constraints in the coverage evaluation strategy are updated to obtain cluster collaborative decision rules, including: analyzing the execution deviation of the coverage evaluation strategy in the heterogeneous search sub-region based on the confidence level of the reconstructed search trajectory; calculating the amount of collaborative strategy adjustment required by the UAV cluster based on the execution deviation; combining the collaborative strategy adjustment amount with the real-time location information and the environmental state information to generate correction parameters; and asynchronously adjusting the constraint terms corresponding to the collaborative decision constraints based on the correction parameters to form cluster collaborative decision rules. Distributed trajectory adjustments are performed according to the cluster collaborative decision-making rules to form a cluster motion pattern to complete multiple rounds of iterative search of the area to be searched.

2. The method according to claim 1, characterized in that, Obtain environmental status information of the area to be searched and real-time location information of the drone swarm, including: Raw sensing data of the area to be searched is collected through an airborne sensing unit. The raw sensing data is combined with the navigation data of the UAV swarm to obtain a local environment description; The local environment description is exchanged through a wireless communication network to construct a global environment map; Based on the global environment map, passable areas and potential threat areas in the area to be searched are identified to obtain environmental status information; The real-time location information of the drone swarm is obtained by aggregating the real-time locations of the drone swarm through a wireless communication network.

3. The method according to claim 1, characterized in that, Based on the coverage evaluation strategy, heterogeneous search sub-regions are dynamically allocated, including: Based on the coverage evaluation strategy, the area to be searched is divided into multiple sub-regions; Based on the potential threat areas in the environmental status information, the sub-region is labeled with a dynamic threat level; Based on the real-time location information, calculate the movement cost of the drone swarm to different sub-regions; By fusing the coverage characteristics of different sub-regions, the dynamic threat level, and the mobility cost, the search adaptability of the drone swarm to different sub-regions can be obtained. Based on the search fitness, the multiple sub-regions are dynamically allocated to the drone cluster.

4. The method according to claim 1, characterized in that, In the heterogeneous search sub-region, environmental awareness segments are shared in real time, and the search trajectory confidence is reconstructed using these environmental awareness segments, including: Within the heterogeneous search sub-region, the drone swarm continuously acquires environmental perception fragments; The environmental perception segments are shared within the UAV cluster via a wireless communication network to form a shared perception data pool. The environmental perception segments in the shared perception data pool are integrated with the environmental perception segments collected by the UAV cluster itself to generate comprehensive environmental perception information. Based on the comprehensive environmental perception information, the actual search trajectory of the UAV cluster in the heterogeneous search sub-region is analyzed. The actual search trajectory is compared with the expected trajectory in the coverage evaluation strategy, and the confidence level of the search trajectory is reconstructed based on the comparison results.

5. The method according to claim 1, characterized in that, Distributed trajectory adjustments are executed according to the aforementioned cluster collaborative decision-making rules to form a cluster motion pattern for completing multiple rounds of iterative searching of the area to be searched, including: Based on the cluster collaborative decision-making rules and the real-time location information, a preliminary flight path is generated for the UAV cluster; Based on the potential threat areas in the environmental status information, the initial flight path is modified to avoid threats, generating an individual modified flight path; Based on the real-time status information of the individual corrected trajectory and neighboring UAVs, the interaction between the individual corrected trajectory is evaluated; Based on the interaction effects and the cluster collaborative decision-making rules, the individual correction trajectory is adjusted to generate the final trajectory; Based on the final flight path, the drone swarm is controlled to fly along the final flight path to form a swarm movement pattern; Based on the cluster movement pattern, a new round of search is performed on the area to be searched to complete multiple rounds of iterative search.

6. A drone swarm cooperative search system, applied to the drone swarm cooperative search method according to any one of claims 1-5, characterized in that, include: The acquisition module is used to acquire environmental status information of the area to be searched and the real-time location information of the drone cluster; The generation module is used to generate a coverage assessment strategy based on the environmental state information and the real-time location information by interacting with historical decision data; The allocation module is used to dynamically allocate heterogeneous search sub-regions according to the coverage evaluation strategy; The reconstruction module is used to share environmental perception segments in real time in the heterogeneous search sub-regions and reconstruct the search trajectory confidence using the environmental perception segments; The update module is used to asynchronously update the collaborative decision constraints in the coverage evaluation strategy based on the confidence of the reconstructed search trajectory, so as to obtain the cluster collaborative decision rules. The adjustment module is used to perform distributed trajectory adjustment according to the cluster collaborative decision-making rules to form a cluster motion pattern to complete multiple rounds of iterative search of the area to be searched.

7. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement the UAV swarm cooperative search method as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements a drone swarm cooperative search method as described in any one of claims 1 to 5.

Images