Unmanned aerial vehicle cluster cooperative dynamic replacement endurance control method and system

Abstract

The application discloses a kind of unmanned aerial vehicle cluster cooperative dynamic replacement endurance control method and system, it is related to unmanned aerial vehicle cluster cooperative dynamic replacement endurance control field, for improving the task execution efficiency and security of unmanned aerial vehicle cluster under complex, dynamic environment, including: when meeting the first condition, first unmanned aerial vehicle sends emergency task request to accessible adjacent unmanned aerial vehicle;Adjacent unmanned aerial vehicle determines the value measurement of emergency task according to the information of sudden target;Determine the cost measurement of adjacent unmanned aerial vehicle abandoning current task;When the value measurement of emergency task is higher than cost measurement, adjacent unmanned aerial vehicle autonomously decides to replace emergency task;And through intercommunication to cluster informs replacement intention;First unmanned aerial vehicle receives the replacement intention of adjacent unmanned aerial vehicle, and sends replacement confirmation information to adjacent unmanned aerial vehicle;In response to replacement confirmation information, adjacent unmanned aerial vehicle switches own task state to execute emergency task.

Description

Technical Field

[0001] This invention relates to the field of collaborative dynamic replacement endurance control for unmanned aerial vehicle (UAV) swarms, and particularly to a method and system for collaborative dynamic replacement endurance control for UAV swarms. Background Technology

[0002] When drone swarms perform long-duration missions, the endurance of individual drones is often limited, and traditional fixed formations are ill-suited to the dynamic adjustment requirements of missions. Especially in complex and communication-unstable environments, such as disaster sites, when a drone's battery is about to run out and it unexpectedly discovers a critical and urgent target, existing replacement endurance solutions are often ineffective. This could lead to drones crashing while performing emergency missions, losing critical information, or missing valuable rescue opportunities while prioritizing their own safe return. To address these issues, this application proposes a collaborative dynamic replacement endurance control method for drone swarms. This method aims to effectively address the challenges of communication interruptions, insufficient energy, and the overlapping of sudden high-priority tasks in complex environments through intelligent decision-making and task handover mechanisms, ensuring mission continuity and the integrity of critical information. Summary of the Invention

[0003] This application discloses a collaborative dynamic replacement endurance control method and system for unmanned aerial vehicle (UAV) swarms, which aims to solve the problem that UAV swarms cannot effectively perform dynamic task adjustment and collaborative decision-making when communication is interrupted, replacement endurance fails, or sudden high-priority tasks overlap in complex environments.

[0004] In a first aspect, this application discloses a method for collaborative dynamic replacement of endurance control in a drone swarm, comprising the following steps:

[0005] When the first UAV in the cluster meets the first condition, the first UAV sends an emergency mission request to the reachable neighboring UAVs through inter-UAV communication; the first condition is that the first UAV's communication with the external command is interrupted, its own energy reserves are lower than the preset level, and it identifies a sudden target that requires immediate response; the emergency mission request includes information about the sudden target and the status information of the first UAV.

[0006] A nearby drone receives an emergency mission request and, based on information about the sudden target, determines the value of the emergency mission; and determines the cost of the nearby drone abandoning its current mission.

[0007] Nearby drones compare the value and cost of an emergency mission; when the value of the emergency mission is higher than the cost, the nearby drones autonomously decide to take over the emergency mission and notify the cluster of their intention to take over via inter-drone communication.

[0008] The first UAV receives the takeover intention from a neighboring UAV and sends a takeover confirmation message to the neighboring UAV; in response to the takeover confirmation message, the neighboring UAV switches its own mission status to perform an emergency mission; and before switching mission status, it transmits the key data of the neighboring UAV's original mission to other UAVs in the cluster.

[0009] After sending a takeover confirmation message, the first UAV initiates a safe return-to-home process. The safe return-to-home process includes planning a path that allows the first UAV to land safely before its energy reserves are depleted, and transmitting the key data collected by the first UAV during the return process.

[0010] Optionally, the value measure of an emergency mission satisfies the following relationship:

[0011] ;

[0012] Wherein, Score_Emergency is the value measure of the urgent task, Urgency is the urgency level of the urgent task, Timeliness is the timeliness of the response, and Importance is the importance level of the overall task objective;

[0013] W1, W2, and W3 are preset weighting coefficients.

[0014] Optionally, the cost metric for a nearby drone abandoning its current mission satisfies the following relationship:

[0015] ;

[0016] Here, Cost_Current is the cost metric, Risk_Level is the risk level of the current task area, Blank_Duration is the duration of the monitoring gap caused by neighboring drones abandoning their current tasks, and Inherent_Importance is the inherent importance of the neighboring drones' own current tasks.

[0017] W4, W5, and W6 are preset weighting coefficients.

[0018] Optionally, when the value metric of an emergency task equals its cost metric, the method further includes:

[0019] The neighboring drone extracts the type of emergency mission from the emergency mission request and obtains the type of its own current mission from its own mission management module;

[0020] Based on a preset mapping relationship, the neighboring drones match the type of emergency mission with the type of their own current mission to determine the priority of the emergency mission and the priority of their own current mission; the preset mapping relationship includes the mapping relationship between different mission types and different priorities.

[0021] The priority of the urgent task of the neighboring drone is compared with the priority of the current task of the neighboring drone, and the task with higher priority is selected as the decision result.

[0022] The neighboring drones issue action instructions corresponding to the decision results and notify the cluster of the intention to take over through inter-drone communication.

[0023] Optionally, there may be multiple neighboring drones. When multiple neighboring drones determine that the value of the emergency mission is higher than the cost, the method may further include:

[0024] For each of the multiple neighboring drones, each neighboring drone encapsulates an intent broadcast data packet; the intent broadcast data packet contains the comparison result, its own identifier, and current location information; the comparison result is used to compare the value measure and cost measure of the emergency mission.

[0025] For each of the multiple neighboring drones, the neighboring drone continuously listens for intent broadcast packets from other neighboring drones;

[0026] When a neighboring drone receives an intent broadcast data packet from another neighboring drone, if the other neighboring drone is closer to the first drone or has an earlier broadcast timestamp, the neighboring drone cancels its own takeover intent.

[0027] Optionally, after determining the value measure of the emergency task, the method may also include:

[0028] Nearby drones continuously monitor key factors in the environment of a sudden target; key factors include the rate of fire spread, the target's movement trajectory, and changes in weather conditions; the target's movement trajectory is the movement trajectory of the sudden target.

[0029] The nearby drone dynamically adjusts the value of emergency missions based on changes in key factors; in particular, it increases the value of emergency missions when the fire spreads faster, the target's movement trajectory becomes irregular, or weather conditions worsen.

[0030] Optional key factors for nearby drones to continuously monitor the environment of sudden targets include:

[0031] Nearby drones collect visual images, temperature distribution data, wind speed, wind direction, and air pressure and temperature data of the environment in which the sudden target is located;

[0032] The nearby drone processes the acquired visual images, extracts the flame areas in the images, and calculates the fire spread rate based on the shape changes and pixel growth rate of the flame areas.

[0033] Nearby drones analyze the collected temperature distribution data, identify the boundaries of high-temperature areas, and correct the fire spread rate based on the expansion rate of the high-temperature area boundaries.

[0034] The nearby UAV performs target recognition and tracking on the acquired visual images, obtains the position sequence of the sudden target in the image coordinate system, and combines its own flight attitude and position information to convert the position sequence into the target movement trajectory in the spatial coordinate system.

[0035] Nearby drones use wind speed, wind direction, air pressure, and temperature data to obtain information on changes in meteorological conditions in the airspace where sudden targets are located.

[0036] Optionally, the nearby UAV performs target recognition and tracking on the acquired visual images, obtains the position sequence of the sudden target in the image coordinate system, and combines it with its own flight attitude and position information to convert the position sequence into the target's movement trajectory in the spatial coordinate system, including:

[0037] The nearby drone applies target detection functionality to visual images to identify sudden targets in the images and obtain the initial position of the sudden targets;

[0038] Based on the initial location of the sudden target, the nearby UAV constructs a target appearance description; the target appearance description includes the color distribution information and texture feature information of the sudden target;

[0039] The neighboring drone continuously tracks the sudden target. When the sudden target is obstructed, deformed, or the lighting changes, the neighboring drone predicts the position of the sudden target in the current frame based on the target's appearance description and the data of the neighboring drone continuously tracking the sudden target.

[0040] Nearby drones conduct local searches around the predicted location to update the position sequence of sudden targets in the image coordinate system;

[0041] By combining its own flight attitude and position information, the nearby UAV can convert the updated position sequence of the sudden target in the image coordinate system into the target's movement trajectory in the spatial coordinate system.

[0042] Optionally, nearby drones perform a local search around the predicted location to update the position sequence of the sudden target in the image coordinate system, including:

[0043] The nearby drone extracts features from the image region within a preset range of the predicted location to obtain the local feature set of that region;

[0044] The neighboring UAV performs a multi-dimensional similarity comparison between the local feature set and the target appearance description; the multi-dimensional similarity comparison includes color distribution similarity comparison and texture feature similarity comparison;

[0045] Based on multi-dimensional similarity comparison results, the neighboring drones calculate a comprehensive matching score;

[0046] When there are multiple regions with matching scores higher than the matching threshold, the neighboring drones combine data from continuous tracking of the sudden target and select the region with the most matching motion trajectory as the new location of the target.

[0047] Based on the new target location, nearby UAVs update the position sequence of the sudden target in the image coordinate system.

[0048] Secondly, this application also discloses a collaborative dynamic replacement endurance control system for unmanned aerial vehicle (UAV) swarms, the system comprising:

[0049] The emergency mission request module is used to send an emergency mission request to nearby drones via inter-drone communication when the first drone in the cluster meets a first condition. The first condition is that the first drone's communication with the external command is interrupted, its own energy reserves are lower than a preset level, and it has identified a sudden target that requires immediate response. The emergency mission request includes information about the sudden target and the status information of the first drone.

[0050] The value assessment module is used to receive emergency mission requests from nearby drones and determine the value of the emergency mission based on the information of the sudden target; and to determine the cost of nearby drones abandoning their current mission.

[0051] The task decision module is used to compare the value and cost of an emergency task with that of a neighboring drone. When the value of the emergency task is higher than the cost, the neighboring drone autonomously decides to take over the emergency task and notifies the cluster of its intention to take over through inter-drone communication.

[0052] The task handover module is used for the first UAV to receive the takeover intention of the neighboring UAV and send the takeover confirmation information to the neighboring UAV; in response to the takeover confirmation information, the neighboring UAV switches its own task status to perform the emergency task; and before switching the task status, it transmits the key data of the original task of the neighboring UAV to other UAVs in the cluster.

[0053] The safe return module is used to initiate the safe return process after the first UAV sends the takeover confirmation information. The safe return process includes planning a path that allows the first UAV to land safely before its energy reserves are depleted, and transmitting the key data collected by the first UAV during the return process.

[0054] Beneficial effects

[0055] The UAV swarm collaborative dynamic replacement endurance control method disclosed in this application solves the problem in existing technologies where UAV swarms cannot effectively perform dynamic task adjustment and collaborative decision-making when communication is interrupted, replacement endurance fails, and sudden high-priority tasks occur. Specifically, when the first UAV in the swarm faces multiple dilemmas such as communication interruption, insufficient energy reserves, and detection of a sudden target, it can proactively send an emergency task request to neighboring UAVs. Upon receiving the request, the neighboring UAVs autonomously assess the value of the emergency task and the cost of abandoning their current task, and decide whether to take over the task based on the assessment results. This mechanism enables the UAV swarm to maintain a high degree of autonomy and adaptability even in the absence of ground command, ensuring timely response to emergencies and continuous acquisition of critical information. This method not only avoids the risk of UAVs crashing or losing critical information due to forced task execution, but also avoids the regret of missing valuable rescue opportunities due to returning to base, thereby effectively improving the task execution efficiency and safety of UAV swarms in complex and dynamic environments, and overcoming the shortcomings of existing replacement endurance schemes in extreme scenarios. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of a method for collaborative dynamic replacement of endurance control in a drone swarm provided by an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of another method for collaborative dynamic replacement of endurance control in unmanned aerial vehicle (UAV) clusters provided in this embodiment of the invention.

[0058] Figure 3 This is a schematic diagram of a drone swarm collaborative dynamic replacement endurance control system provided in an embodiment of the present invention. Detailed Implementation

[0059] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0060] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] To better understand the UAV swarm collaborative dynamic replacement endurance control method proposed in this application, the following will elaborate on some key terms and implementation environments involved.

[0062] "First drone" refers to a drone in a swarm that needs to be replaced due to its own condition (such as low energy reserves or communication interruption) or the discovery of a sudden target.

[0063] "Nearby drones" refers to other drones in the cluster that can receive the emergency mission request of the first drone and have the ability to take over the mission when the first drone needs to be replaced.

[0064] The "first condition" is a comprehensive set of conditions that trigger the first UAV to send an emergency mission request. It includes communication interruption, energy reserves falling below a preset level, and the identification of a sudden target requiring immediate response. These conditions collectively indicate that the first UAV is in a state requiring emergency support.

[0065] An "emergency mission request" is a data packet sent by the first UAV to a neighboring UAV. It contains detailed information about the sudden target (such as location, type, and urgency) as well as the current status information of the first UAV itself (such as remaining battery power and mission progress).

[0066] The "Value Measurement of Emergency Missions" is a quantitative metric used by nearby drones to assess the importance and priority of emergency missions, and to measure the overall benefits that can be brought about by taking over the mission.

[0067] "Cost Measure" is a quantitative indicator used by neighboring drones to assess the negative impact of abandoning their current mission, such as mission interruption losses and resource waste.

[0068] The "safe return process" refers to the planned and executed process of the first UAV returning to its base or safe landing point after the mission handover is completed. It aims to ensure that the UAV is safely recovered before its energy is depleted and to transmit the critical data that has been collected.

[0069] The following specific embodiments will provide a detailed description and explanation of the UAV swarm collaborative dynamic replacement endurance control method provided in this application.

[0070] Reference Figure 1 This invention provides a method for collaborative dynamic replacement of endurance control in a drone swarm, comprising the following steps:

[0071] S1, when the first drone in the cluster meets the first condition, the first drone sends an emergency mission request to the reachable neighboring drones through inter-drone communication.

[0072] The first condition is that the first UAV loses communication with external command, its own energy reserves are lower than a preset level, and it identifies a sudden target that requires immediate response; the emergency mission request includes information about the sudden target and the status information of the first UAV.

[0073] Specifically, when the first drone in the cluster meets the first condition—that is, when the first drone's communication with external command is interrupted, its own energy reserves are lower than a preset level, and it identifies a sudden target requiring immediate response—the first drone will send an emergency mission request to nearby drones via inter-drone communication. For example, the first drone can periodically check the status of its communication link with the ground control center; if it does not receive a response to multiple consecutive heartbeat packets, it can determine that the communication is interrupted. Simultaneously, the first drone can monitor its battery level in real time; when the battery level is lower than a preset threshold (e.g., only 1.2 times the power required for return), it is considered that its energy reserves are lower than a preset level.

[0074] Furthermore, the first UAV can scan the mission area using its onboard sensors (such as visual sensors and thermal imaging sensors) and identify sudden targets requiring immediate response through its built-in target recognition algorithm, such as trapped personnel at a fire scene or a sudden explosion point. When these three conditions are met simultaneously, the first UAV will trigger the transmission of an emergency mission request. The emergency mission request will contain detailed information about the sudden target, such as its geographical coordinates, the type of target identified (such as personnel, vehicles, fire sources, etc.), and the first UAV's remaining battery power and current mission progress. This request can be broadcast via short-range radio links, satellite communication links, or the self-organizing network within the cluster.

[0075] S2. The neighboring drone receives an emergency mission request and determines the value of the emergency mission based on the information of the sudden target; and determines the cost of the neighboring drone abandoning its current mission.

[0076] For example, nearby drones can calculate the value of an emergency mission based on factors such as the urgency, importance, and timeliness of the response to the sudden target, using a pre-set evaluation model. The urgency can be determined based on the target type (e.g., life rescue is more important than property protection) and time sensitivity (e.g., the speed of fire spread); importance can be assessed based on the mission's impact on overall strategic objectives; and timeliness of the response can be measured based on the time window required from target detection to response.

[0077] Simultaneously, nearby drones also assess the cost of abandoning their current mission. For example, if a nearby drone is performing a monitoring mission on critical infrastructure, abandoning that mission could result in a monitoring gap, leading to significant costs. Cost metric considerations include the risk level of the current mission, potential losses from mission interruption, and the inherent importance of the mission itself. These assessments can be performed autonomously within the nearby drone's mission management module.

[0078] The value of an emergency mission is determined by the following relationship:

[0079] ;

[0080] Wherein, Score_Emergency is the value measure of the urgent task, Urgency is the urgency level of the urgent task, Timeliness is the timeliness of the response, and Importance is the importance level of the overall task objective; W1, W2, and W3 are preset weight coefficients.

[0081] Score_Emergency can be understood as a quantitative assessment of the overall importance of an emergency task; the higher the value, the more worthy the task is of being taken over. Urgency refers to the urgency level of the emergency task; for example, sudden events such as fires and earthquakes typically have a high urgency level, requiring immediate response. Timeliness refers to the response timeliness, reflecting the task's time sensitivity requirements; for example, in search and rescue missions, shorter response times lead to higher success rates, hence a higher timeliness value. Importance refers to the overall importance of the mission objective; for example, tasks protecting critical infrastructure or saving lives typically have a high importance value. W1, W2, and W3 are preset weighting coefficients used to adjust the relative contributions of Urgency, Timeliness, and Importance in the Score_Emergency calculation. These weighting coefficients can be flexibly configured according to specific application scenarios, task types, and cluster strategy preferences to ensure that the value measurement accurately reflects the actual priority of the task.

[0082] This application's solution introduces a Score_Emergency calculation formula, enabling neighboring UAVs to quantitatively assess the intrinsic value of emergency missions upon receiving them. By comprehensively considering urgency, response timeliness, and the overall importance of the mission objective, combined with preset weighting coefficients, the priority of emergency missions can be objectively and comprehensively measured. This quantitative assessment mechanism provides a scientific basis for neighboring UAVs to autonomously decide whether to take over emergency missions, avoiding misjudgments or inefficient decisions that may result from relying on a single factor. This ensures that, in the face of emergencies, the swarm can prioritize missions that have the greatest impact on the overall mission objective, are the most urgent, and require the most timely response.

[0083] The cost of a nearby drone abandoning its current mission is measured by the following relationship:

[0084] ;

[0085] Here, Cost_Current is the cost metric, Risk_Level is the risk level of the current task area, Blank_Duration is the duration of the monitoring gap caused by neighboring drones abandoning their current tasks, and Inherent_Importance is the inherent importance of the neighboring drones' own current tasks.

[0086] W4, W5, and W6 are preset weighting coefficients.

[0087] Specifically, Cost_Current represents the overall cost or negative impact of a neighboring drone abandoning its current mission. Risk_Level refers to the inherent danger level of the area where the neighboring drone is currently performing its mission. For example, when performing a reconnaissance mission, a higher Risk_Level value will be generated if there is high-risk hostile activity or complex terrain in the current area. Blank_Duration refers to the period during which the original mission area will lose surveillance or coverage when a neighboring drone abandons its current mission to perform an emergency mission. The longer this duration, the greater the potential loss. Inherent_Importance refers to the importance of the neighboring drone's current mission itself. For example, if the current mission is to protect critical infrastructure, its Inherent_Importance value will be higher. W4, W5, and W6 are preset weighting coefficients used to adjust the relative importance of Risk_Level, Blank_Duration, and Inherent_Importance in calculating Cost_Current. These weighting coefficients can be configured and optimized according to specific application scenarios and mission priorities.

[0088] This application's solution, by introducing the aforementioned cost-based calculation relationship, enables neighboring UAVs to quantify the potential losses of abandoning their current mission when deciding whether to take over an emergency task. By comprehensively considering the risk level of the current mission area, the duration of monitoring gaps that may result from abandoning the mission, and the inherent importance of the current mission itself, combined with preset weighting coefficients, the cost of abandoning the current mission can be calculated more comprehensively and objectively. This quantitative assessment mechanism provides solid data support for UAV autonomous decision-making, avoiding the bias that may result from judging based on a single factor.

[0089] S3. Neighboring drones compare the value and cost of the emergency mission; when the value of the emergency mission is higher than the cost, the neighboring drones autonomously decide to take over the emergency mission and notify the cluster of their intention to take over through inter-drone communication.

[0090] For example, if the assessment shows that the value of taking over an emergency mission to rescue trapped personnel far outweighs the cost of abandoning the current routine patrol mission, then a neighboring drone will make a takeover decision. After making the decision, the neighboring drone will broadcast its takeover intention to other drones in the swarm via inter-drone communication, informing the swarm that it will take over the emergency mission from the first drone. This broadcast may include information such as the neighboring drone's identifier, the type of emergency mission it is taking over, and its estimated arrival time, so that other drones in the swarm can update their mission assignment status.

[0091] S4. The first UAV receives the takeover intention from the neighboring UAV and sends a takeover confirmation message to the neighboring UAV; in response to the takeover confirmation message, the neighboring UAV switches its own task status to perform an emergency task; and before switching task status, it transmits the key data of the neighboring UAV's original task to other UAVs in the cluster.

[0092] Specifically, after receiving a takeover intention from a neighboring drone, the first drone will send a takeover confirmation message to that drone. For example, upon receiving the takeover intention, the first drone will perform a simple verification to confirm that the neighboring drone is indeed reachable and capable of taking over, and then send a confirmation signal. In response to the takeover confirmation message, the neighboring drone will switch its mission status to perform the emergency mission. Before switching mission status, the neighboring drone will also transmit key data of its original mission to other drones in the cluster. For example, if the neighboring drone was originally responsible for a reconnaissance mission in area A, it will send the latest reconnaissance data for area A, information on discovered targets, and key data such as incomplete reconnaissance plans to other suitable drones in the cluster via inter-drone communication, so that other drones can seamlessly take over or assist in completing the original mission, avoiding data loss and mission interruption.

[0093] S5. After sending the handover confirmation information, the first UAV initiates the safe return process.

[0094] The safe return process includes planning a path that allows the first drone to land safely before its energy reserves are depleted, and transmitting the critical data collected by the first drone during the return process.

[0095] For example, the first UAV can calculate an optimal return path based on its current location, remaining battery power, and the location of known safe landing points using its built-in path planning algorithm, ensuring a safe landing before its battery runs out. During the return journey, the first UAV will also transmit all critical data collected during the mission, such as detailed images, videos, and sensor readings of potential targets, to other UAVs in the cluster or the ground control center via inter-UAV communication or a restored external communication link, ensuring the integrity and traceability of the information.

[0096] The UAV swarm collaborative dynamic replacement endurance control method proposed in this application effectively addresses the challenges of communication interruptions, insufficient energy, and the overlapping of sudden high-priority tasks in complex environments through a series of closely coordinated steps. When the first UAV faces difficulties, it no longer passively waits for ground command but can proactively request assistance from the swarm. Neighboring UAVs can then make optimal decisions based on autonomous assessment, weighing the value of the emergency task against the cost of their own mission. This distributed, adaptive decision-making mechanism significantly improves the response speed and mission execution efficiency of the UAV swarm in emergencies.

[0097] Compared to traditional drone endurance replacement solutions, the innovation of this application lies in introducing autonomous decision-making and dynamic task handover capabilities into the drone swarm. In existing technologies, when drones face communication interruptions or sudden emergency tasks, they often cannot effectively hand over tasks, leading to task interruptions or information loss. For example, in a forest fire scenario, when a drone discovers trapped people due to low battery and communication interruption, traditional solutions may not be able to dispatch other drones in time to take over, thus missing rescue opportunities. This application, however, ensures mission continuity and the preservation of critical information under extreme conditions by having the first drone proactively send an emergency task request, and nearby drones autonomously assess the task's value and cost, and then hand over the task. This mechanism enables the drone swarm to shift from passively executing commands to actively adapting to the environment, significantly improving its robustness and intelligence in complex and dynamic task environments.

[0098] like Figure 2 As shown, specifically, when the value measure of an emergency task equals its cost measure, the method also includes:

[0099] S101. The neighboring UAV extracts the type of emergency mission from the emergency mission request and obtains the type of its own current mission from its own mission management module.

[0100] The types of emergency missions and the types of current missions of nearby drones can be understood as classifications of mission nature, such as "fire reconnaissance," "search and rescue operations," "area patrol," and "environmental monitoring." These mission types are typically defined during drone system initialization and stored in the mission management module. Nearby drones extract the type of emergency mission from emergency mission requests, for example, by parsing specific fields or identifiers in the request data packet. Simultaneously, nearby drones retrieve the type of mission they are currently performing from their own mission management module.

[0101] S102. Based on a preset mapping relationship, the neighboring drone matches the type of the emergency task with the type of its own current task to determine the priority of the emergency task and the priority of its own current task.

[0102] The preset mapping relationships include mapping relationships between different task types and different priorities.

[0103] For example, "search and rescue operations" can be defined as the highest priority, "fire reconnaissance" as a high priority, "area patrol" as a medium priority, and "environmental monitoring" as a low priority. This mapping relationship can be stored in the UAV's mission management module in the form of a lookup table, database, or algorithm. Through this mapping relationship, neighboring UAVs can convert the extracted emergency mission type and their own current mission type into specific priority values ​​or levels.

[0104] S103. The priority of the urgent task of the neighboring drone is compared with the priority of the current task of the neighboring drone, and the task with higher priority is selected as the decision result.

[0105] If the priority of the emergency mission is higher than the priority of its current mission, the neighboring drone will decide to take over the emergency mission; conversely, if the priority of its current mission is higher, the neighboring drone will maintain its current mission.

[0106] S104. The neighboring drones issue action instructions corresponding to the decision results and notify the cluster of the intention to take over through inter-drone communication.

[0107] Specifically, after comparing and determining the decision, the neighboring drones will issue corresponding action instructions, such as initiating a task switching procedure or continuing to perform the current task, and notify the cluster of their intention to take over or not to take over through inter-drone communication.

[0108] This application's solution addresses the decision-making dilemma that traditional methods may face when the value and cost of an urgent task are equal by introducing a task type and priority comparison mechanism. This mechanism enables UAVs to make deeper judgments based on pre-defined strategic importance or task nature when numerical assessments cannot distinguish between superior and inferior tasks, thereby avoiding task delays or improper resource allocation due to ambiguous decision-making.

[0109] As a specific implementation, suppose a first UAV, while performing a forest fire reconnaissance mission, experiences a communication interruption and identifies a rapidly spreading fire, sending an emergency mission request to a nearby second UAV. The second UAV is currently performing a routine area patrol mission. After value assessment, the second UAV finds that the value of taking over the fire reconnaissance mission is exactly equal to the cost of abandoning its own area patrol mission. At this point, the second UAV will initiate a priority comparison mechanism. Specifically, the second UAV extracts the emergency mission type as "fire reconnaissance" from the emergency mission request and obtains the current mission type as "area patrol" from its own mission management module. Based on a preset mapping relationship, for example, "fire reconnaissance" is mapped to high priority, and "area patrol" is mapped to medium priority. The second UAV compares and finds that "fire reconnaissance" has a higher priority than "area patrol," therefore it autonomously decides to take over the fire reconnaissance mission and notifies the cluster of its intention to take over.

[0110] In some embodiments described above, when the first UAV in the cluster meets certain conditions, neighboring UAVs decide whether to take over the emergency task by evaluating its value and the cost of abandoning their current task. However, in practical applications, when multiple reachable neighboring UAVs all determine that the value of the emergency task is higher than their own cost, multiple UAVs may attempt to take over the same emergency task simultaneously. This can lead to redundancy in task allocation, waste of resources, and potential conflicts within the cluster, thereby reducing the overall collaborative efficiency of the cluster.

[0111] To address this issue, this application proposes an optimization scheme to resolve the coordination problem when multiple neighboring UAVs simultaneously respond to emergency tasks, ensuring effective task allocation and rational utilization of cluster resources. This scheme introduces intent broadcasting and a competition mechanism, enabling multiple neighboring UAVs willing to take over the task to autonomously coordinate, ultimately allowing the most suitable UAV to take over the task.

[0112] When there are multiple neighboring drones, and when all the neighboring drones determine that the value of the emergency mission is higher than the cost, the method also includes:

[0113] S201. For each of the multiple neighboring drones, each neighboring drone encapsulates an intent-to-broadcast data packet.

[0114] The intent broadcast data packet contains the comparison result, its own identifier, and current location information; the comparison result is used to compare the value measure and cost measure of the emergency task.

[0115] Specifically, when the first drone in the cluster issues an emergency mission request, if multiple neighboring drones, after their respective evaluations, all consider taking over the emergency mission to be the better option—that is, the value of the emergency mission outweighs its cost—then these neighboring drones will not immediately perform the takeover operation. Instead, each neighboring drone intending to take over will encapsulate an intent broadcast data packet. This intent broadcast data packet contains the neighboring drone's comparison of the emergency mission's value with its own cost, indicating its intention to take over; simultaneously, for subsequent coordination and decision-making, the data packet also includes the neighboring drone's unique identifier and its current position information relative to the first drone.

[0116] S202. For each of the multiple neighboring drones, the neighboring drone continuously listens for intent broadcast packets from other neighboring drones.

[0117] In this mechanism, while encapsulating and broadcasting intent data packets, each neighboring drone intending to take over continuously listens for similar intent broadcast data packets from other neighboring drones. This listening mechanism enables each drone to perceive the presence and intentions of other competitors.

[0118] S203. When a neighboring drone receives an intent broadcast data packet from another neighboring drone, if the other neighboring drone is closer to the first drone or has an earlier broadcast timestamp, the neighboring drone cancels its own replacement intent.

[0119] When a neighboring drone receives an intent broadcast packet from another neighboring drone, it determines its priority based on predefined rules. Specifically, if the received packet indicates that another neighboring drone is closer to the first drone, it means that the first drone has a geographical advantage and can arrive and respond to the emergency more quickly; or, if the broadcast timestamp of another neighboring drone is earlier, this usually means that the first drone made the takeover decision and broadcast earlier. In both cases, the current neighboring drone cancels its own takeover intent, thus avoiding duplicate responses and wasted resources.

[0120] This application's solution effectively addresses the coordination problem when multiple neighboring drones simultaneously respond to an emergency task by introducing an intent broadcasting and competitive cancellation mechanism. Because each neighboring drone willing to take over the task broadcasts its intent and listens to the broadcasts of other drones, the drones in the cluster are able to autonomously exchange information and make decisions. By comparing the distances of other drones to the first drone and the broadcast timestamps, it ensures that the drone with a better geographical location or faster response time takes over the task first. This distributed decision-making process avoids the communication delays and single points of failure risks that centralized coordination may bring, while ensuring the efficiency and rationality of task allocation.

[0121] In some preferred embodiments, a specific example is given below. Suppose that the first UAV in the cluster, while performing a mission, suddenly encounters a communication outage, insufficient energy reserves, and identifies a sudden target—a forest fire. It then sends an emergency mission request to nearby UAVs. At this time, there are three nearby UAVs, UAV A, UAV B, and UAV C. They all receive the request and, after evaluation, determine that the value of taking over the emergency forest fire mission is higher than the cost of abandoning their current mission.

[0122] Specifically, drones A, B, and C will each encapsulate an intent broadcast data packet. For example, drone A's intent broadcast data packet includes its comparison result (indicating willingness to take over), its own identifier "UAV-A," and its current location information (e.g., 5 kilometers away from the first drone). Similarly, drones B and C also encapsulate and broadcast their respective data packets, with drone B 4 kilometers away from the first drone and drone C 6 kilometers away.

[0123] While broadcasting intent packets, the three drones continuously listened for broadcasts from other drones.

[0124] Suppose that drone B's broadcast timestamp is slightly earlier than drones A and C. When drones A and C receive drone B's intended broadcast data packet, they will make a judgment.

[0125] Drone A discovers that Drone B is closer to the first drone (4 km < 5 km), therefore Drone A will cancel its intention to take over.

[0126] Drone C discovers that Drone B is closer to the first drone (4 km < 6 km), therefore Drone C will also cancel its intention to take over.

[0127] Ultimately, only Drone B maintained its intention to take over and notified the cluster of this intention, awaiting confirmation from the first Drone. This mechanism ensured that Drone B, closest to the emergency target, took over the task first, thus achieving rapid task response and optimized resource allocation.

[0128] In some embodiments described above, this application proposes determining the value of an emergency task based on information about the sudden target. However, in practical applications, the environment in which the sudden target is located is often dynamically changing, such as fire spread, target movement, or deteriorating weather conditions. If the value of the emergency task remains unchanged after determination, it may not be able to reflect the actual urgency and importance of the task in a timely manner, thereby affecting the decision-making efficiency and mission response accuracy of the UAV swarm. Therefore, this application further proposes a scheme that, after determining the value of the emergency task, continuously monitors key factors of the environment in which the sudden target is located and dynamically adjusts the value of the emergency task based on changes in these key factors.

[0129] After determining the value metric for the emergency mission, the methodology also includes:

[0130] S301, a key factor in the continuous monitoring of the environment of sudden targets by nearby drones.

[0131] Key factors include the speed of fire spread, the trajectory of the target, and changes in weather conditions; the trajectory of the target refers to the movement of the sudden target.

[0132] Specifically, the nearby UAV collects visual images, temperature distribution data, wind speed and direction, and air pressure and temperature data of the environment where the sudden target is located; the nearby UAV processes the collected visual images, extracts the flame areas in the images, and calculates the fire spread rate based on the shape changes and pixel growth rate of the flame areas; the nearby UAV analyzes the collected temperature distribution data, identifies the boundaries of high-temperature areas, and corrects the fire spread rate based on the expansion rate of the high-temperature area boundaries; the nearby UAV performs target recognition and tracking on the collected visual images, obtains the position sequence of the sudden target in the image coordinate system, and converts the position sequence into the target movement trajectory in the spatial coordinate system by combining its own flight attitude and position information; the nearby UAV uses wind speed, direction, and air pressure and temperature data to obtain changes in meteorological conditions in the airspace where the sudden target is located.

[0133] Specifically, to achieve comprehensive monitoring of the environment surrounding a sudden target, nearby drones are equipped with a variety of sensors. For example, they can be equipped with high-definition visible light cameras to acquire visual images, infrared thermal imagers to obtain temperature distribution data, and integrated meteorological sensors to measure wind speed, wind direction, and air pressure and temperature data. These sensors work together to ensure multi-dimensional acquisition of environmental data.

[0134] When processing the acquired visual images, image segmentation algorithms can be used to identify flame regions within the images. Morphological changes in the flame regions, such as their area expansion, shape distortion, and pixel growth rate (i.e., the rate at which the number of pixels occupied by the flame region increases in consecutive frames), can all be used as a basis for calculating the fire spread rate. For example, the spread rate can be assessed by calculating the rate of change of the perimeter and area of ​​the flame region over time.

[0135] Furthermore, when analyzing the collected temperature distribution data, thermodynamic models and image processing techniques can be used to identify the boundaries of high-temperature regions. The expansion rate of the high-temperature region boundary, i.e., the spatial diffusion speed of the high-temperature region, can provide corrections for calculating the fire spread rate. For example, when there is a difference between the fire spread rate obtained from visual image analysis and the boundary expansion rate obtained from temperature distribution data analysis, a weighted average of the two or a correction based on a more reliable data source can be performed to obtain a more accurate fire spread rate.

[0136] Furthermore, to obtain the position sequence of sudden targets in the image coordinate system and convert it into a target trajectory in the spatial coordinate system, nearby UAVs can utilize deep learning models for target recognition, such as YOLO and Faster R-CNN, to accurately identify sudden targets in the image. Target tracking algorithms, such as Kalman filtering, particle filtering, or deep learning trackers (such as DeepSORT), can be used to track targets in consecutive frames, thereby obtaining their position sequence in the image coordinate system. Subsequently, combining the UAV's own flight attitude information (such as pitch angle, roll angle, and yaw angle) and position information (such as GPS coordinates), a coordinate transformation algorithm is used to convert the position sequence in the image coordinate system into a three-dimensional trajectory in the actual spatial coordinate system.

[0137] Finally, nearby drones can use data acquired by wind speed and direction sensors, as well as air pressure and temperature sensors, to monitor changes in meteorological conditions in the airspace where the target is located in real time. This data can be used to assess the impact of wind on the spread of fire, the impact of air pressure changes on drone flight, and the impact of temperature changes on mission execution, thus providing important environmental parameters for subsequent decision-making.

[0138] S302, neighboring drones dynamically adjust the value measurement of emergency missions based on changes in key factors.

[0139] In particular, when the fire spreads faster, the target's movement trajectory becomes irregular, or weather conditions worsen, the value of emergency missions should be increased.

[0140] This application's solution effectively addresses the limitation of the basic solution where the value of an emergency task may not reflect dynamic environmental changes in a timely manner by introducing a mechanism for continuous monitoring and dynamic adjustment of key environmental factors of the target's location. Specifically, in the basic solution, the value of an emergency task may remain fixed after it is determined, while actual emergencies are often accompanied by rapid environmental evolution. For example, a fire may spread rapidly, a trapped target may move within a danger zone, or severe weather may occur. If the value is not adjusted accordingly, the drone swarm may make decisions based on outdated or inaccurate assessments, leading to slow response or improper resource allocation. By continuously monitoring key factors such as the rate of fire spread, target movement trajectory, and changes in weather conditions, nearby drones can obtain the latest environmental dynamics in real time. When these key factors show a deteriorating trend, such as an accelerated fire spread, irregular target movement trajectory, or worsening weather conditions, the system immediately recognizes the increased urgency and importance of the task. Based on this, the value of the emergency task is dynamically increased, prompting nearby drones to prioritize taking over the emergency task. This mechanism ensures that the decision-making process of the drone swarm can fully consider the real-time changes in the mission environment, making the priority assessment of emergency missions more accurate and flexible, and avoiding decision-making errors caused by information lag.

[0141] In some preferred embodiments, assuming that during a forest fire monitoring mission, a first UAV, due to communication interruption and insufficient energy reserves, simultaneously identifies a new fire spot and issues an emergency mission request. Upon receiving the request, a neighboring UAV initially calculates the value of the emergency mission. While the neighboring UAV is preparing to make a decision, it continuously monitors that the fire's spread rate in the fire spot area is significantly accelerating, and wind speeds are suddenly increasing, indicating deteriorating weather conditions. According to the scheme of this application, the neighboring UAV will immediately identify these changes in key factors and dynamically increase the value of the emergency mission according to preset adjustment rules. For example, if the initial value is 80 points, it may be adjusted to 95 points after detecting the fire and worsening weather. This dynamic adjustment significantly elevates the priority of the emergency mission in the neighboring UAV's internal decision-making process. Even if its current mission has a high cost, it is more likely to prompt the neighboring UAV to autonomously decide to take over the emergency mission, thereby ensuring a timely response and effective control of the rapidly deteriorating fire situation.

[0142] Specifically, the aforementioned neighboring UAV performs target recognition and tracking on the acquired visual images, obtains the position sequence of the sudden target in the image coordinate system, and combines its own flight attitude and position information to convert the position sequence into the target movement trajectory in the spatial coordinate system. This can be achieved by following these steps.

[0143] S401, the nearby UAV applies target detection function to visual images to identify sudden targets in the image and obtain the initial position of the sudden targets.

[0144] Specifically, when applying target detection functionality to visual images using nearby UAVs, deep learning-based target detection models, such as the YOLO (You Only Look Once) series of algorithms or Faster R-CNN, can be employed to achieve rapid and accurate identification of sudden targets in the image. This function aims to accurately locate and identify sudden targets from complex backgrounds and obtain their initial position in the image, such as the target's bounding box coordinates or center point coordinates, providing a reliable starting point for subsequent tracking.

[0145] S402. The nearby UAV constructs a target appearance description based on the initial position of the sudden target.

[0146] The target appearance description includes information on the color distribution and texture features of the sudden target.

[0147] This target appearance description is crucial information for re-identifying and tracking the target in subsequent image frames. Color distribution information can be obtained by calculating the color histogram of the target region, for example, by performing histogram statistics in the HSV color space to capture the overall color characteristics of the target. Texture features can be extracted using feature descriptors such as Local Binary Pattern (LBP), Gabor filters, or Scale Invariant Feature Transform (SIFT) / Speeded Robust Feature Transform (SURF) to reflect the details and structure of the target surface. These descriptors provide a unique "fingerprint" of the target, maintaining robustness in identification even with slight changes to the target.

[0148] S403. The neighboring UAV continuously tracks the sudden target. When the sudden target is obstructed, deformed, or the lighting changes, the neighboring UAV predicts the position of the sudden target in the current frame based on the target's appearance description and the data from the neighboring UAV's continuous tracking of the sudden target, and obtains the predicted position.

[0149] For example, based on data from nearby drones continuously tracking a sudden target, the direction and speed of the sudden target can be determined, and based on the direction and speed of the sudden target, its position in the current frame can be predicted to obtain the predicted position.

[0150] S404. Nearby UAVs conduct local searches around the predicted location to update the position sequence of sudden targets in the image coordinate system.

[0151] Local search refers to precisely locating a target within a predefined region near the predicted location by comparing its appearance description with image features within that region. This local search strategy avoids time-consuming global searches across the entire image and can correct prediction errors, ensuring tracking accuracy. Updating the location sequence means adding newly identified target locations to the historical record, forming a complete motion trajectory.

[0152] S405, by combining its own flight attitude and position information, transforms the updated position sequence of a sudden target in the image coordinate system into the target's movement trajectory in the spatial coordinate system.

[0153] The position sequence in the image coordinate system is two-dimensional, reflecting only the pixel position of the target in the image. To obtain the target's movement trajectory in real three-dimensional space, it is necessary to combine the UAV's own flight attitude (such as pitch angle, roll angle, and yaw angle) and precise position information (such as GPS data and inertial navigation system data). Through camera intrinsic and extrinsic parameter calibration, the image coordinates can be converted into camera coordinates. Then, combined with the UAV's attitude and position, the target position is finally transformed into the global spatial coordinate system, thus obtaining the true movement trajectory of the sudden target in three-dimensional space.

[0154] This application's solution first applies target detection functionality to visual images, enabling rapid and accurate identification of sudden targets and acquisition of their initial positions, laying the foundation for subsequent precise tracking. Subsequently, based on the initial position, a target appearance description incorporating color distribution and texture features is constructed. This allows the UAV to utilize robust feature information for prediction and local search in complex environments such as target occlusion, deformation, or lighting changes, overcoming the limitation of traditional tracking methods that easily lose targets in complex scenes. By performing local searches around the predicted position and updating the target's position sequence in the image coordinate system, the continuity and accuracy of tracking are ensured. Finally, by combining the UAV's own flight attitude and position information, the two-dimensional image coordinates are converted into three-dimensional spatial coordinates, thereby obtaining a precise trajectory of the sudden target's movement. This refined target identification and tracking mechanism provides reliable and accurate input data for the aforementioned dynamic adjustment of emergency mission value metrics, enabling UAV swarms to respond to emergencies more promptly and accurately, improving the overall mission adaptability and decision-making quality.

[0155] The aforementioned nearby UAVs perform a local search around the predicted location to update the position sequence of the sudden target in the image coordinate system, including:

[0156] S501. The nearby UAV extracts features from the image region within the preset range of the predicted location to obtain the local feature set of the region.

[0157] The preset range can be a rectangular or circular region of a specific size centered on the predicted location. Its size can be dynamically adjusted according to factors such as target size, movement speed, and environmental complexity. Feature extraction aims to obtain visual information that can characterize the target's properties from these candidate regions. For example, local feature descriptors such as Scale Invariant Feature Transform (SIFT), Speed-Up Robust Features (SURF), Oriented Fast and Rotationally Brief Description (ORB) can be used, or high-dimensional feature vectors can be generated using deep learning-based feature extraction networks.

[0158] S502, the neighboring UAV will perform a multi-dimensional similarity comparison between the local feature set and the target appearance description.

[0159] The multi-dimensional similarity comparison includes color distribution similarity comparison and texture feature similarity comparison.

[0160] Multidimensional similarity comparison means not only relying on a single feature for matching, but also comprehensively considering multiple visual dimensions to improve matching accuracy. For example, color distribution similarity comparison can be achieved by calculating histogram cross, Bach distance, or chi-square distance; texture feature similarity comparison can be achieved by using Gabor filters, local binary patterns (LBP), or gray-level co-occurrence matrices to extract texture features and calculate their similarity.

[0161] S503 and neighboring drones calculate a comprehensive matching score based on multi-dimensional similarity comparison results.

[0162] The comprehensive matching score is the result of a weighted sum or fusion of color similarity and texture similarity scores, aiming to quantify the overall degree of matching between each candidate region and the target appearance description. The weighting coefficients can be adjusted according to the actual application scenario and target characteristics to highlight the feature dimensions that are more important for target recognition.

[0163] S504. When there are multiple areas with matching scores higher than the matching threshold, the neighboring drones combine the data from continuously tracking the sudden target and select the area with the most matching motion trajectory as the new location of the target.

[0164] This involves combining data from continuous tracking of sudden targets to predict the target's possible direction and velocity in the current frame. By comparing the relative position changes of candidate regions with the predicted motion trends, the region that best matches the target's dynamic behavior can be selected, thus effectively resolving tracking ambiguity issues under interference from multiple or similar targets.

[0165] S505. Based on the new target location, nearby UAVs update the position sequence of the sudden target in the image coordinate system.

[0166] Therefore, based on the determined new target location, the nearby UAV updates the position sequence of the sudden target in the image coordinate system. This updated position sequence will be used for subsequent target tracking and trajectory generation to ensure the continuity and accuracy of tracking.

[0167] This application's solution effectively addresses the problem of inaccurate target tracking caused by multiple similar candidate regions generated during local search in complex visual environments. Specifically, by introducing a multi-dimensional similarity comparison and a decision-making mechanism that combines target motion trend prediction, it comprehensively evaluates the matching degree between candidate regions and the target from multiple dimensions, improving matching accuracy. Furthermore, by calculating a comprehensive matching score, multi-dimensional information is effectively fused, providing a more reliable quantitative basis for target recognition. Especially in the case of ambiguity with multiple high-matching-score regions, combining the predicted target motion trend information allows for auxiliary judgment using the target's dynamic characteristics, eliminating visually similar but unpredictable motion trajectories, thus accurately selecting the true new target location. This mechanism enables the tracking system to maintain high robustness and accuracy even when facing challenges such as occlusion, deformation, or changes in illumination.

[0168] In some preferred embodiments, a specific example is given below. Suppose that during a city security patrol mission, a first drone identifies an emergency requiring immediate response, such as discovering a suspicious person engaging in sabotage, and sends an emergency mission request to a neighboring drone. After taking over the mission, the neighboring drone needs to continuously track the suspicious person. At some point, due to the suspicious person entering a densely populated area, moving between buildings causing partial obstruction of their view, or a drastic change in lighting conditions, the neighboring drone predicts the possible location of the suspicious person and conducts a localized search around that location.

[0169] Specifically, the nearby UAV first divides the image region around the predicted location into a grid, and extracts color histograms and Local Binary Pattern (LBP) texture features from each grid region to form a set of local features. At the same time, the nearby UAV has stored target appearance descriptions of suspicious persons, including the distribution of clothing color and texture features of their face or body.

[0170] Subsequently, the neighboring drone compares the local feature set of each grid region with the target appearance description of the suspicious person for color distribution similarity and texture feature similarity, and calculates the comprehensive matching score for each grid region. For example, the comprehensive score is obtained by calculating the Bach distance of the color histogram and the Euclidean distance of the LBP feature and performing a weighted sum.

[0171] Suppose that within a local search area, three grid regions all have a combined matching score higher than a preset matching threshold, indicating that these three regions may contain suspicious individuals. In this case, relying solely on visual similarity would be insufficient to determine the true location of the suspicious individual. The solution proposed in this application addresses this issue: a neighboring UAV combines historical movement trend prediction information of the suspicious individual. For example, based on the movement direction and speed of the suspicious individual in previous frames, it predicts the area the suspicious individual is most likely to move to in the current frame. By comparing the relative positions of these three high-scoring regions with the predicted movement trends, the neighboring UAV selects the region whose movement trajectory best matches the new location of the suspicious individual. For example, if the suspicious individual is predicted to move northwest, the region among the three high-scoring regions that is located in the predicted northwest direction and has the smallest deviation from the predicted trajectory is selected.

[0172] Based on the newly selected location of the suspect, the nearby drone updates the suspect's position sequence in the image coordinate system, thereby ensuring that the tracking of suspects can remain accurate in complex and ever-changing environments, providing reliable data support for subsequent emergency response and law enforcement operations.

[0173] like Figure 3 As shown in the figure, this embodiment of the invention also provides a collaborative dynamic replacement endurance control system for unmanned aerial vehicle (UAV) swarms. The system includes:

[0174] The emergency mission request module is used to send an emergency mission request to nearby drones via inter-drone communication when the first drone in the cluster meets a first condition. The first condition is that the first drone's communication with the external command is interrupted, its own energy reserves are lower than a preset level, and it has identified a sudden target that requires immediate response. The emergency mission request includes information about the sudden target and the status information of the first drone.

[0175] The value assessment module is used to receive emergency mission requests from nearby drones and determine the value of the emergency mission based on the information of the sudden target; and to determine the cost of nearby drones abandoning their current mission.

[0176] The task decision module is used to compare the value and cost of an emergency task with that of a neighboring drone. When the value of the emergency task is higher than the cost, the neighboring drone autonomously decides to take over the emergency task and notifies the cluster of its intention to take over through inter-drone communication.

[0177] The task handover module is used for the first UAV to receive the takeover intention of the neighboring UAV and send the takeover confirmation information to the neighboring UAV; in response to the takeover confirmation information, the neighboring UAV switches its own task status to perform the emergency task; and before switching the task status, it transmits the key data of the original task of the neighboring UAV to other UAVs in the cluster.

[0178] The safe return module is used to initiate the safe return process after the first UAV sends the takeover confirmation information. The safe return process includes planning a path that allows the first UAV to land safely before its energy reserves are depleted, and transmitting the key data collected by the first UAV during the return process.

[0179] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by a computer program instructing related hardware. This program can be stored in the computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be an internal storage unit of the task execution device (including a data sending end and / or a data receiving end) of any of the foregoing embodiments, such as the hard disk or memory of the task execution device. The computer-readable storage medium can also be an external storage device of the terminal device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device. Further, the computer-readable storage medium can include both the internal storage unit of the task execution device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the task execution device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0180] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0181] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0182] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A method for collaborative dynamic replacement of endurance control in a drone swarm, characterized in that, include: When the first drone in the cluster meets the first condition, the first drone sends an emergency mission request to the reachable neighboring drones through inter-drone communication. The first condition is that the first UAV loses communication with external command, its own energy reserves are lower than a preset level, and it identifies a sudden target that requires immediate response; the emergency mission request includes information about the sudden target and the status information of the first UAV; The nearby drone receives the emergency mission request and determines the value of the emergency mission based on the information of the sudden target. And, determine the cost metric for the neighboring drone abandoning its current mission; The neighboring drones are compared with the value measure of the emergency mission and the cost measure; When the value of the emergency mission is higher than the cost, the neighboring drone autonomously decides to take over the emergency mission. And notify the cluster of the takeover intention through inter-machine communication; The first UAV receives the replacement intention from the neighboring UAV and sends a replacement confirmation message to the neighboring UAV; in response to the replacement confirmation message, the neighboring UAV switches its own task status to perform the emergency task; Before switching mission states, the key data of the original mission of the neighboring drones are transmitted to other drones in the cluster. After sending the handover confirmation information, the first drone initiates the safe return-to-home process; The safe return process includes planning a path that allows the first UAV to land safely before its energy reserves are depleted, and transmitting the key data collected by the first UAV during the return process.

2. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The value of the emergency mission satisfies the following relationship: ； Wherein, Score_Emergency is the value measure of the emergency task, Urgency is the urgency level of the emergency task, Timeliness is the response timeliness value, and Importance is the importance value of the overall task objective; W1, W2, and W3 are preset weight coefficients.

3. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The cost of a neighboring drone abandoning its current task satisfies the following relationship: ； Wherein, Cost_Current is the cost metric, Risk_Level is the risk level of the current task area, Blank_Duration is the duration of the monitoring gap caused by the neighboring drone abandoning its current task, and Inherent_Importance is the inherent importance of the neighboring drone's own current task. W4, W5, and W6 are preset weighting coefficients.

4. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, When the value measure of the emergency task is equal to the cost measure, the method further includes: The neighboring drone extracts the type of the emergency task from the emergency task request and obtains the type of its own current task from its own task management module; The neighboring drone matches the type of the emergency task with the type of its own current task according to a preset mapping relationship to determine the priority of the emergency task and the priority of its own current task; the preset mapping relationship includes the mapping relationship between different task types and different priorities. The neighboring drone compares the priority of the emergency task with the priority of its own current task, and selects the task with the higher priority as the decision result. The neighboring drones issue action commands corresponding to the decision result and notify the cluster of the replacement intention through inter-drone communication.

5. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The number of neighboring drones is multiple. When all the neighboring drones determine that the value of the emergency mission is higher than the cost value, the method further includes: For each of the multiple neighboring drones, each neighboring drone encapsulates an intent broadcast data packet; the intent broadcast data packet includes a comparison result, its own identifier, and current location information; the comparison result is used to compare the value measure of the emergency mission with the cost measure. For each of a plurality of neighboring drones, the neighboring drone continuously listens for intent broadcast packets from other neighboring drones; When a neighboring drone receives an intent broadcast data packet from another neighboring drone, if the other neighboring drone is closer to the first drone or has an earlier broadcast timestamp, the neighboring drone cancels its own replacement intent.

6. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 2, characterized in that, After determining the value measure of the emergency task, the method further includes: The nearby drone continuously monitors key environmental factors of the sudden target; these key factors include the fire spread rate, the target's movement trajectory, and changes in weather conditions; the target's movement trajectory is the movement trajectory of the sudden target. The nearby drone dynamically adjusts the value of the emergency mission based on changes in the key factors; specifically, the value of the emergency mission is increased when the fire spreads faster, the target's movement trajectory becomes irregular, or the weather conditions worsen.

7. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 6, characterized in that, The key factors that the nearby drone continuously monitors in the environment of the sudden target include: The nearby drone collects visual images, temperature distribution data, wind speed and direction, and air pressure and temperature data of the environment where the sudden target is located. The nearby UAV processes the acquired visual images, extracts the flame areas in the images, and calculates the fire spread rate based on the shape changes and pixel growth rate of the flame areas. The nearby drone analyzes the collected temperature distribution data, identifies the boundaries of high-temperature areas, and corrects the fire spread rate based on the expansion rate of the high-temperature area boundaries. The nearby UAV performs target recognition and tracking on the acquired visual images, obtains the position sequence of the sudden target in the image coordinate system, and combines its own flight attitude and position information to convert the position sequence into the target movement trajectory in the spatial coordinate system. The nearby UAV uses the wind speed, wind direction, and air pressure and temperature data to obtain information on changes in meteorological conditions in the airspace where the sudden target is located.

8. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 7, characterized in that, The nearby UAV performs target recognition and tracking on the acquired visual images, obtains the position sequence of the sudden target in the image coordinate system, and combines its own flight attitude and position information to convert the position sequence into the target's movement trajectory in the spatial coordinate system, including: The nearby UAV applies target detection functionality to the visual images to identify sudden targets in the images and obtain the initial position of the sudden targets; The neighboring UAV constructs a target appearance description based on the initial position of the sudden target; the target appearance description includes the color distribution information and texture feature information of the sudden target; The neighboring UAV continuously tracks the sudden target. When the sudden target is obstructed, deformed, or the lighting changes, the neighboring UAV predicts the position of the sudden target in the current frame based on the target appearance description and the data of the neighboring UAV continuously tracking the sudden target, and obtains the predicted position. The nearby UAVs perform a local search around the predicted location to update the position sequence of the sudden target in the image coordinate system; The neighboring UAV, combining its own flight attitude and position information, converts the updated position sequence of the sudden target in the image coordinate system into the target's movement trajectory in the spatial coordinate system.

9. The method for collaborative dynamic replacement endurance control of unmanned aerial vehicle (UAV) swarms according to claim 8, characterized in that, The nearby UAVs perform a local search around the predicted location to update the position sequence of the sudden target in the image coordinate system, including: The neighboring UAV extracts features from an image region within a preset range of the predicted location to obtain a set of local features for that region. The neighboring UAV performs a multi-dimensional similarity comparison between the local feature set and the target appearance description; the multi-dimensional similarity comparison includes color distribution similarity comparison and texture feature similarity comparison; The neighboring drones calculate a comprehensive matching score based on the multi-dimensional similarity comparison results; When there are multiple regions with matching scores higher than the matching threshold, the neighboring UAV combines the data from continuously tracking the sudden target and selects the region with the most matching motion trajectory as the new location of the target. Based on the new location of the target, the neighboring UAV updates the position sequence of the sudden target in the image coordinate system.

10. A collaborative dynamic replacement endurance control system for unmanned aerial vehicle (UAV) swarms, characterized in that, The system includes: An emergency mission request module is used to send an emergency mission request to nearby drones via inter-drone communication when the first drone in the cluster meets a first condition. The first condition is that the first drone's communication with external command is interrupted, its own energy reserves are lower than a preset level, and it has identified a sudden target that requires immediate response. The emergency mission request includes information about the sudden target and the status information of the first drone. The value assessment module is used to determine the value of the emergency mission when the nearby UAV receives the emergency mission request, and to determine the cost of the nearby UAV abandoning its current mission based on the information of the sudden target; The task decision module is used by the neighboring UAVs to compare the value of the emergency task with the cost of the emergency task; when the value of the emergency task is higher than the cost of the emergency task, the neighboring UAVs autonomously decide to take over the emergency task; and notify the cluster of the takeover intention through inter-UAV communication. The task handover module is used for the first UAV to receive the takeover intention of the neighboring UAV and send the takeover confirmation information to the neighboring UAV; in response to the takeover confirmation information, the neighboring UAV switches its own task status to execute the emergency task; and before switching the task status, it transmits the key data of the original task of the neighboring UAV to other UAVs in the cluster. The safe return-to-home module is used to initiate a safe return-to-home process after the first UAV sends the handover confirmation information. The safe return-to-home process includes planning a path that allows the first UAV to land safely before its energy reserves are depleted, and transmitting the key data collected by the first UAV during the return-to-home process.

Images