A conflict avoidance method and system for cooperative work of a drone cluster

Abstract

The application discloses a conflict avoidance method and system for cooperative work of a UAV cluster, and the method comprises the following steps: acquiring a task time limit parameter and motion state information; determining a task urgency and converting the task urgency into a negotiation priority weight; generating a time-space flight pipeline with preemption characteristics as intention information and broadcasting; combining neighborhood intention information to predict a conflict space region and a time window; judging whether the conflict belongs to a negotiable conflict according to a priority weight difference and a time window urgency, and triggering a distributed negotiation protocol when the conflict is negotiable; adjusting the pipeline by a low-priority UAV, checking by a high-priority UAV, and iteratively generating a conflict-free cooperative trajectory; and executing the cooperative trajectory and triggering a closed-loop re-negotiation by using a flight deviation. The application links the task urgency and the distributed intention negotiation, which is helpful to guarantee the priority passing of an emergency task and reduce redundant avoidance actions.

Description

Technical Field

[0001] This application belongs to the field of unmanned aerial vehicle (UAV) swarm control, specifically relating to a conflict avoidance method and system for UAV swarm collaborative operations. Background Technology

[0002] Collaborative drone swarm operations have been applied in areas such as regional inspection, logistics transportation, and environmental monitoring. Common conflict avoidance methods mainly include centralized path planning and situational reactive collision avoidance. Centralized path planning typically involves a ground station or central node uniformly calculating the flight trajectories of each drone. When the number of tasks increases or changes dynamically, the computational load and communication overhead increase significantly, and the risk of single-point failure is high. Situational reactive collision avoidance relies on each drone's perception of its surrounding environment to perform real-time avoidance. It has a fast response speed, but often lacks the prediction and coordination of future mission intentions, making it difficult to distinguish differences in mission urgency. This can easily lead to delays in critical missions or unnecessary chain avoidance actions between multiple drones. Especially when missions have dynamically changing time constraints and inter-drone communication may be intermittent, the difficulty of balancing operational efficiency and flight safety with the above solutions increases further.

[0003] A key reason for the aforementioned problems is that existing solutions fail to effectively link real-time task urgency with distributed intent negotiation. In a distributed environment, the lack of a unified measure of task urgency among drones, along with a corresponding negotiation priority, prevents them from proactively reserving sufficient time and space leeway for high-urgency tasks in the event of potential conflicts. Simultaneously, low-urgency tasks lack a basis for orderly concession, leading to unnecessary evasive maneuvers or negotiation stalemates. Therefore, providing a method that can map task urgency to negotiation priority in real time and achieve differentiated conflict resolution through distributed negotiation has become a pressing technical problem. Summary of the Invention

[0004] This application provides a conflict avoidance method and system for collaborative operation of unmanned aerial vehicle (UAV) swarms, which at least addresses the problems existing in the prior art.

[0005] A first aspect of this application provides a conflict avoidance method for collaborative operation of unmanned aerial vehicle (UAV) swarms, comprising the following steps: S1. Obtain the task time limit parameters and real-time motion status information of each drone in the drone cluster; S2. Based on the mission time limit parameters and the current time, determine the mission urgency of the drone and convert it into a negotiated priority weight; S3. Based on real-time motion state information and negotiation priority weights, generate a spatiotemporal flight pipeline with preemption characteristics as intent information, and broadcast it to neighboring UAVs within the communication range. S4. Receive intent information broadcast by neighboring drones, combine it with the local planned mission path and motion status, and predict potential conflict areas and time windows. S5. Utilize the difference in negotiation priority weights between the local drone and neighboring drones and the urgency of the time window to determine whether the conflict is a negotiable conflict, and trigger the distributed negotiation protocol when it is determined to be a negotiable conflict. S6. In the distributed negotiation protocol, the drone with the lower negotiation priority weight adjusts its own spatiotemporal flight path and sends the adjusted intention information. The drone with the higher negotiation priority weight verifies whether the adjusted intention information eliminates the conflict. The process is iterated until a conflict-free cooperative flight trajectory is generated. S7. The UAV flies along a conflict-free cooperative flight trajectory and monitors flight deviations using real-time motion status. When the flight deviation exceeds the preset tolerance, it returns to S2 to redetermine the mission urgency and negotiate priority weights, and then executes subsequent steps.

[0006] This application transforms task time-limit parameters into negotiation priority weights in real time, and uses these to generate spatiotemporal flight pipelines with preemptive attributes for intent broadcasting, enabling the explicit expression and transmission of the task urgency of each UAV. Based on this, negotiable conflicts are distinguished according to priority weight differences and time window urgency. Conflicts are then resolved through an iterative process where lower-priority parties concede and higher-priority parties verify, ultimately forming a conflict-free collaborative flight trajectory. This scheme effectively links task urgency-driven processes with distributed intent negotiation, ensuring that urgent tasks pass through conflict zones first while reducing unnecessary evasive actions, thus improving the overall efficiency and robustness of cluster task execution. The deviation feedback mechanism further introduces closed-loop correction capabilities, enhancing adaptability in dynamic environments.

[0007] In this embodiment of the application, step S2 determines the mission urgency of the UAV based on the mission time limit parameter and the current time, and converts it into a negotiated priority weight, specifically including: S21. Obtain the task value weight corresponding to the task time limit parameter, and calculate the remaining execution time from the current time to the task deadline. S22. Use the ratio of task value weight to remaining execution time as the basic task urgency. S23. Using a preset monotonic nonlinear mapping function, the urgency of the basic task is converted into a normalized negotiation priority weight. The mapping function makes the rate of change of the negotiation priority weight greater when the remaining execution time is shorter.

[0008] This application constructs a basic task urgency by introducing the ratio of task value weight to remaining execution time, and uses a nonlinear mapping for normalization, so that the negotiation priority weight has higher sensitivity and discriminativeness when the task is close to the deadline, and more accurately reflects the true urgency of the task, providing a reliable quantitative basis for subsequent differentiated negotiation.

[0009] In this embodiment of the application, S3 generates a spatiotemporal flight pipeline with preemption characteristics as intent information based on real-time motion state information and negotiation priority weights, specifically including: S31. Based on real-time motion status and planned mission path, generate a nominal safe flight corridor for the aircraft within a future preset time domain, defined by position boundaries and speed range; S32. Based on the value of the negotiation priority weight, the expansion margin is superimposed on the spatial dimension of the nominal safe flight corridor, and the available time period is extended in the time dimension. The expansion range increases with the increase of the negotiation priority weight, forming a spatiotemporal flight channel with preemption characteristics. S33. Write priority weight flags into the broadcast message of the spacetime flight pipeline and encapsulate them as intent information.

[0010] This application expands the nominal safe flight corridor in space and time based on negotiation priority weights. High-priority tasks obtain greater space and time occupancy margins, which reflects the preemptive characteristics from the perspective of intent expression. This allows other UAVs to perceive in advance and reserve avoidance space, reducing the frequency and cost of subsequent conflict resolution negotiations.

[0011] In this embodiment of the application, S4 receives the intent information broadcast by neighboring UAVs, and combines it with the local planned task path and motion state to predict the spatial areas and time windows with potential conflicts, specifically including: S41. Parse the received intent information and construct the occupancy grid set of the corresponding neighboring UAV in the time-space dimension. The occupancy grid set includes a three-dimensional spatial grid index and the corresponding occupancy time slot. S42. Discretize the local planned task path into a sequence of track points carrying timestamps according to a preset time step; S43. Perform spatiotemporal intersection detection on the track point sequence and the occupied grid set, and take the overlapping spatial grids and their covered time intervals as potential conflict spatial regions and time windows.

[0012] This application employs a spatiotemporal intersection detection method based on the occupancy grid set and discrete track point sequence, which can accurately obtain the specific spatial range and time interval of the conflict occurrence. This provides accurate input for the subsequent fine-grained determination of the conflict type and the quantitative calculation of the negotiation step size, avoiding misjudgment or over-avoidance caused by rough prediction.

[0013] In this embodiment, step S5 utilizes the difference in negotiation priority weights between the local drone and neighboring drones, as well as the urgency of the time window, to determine whether a conflict is a negotiable conflict. Specifically, this includes: S51. Calculate the negotiation priority weight difference between the local machine and the neighboring UAVs that caused the conflict; S52, Get the duration of the time window; S53. When the difference in negotiation priority weights is less than the preset negotiation difference threshold and the duration is greater than the minimum negotiation window threshold, the conflict is determined to be a negotiable conflict; otherwise, it is determined to be an emergency conflict. S54. When an emergency conflict is identified, an emergency avoidance flag is generated and the distributed negotiation protocol is suspended.

[0014] This application distinguishes between negotiable conflicts and urgent conflicts by combining the priority weight difference and the conflict time window duration. Conflicts with similar weights and sufficient negotiation time are allowed to enter the negotiation process, while conflicts with disparate weights or tight time are directly transferred to emergency avoidance. This not only improves the negotiation success rate but also ensures the safe passage of high-priority tasks.

[0015] In this embodiment of the application, when an emergency conflict is determined, the method further includes: S5a: Trigger a priority-based emergency avoidance protocol, commanding the drone with the lower priority weight to perform the maximum available maneuver to avoid the obstacle under its own constraints; S5b: UAVs with higher negotiation priority weights maintain their spatiotemporal flight pipeline unchanged within the time window until the time window ends.

[0016] This application mandates that low-priority drones perform the maximum available maneuvers during emergency conflicts, while high-priority drones maintain their original routes. This ensures that high-urgency missions can still pass through conflict zones unaffected even if negotiations fail, while avoiding additional uncertainties caused by two-way adjustments.

[0017] In this embodiment of the application, in S6, the drone with the lower negotiation priority weight adjusts its own spatiotemporal flight path and sends the adjusted intent information. The drone with the higher negotiation priority weight verifies whether the adjusted intent information eliminates the conflict. Specifically, this includes: S61. For UAVs with lower negotiation priority weights, the amount of spatiotemporal flight pipeline space contraction or time offset required to eliminate the conflict is calculated based on the conflict geometry. S62. Modify the spatial boundary parameters or time interval parameters of the space-time flight pipeline step by step according to the preset adjustment step size. After each step of modification, keep the pipeline continuous and do not cause new conflicts with its own mission path. S63. After each modification is completed, the updated intent information is broadcast, and the drone with the higher priority weight in the negotiation performs conflict detection and verification. S64. If the conflict disappears, the current parameters are fixed as part of the cooperative trajectory. If the conflict does not disappear and the cumulative adjustment steps have not reached the preset concession limit, return to S62 and continue to modify according to the adjustment step size. If the conflict still exists after reaching the preset concession limit, pause the adjustment and send a concession limit notification to the high-priority weight drone.

[0018] This application adopts an iterative negotiation mechanism of step-by-step adjustment and step-by-step verification. The low-priority party makes concessions in a controllable step size, and the high-priority party verifies in real time. This not only ensures the convergence of the conflict resolution process, but also avoids the low-priority party from making unlimited concessions through the concession upper limit mechanism, thereby improving the fairness and controllability of the negotiation.

[0019] In this embodiment of the application, before S5 triggers the distributed negotiation protocol, it further includes: S50. Detect the quality of the current communication link with neighboring drones; if the quality of the communication link is lower than the preset reliability threshold, determine the drone with the highest weight value from the drones that are accessible to the communication as a temporary coordination node according to the negotiation priority weight. The temporary coordination node collects the intention information of each drone, generates a safe trajectory using a centralized conflict resolution algorithm, and broadcasts it to each drone.

[0020] When the communication link quality deteriorates to an unreliable level, this application automatically switches to a centralized resolution mode in which the highest priority drone acts as a temporary coordination node, ensuring the continuous availability of the conflict avoidance function under conditions of limited or intermittent communication and enhancing the robustness of the system.

[0021] In this embodiment of the application, before the UAV with the lower negotiation priority weight adjusts its own spatiotemporal flight path in step S6, it further includes: Query the historical concession records between this machine and the high-priority drone involved in the current conflict; if the historical concession records show that the number of concessions made by this machine exceeds the preset value, reduce the upper limit of the adjustment range, or request temporary compensation for the priority weight from the high-priority drone.

[0022] This application introduces a historical concession memory mechanism to prevent the same low-priority drone from continuously yielding in multiple rounds of conflict, which would cause serious delays to its own mission. By limiting the concession range or requesting priority compensation, the fairness of distributed negotiation among different missions is improved.

[0023] In this embodiment of the application, when a drone simultaneously encounters multiple conflicts with multiple neighboring drones, the method further includes: Obtain the start time of the time window corresponding to each conflict; Multiple conflicts are sorted according to their starting time; for conflicts with the same starting time, they are further sorted from largest to smallest according to the difference in the negotiation priority weight involved. Based on the sorting results, a distributed negotiation protocol is initiated for each conflict in turn, and during the processing of the current conflict, the spatiotemporal flight pipeline associated with the conflict later in the time window is locked.

[0024] This application introduces a sorting mechanism based on time window order and weight difference for multi-conflict scenarios, and locks the subsequent conflict pipeline to avoid trajectory oscillation and repeated adjustments caused by parallel negotiation of multiple conflicts, so that the multi-conflict resolution process converges in an orderly manner.

[0025] In this embodiment of the application, when returning to S2 in S7 to re-determine the task urgency and negotiate the priority weight, if it is detected that the current task deadline has expired or the task has been completed, the determination of task urgency is stopped, the machine's spacetime flight pipeline is released, and the cooperative conflict avoidance process is exited.

[0026] This application releases the spacetime flight pipeline and exits the process in a timely manner when the mission ends or is completed, avoiding unnecessary continuous constraints on other drones in the cluster, which is conducive to improving the overall utilization efficiency of airspace resources.

[0027] In this embodiment of the application, when determining the task urgency of the UAV in S2, the remaining energy status of the UAV is also incorporated; if the remaining energy status is lower than the preset energy threshold, the determined task urgency is adjusted to reduce the negotiation priority weight accordingly.

[0028] This application incorporates the remaining energy status into the correction factor for mission urgency, which appropriately reduces the priority of low-energy UAVs and reduces the number of complex avoidance operations that they perform beyond their energy support range due to high priority settings. This helps to improve the mission completion rate of individual UAVs and the overall rationality of cluster energy use.

[0029] In this embodiment, before the potential conflict is predicted in S4, the predicted intent information of the neighboring UAVs in the future is generated using a preset intent inference model based on the received historical intent information sequence of the neighboring UAVs and their task time limit parameters. The predicted intent information is then checked for consistency with the actual intent information broadcast by the neighboring UAVs. When the deviation between the predicted intent information and the actual intent information is lower than a preset confidence threshold, the neighboring UAV is marked as a predictable object. Based on the predicted intent information of the predictable object, the time starting point of the conflict prediction in S4 is extended to the future by a preset prediction duration. Conflicts that may occur within the extended time period are pre-identified, and the negotiable conflict judgment process in S5 is initiated in advance.

[0030] This step, occurring earlier than the actual occurrence of a conflict, utilizes the historical intent sequences and mission parameters of neighboring drones to construct a behavioral prediction model, anticipating their potential future flight path occupancy trends. By verifying the consistency between the predicted and actual broadcast information, predictable drones with stable behavioral patterns are selected. Conflict assessment is initiated only for these pre-emptively, avoiding additional negotiation overhead caused by inaccurate predictions. Extending the conflict detection time window forward allows for earlier discovery and resolution of negotiable conflicts, providing lower-priority parties with more time to adjust. This helps reduce the proportion of conflicts escalating to emergency conflicts due to insufficient negotiation windows, improving the overall throughput efficiency of the cluster in multi-drone dynamic interaction scenarios.

[0031] In this embodiment, when the drone with the lower priority weight calculates the amount of spatial contraction or time offset required to eliminate the conflict in S6, the delay loss caused by the adjustment to its own task completion time and the incremental collision risk introduced by the pipeline contraction are evaluated simultaneously. Based on the task delay loss and the incremental collision risk, an adjustment cost function is constructed, and the value of the adjustment cost function is compared with the estimated value of the task delay cost that the high priority party may bear when the conflict is not resolved. When the value of the adjustment cost function exceeds the product of the estimated value of the task delay cost of the high priority party and the preset tolerance coefficient, the concession adjustment is suspended and a cost unacceptable notification is sent to the high priority weight drone. After the two parties exchange cost information, they re-evaluate whether to continue the negotiation or switch to other conflict handling branches.

[0032] This step introduces a quantitative assessment of the costs incurred by the lower-priority party before it makes concessions, upgrading the unilateral concession mechanism into a multi-objective trade-off mechanism that considers the overall losses of both parties. The lower-priority party not only calculates the geometric adjustment required to avoid the concession but also simultaneously assesses the task delays and secondary risks brought about by the adjustment, forming a quantitative expression of the adjustment cost. By comparing the potential delay costs with those of the higher-priority party, unreasonable excessive concessions are prevented when the lower-priority party's costs significantly outweigh the higher-priority party's gains, avoiding sacrificing the greater interests of one party to protect the smaller interests of the other. The exchange of cost information between the two parties helps to find a more balanced conflict resolution path based on fairness, improving the overall operational efficiency and task completion balance of the cluster.

[0033] A second aspect of this application provides a conflict avoidance system for collaborative operation of unmanned aerial vehicle (UAV) swarms, comprising a task management unit, a data acquisition unit, a urgency calculation unit, an intent management unit, a conflict detection unit, a negotiation control unit, and a flight monitoring unit. The task management unit is used to store and provide task time limit parameters for each UAV; The data acquisition unit, connected to the task management unit, is used to collect real-time motion status information and obtain task time limit parameters from the task management unit. The urgency calculation unit, connected to the data acquisition unit, is used to determine the urgency of the task based on the acquired task time limit parameters and the current time, and convert it into a negotiated priority weight. The intent management unit, connected to the data acquisition unit and the urgency calculation unit, is used to generate a spatiotemporal flight pipeline with preemption characteristics as intent information based on real-time motion state information and negotiated priority weights, and broadcast it to neighboring UAVs within the communication range. The conflict detection unit, connected to the intent management unit, is used to receive intent information broadcast by neighboring UAVs, and, in combination with the local planned mission path and motion status, predict the spatial areas and time windows where potential conflicts exist. The negotiation control unit, connected to the conflict detection unit and the intent management unit, is used to determine whether a conflict is negotiable by using the difference in negotiation priority weights between the local and neighboring UAVs and the urgency of the time window. When a conflict is determined to be negotiable, the distributed negotiation protocol is triggered, and the UAV with lower priority weights is controlled to adjust its spatiotemporal flight path, while the UAV with higher priority weights verifies the adjustment results. Through iteration, a conflict-free cooperative flight trajectory is generated. The flight monitoring unit, connected to the negotiation control unit and the data acquisition unit, is used to control the UAV to fly along a conflict-free cooperative flight trajectory and to monitor flight deviations using real-time motion status. When the flight deviation exceeds the preset tolerance, the urgency calculation unit is triggered to re-determine the mission urgency and negotiation priority weights.

[0034] The system provided in this application integrates task urgency calculation, preemptive intent broadcasting, conflict spatiotemporal prediction, hierarchical negotiation iteration, and flight deviation closed-loop monitoring into a complete distributed collaborative architecture through orderly connection and functional division between units. The functions of each unit correspond one-to-one with the independent method steps, enabling the system's information flow and processing logic to fully support the execution of conflict avoidance methods and possessing engineering feasibility for dynamic tasks and restricted communication scenarios. Attached Figure Description

[0035] Figure 1 This is a flowchart illustrating a conflict avoidance method for collaborative operation of unmanned aerial vehicle (UAV) swarms, as provided in an embodiment of this application. Detailed Implementation

[0036] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0037] In existing UAV swarm collaborative operations, conflict avoidance schemes mostly adopt centralized path planning or reactive collision avoidance based on local situational awareness. Under conditions of dynamically changing mission time limits and potentially limited inter-UAV communication, centralized schemes have high computational and communication overhead and a high risk of single point of failure; reactive collision avoidance is difficult to distinguish between different mission urgency levels, which can easily lead to delays in high-urgency missions or unnecessary chain avoidance.

[0038] One major reason for the aforementioned problems is that existing solutions fail to effectively map the real-time urgency of tasks into priority metrics that can be used for distributed coordination, and also fail to link this priority with intent expression and negotiation mechanisms, resulting in a lack of differentiated concession criteria during conflict resolution.

[0039] This application proposes a distributed negotiation conflict avoidance scheme based on dynamic urgency. The urgency of the task is calculated in real time through task time limit parameters and mapped to negotiation priority weights. A spatiotemporal flight pipeline with preemption attribute is generated as a broadcast intent. After receiving the broadcast intent, the neighboring machine combines its own path to predict the conflict area and time window. Based on the weight difference and window urgency, hierarchical negotiation is triggered. The low-priority party concedes and the high-priority party verifies. The conflict-free cooperative trajectory is generated iteratively, and the flight deviation feedback is used to trigger renegotiation.

[0040] To make the purpose, technical solution, and advantages of this application clearer, the following will be described in conjunction with the appendix. Figure 1 The following is an explanation using specific examples.

[0041] Please refer to Figure 1 , Figure 1 A first aspect of this application provides a conflict avoidance method for collaborative operation of unmanned aerial vehicle (UAV) swarms, comprising the following steps: S1. Obtain the task time limit parameters and real-time motion status information of each drone in the drone cluster; S2. Based on the mission time limit parameters and the current time, determine the mission urgency of the drone and convert it into a negotiated priority weight; S3. Based on real-time motion state information and negotiation priority weights, generate a spatiotemporal flight pipeline with preemption characteristics as intent information, and broadcast it to neighboring UAVs within the communication range. S4. Receive intent information broadcast by neighboring drones, combine it with the local planned mission path and motion status, and predict potential conflict areas and time windows. S5. Utilize the difference in negotiation priority weights between the local drone and neighboring drones and the urgency of the time window to determine whether the conflict is a negotiable conflict, and trigger the distributed negotiation protocol when it is determined to be a negotiable conflict. S6. In the distributed negotiation protocol, the drone with the lower negotiation priority weight adjusts its own spatiotemporal flight path and sends the adjusted intention information. The drone with the higher negotiation priority weight verifies whether the adjusted intention information eliminates the conflict. The process is iterated until a conflict-free cooperative flight trajectory is generated. S7. The UAV flies along a conflict-free cooperative flight trajectory and monitors flight deviations using real-time motion status. When the flight deviation exceeds the preset tolerance, it returns to S2 to redetermine the mission urgency and negotiate priority weights, and then executes subsequent steps.

[0042] In this embodiment, in S1, the task time limit parameter refers to information related to the time constraints of the task currently being performed by the UAV. For example, it may include the latest completion time of the task, the total time budget for task execution, or the estimated time consumption of the task at different stages. Real-time motion state information refers to instantaneous data that describes the motion of the UAV in three-dimensional space. This typically includes the UAV's position coordinates, flight velocity vector, heading angle, and acceleration and angular velocity acquired by onboard sensors. This information collectively describes the dynamic state of each UAV at the current moment.

[0043] In S2, task urgency is a quantitative indicator used to measure the urgency of a task at the current point in time, or the margin for delay. Task urgency is generally higher when the task deadline parameter indicates that the deadline is approaching or the available task buffer time is decreasing. Negotiation priority weight is a numerical label positively correlated with task urgency, used to quantitatively indicate the drone's priority during subsequent negotiation processes. The process of converting task urgency into negotiation priority weight can be implemented using a preset function or mapping table. For example, the urgency value can be linearly or non-linearly mapped to a standard range, such as [0,1], where a larger value represents a higher priority.

[0044] In S3, a preemptive spatiotemporal flight conduit refers to an intention expression that not only includes the spatial range of the future flight path but also clearly defines its occupied time interval, and its spatial-temporal boundaries have the ability to expand due to priority. This conduit not only informs neighboring aircraft of the four-dimensional spatial-temporal region it will occupy but also implies that, when necessary, the boundaries of high-priority conduits cannot be easily violated, thus possessing a preemptive attribute. Intention information refers to the digital information packet broadcast by the UAV indicating its future movement plan. It can be understood that this information packet carries at least the coordinate parameters, time parameters, and corresponding negotiation priority weights of the aforementioned spatiotemporal flight conduit. Communication range refers to the spatial distance or signal coverage area where the UAV's onboard communication equipment can effectively and reliably transmit and receive data.

[0045] In S4, the planned mission path refers to the expected flight path from the current position to the target position, pre-generated or generated online by the flight planning algorithm to achieve the mission objective. Motion state is a continuation of the real-time motion state information from S1 in the prediction model, used to infer the UAV's position over a future period. Predicting the spatial region and time window with potential conflicts refers to using the intention information of neighboring UAVs and the UAV's pre-planned path to infer the three-dimensional spatial range in which the safe distance between the two UAVs may fall below a preset protection threshold, as well as the corresponding start and end times of this situation.

[0046] In S5, the negotiation priority weight difference refers to the absolute difference in priority weight values ​​between the local machine and its conflicting neighbor, quantifying the differentiation of priorities between the two sides. The urgency of the time window refers to the time margin from the current moment to the predicted time of conflict, or the duration of the conflict window itself; generally, the closer to the conflict, the narrower the window and the higher the urgency. Negotiable conflict refers to a type of conflict that can be resolved through information exchange and mutual concessions when both time and priority conditions meet preset criteria. The distributed negotiation protocol is a set of preset communication and decision-making rules that allow conflicting parties to resolve conflicts without central node scheduling, simply through direct or forwarded information exchange.

[0047] In S6, adjusting one's own spatiotemporal flight path refers to modifying the spatial area and / or time interval originally planned to be occupied. This modification can be spatial contraction or translation, or temporal delay or advancement. Verifying whether the adjusted intent information eliminates conflicts involves a high-priority UAV executing a conflict detection algorithm similar to S4 to check whether the new path adjusted by the low-priority UAV has completely moved out of the safe distance from its own path. A conflict-free cooperative flight trajectory refers to the sequence of flight paths ultimately generated by the conflicting UAVs after iterative negotiation, which enables safe and orderly passage between different priorities and maintains intervals greater than a safe threshold.

[0048] In S7, flight deviation refers to the vector difference between the UAV's position, speed, and other motion states during actual flight and the expected motion states at the corresponding timestamps on the conflict-free cooperative flight trajectory generated by S6. The preset tolerance is a pre-defined deviation allowance threshold, which may include upper limits for position error and speed error. When the deviation is within the tolerance range, it is considered to be executing according to plan; exceeding the tolerance indicates that the cooperative state has been broken, constituting a condition for triggering the renegotiation process.

[0049] The following analysis explains how the above seven steps work together to achieve technical results. The entire process forms a complete closed loop from perception and assessment to collaborative decision-making, and then to execution and correction.

[0050] First, S1 and S2 input dynamic mission times and real-time flight status into the system, with their core output being negotiated priority weights with quantitative standards. This process ensures that each drone, before facing potential conflicts, has the ability to assess the urgency of its own mission, providing a fundamental basis for all subsequent differentiated handling. Without this step, all drones would be treated equally, making it impossible to reserve priority passage for urgent missions.

[0051] Next, S3 combines the priority weights output by S2 with the motion state of S1 to generate a spatiotemporal flight channel with preemption attributes and broadcasts it. This step physicalizes the priority values ​​into specific space-time occupancy boundaries. High-priority tasks not only know they are more urgent, but also announce this intention of priority reservation to other drones in the cluster by occupying a wider spatiotemporal channel. This announcement provides high-fidelity input for S4's neighboring drones to perform conflict prediction.

[0052] Subsequently, based on the intent information received from S3 and its own planned path, S4 accurately predicts the specific conflict area and time window. The contribution of this step is that it enables both parties to reach a consensus on "where and when danger will occur," quantifying avoidance from a vague directional issue into a clearly defined geometric and temporal issue, thus creating conditions for precise negotiation.

[0053] Entering S5, the priority weight difference provided by S2 and the conflict time window information provided by S4 are used to determine the type of conflict. This is a crucial strategy filter that focuses on conflicts with room for negotiation; for extremely urgent situations or those with significant priority differences, it bypasses complex iterative processes for rapid response. This differentiated processing mechanism avoids wasting computational and communication resources on unnecessary negotiations, improving overall response efficiency.

[0054] When a conflict is determined to be negotiable, S6 follows the decision made in S5 and initiates an ordered iterative concession mechanism. The lower-priority party, based on the conflict geometry calculated in S4, actively adjusts the pipeline generated by the principle of S3, while the higher-priority party only needs to verify the adjustments. This unidirectional adjustment plus unidirectional verification mode ensures the directionality of the decision and the speed of convergence. Each adjustment is verified by the higher-priority party, forming an inner loop that progressively approaches a safe state until a conflict-free trajectory that both parties can safely execute is generated.

[0055] Finally, S7 executes the cooperative trajectory generated by S6 and monitors deviations using the real-time status obtained in S1. This monitoring mechanism is crucial for handling environmental disturbances and individual execution errors. Once a deviation exceeding the preset tolerance is detected between the actual flight trajectory and the cooperative plan, it immediately activates the S2 loop, pulling the system back into the closed loop of evaluation, broadcasting, prediction, and negotiation for renegotiation. This allows the entire method to avoid relying on a perfect one-time plan and instead possess the ability to continuously adapt to and correct dynamic uncertainties. Overall, these seven steps, from front-end evaluation and mid-stage game-theoretic decision-making to back-end execution monitoring, are interconnected, smoothly and orderly transmitting and dissipating the dynamic urgency pressure of the task level throughout the cooperative flight of the entire cluster.

[0056] In one or more embodiments of this application, step S2 determines the mission urgency of the UAV based on the mission time limit parameter and the current time, and converts it into a negotiated priority weight, specifically including: S21. Obtain the task value weight corresponding to the task time limit parameter, and calculate the remaining execution time from the current time to the task deadline. S22. Use the ratio of task value weight to remaining execution time as the basic task urgency. S23. Using a preset monotonic nonlinear mapping function, the urgency of the basic task is converted into a normalized negotiation priority weight. The mapping function makes the rate of change of the negotiation priority weight greater when the remaining execution time is shorter.

[0057] In this embodiment, in step S21, the task value weight is a pre-assigned or dynamically evaluated value used to distinguish the importance of different tasks. Even if two tasks have similar deadlines, the loss or impact of the higher-value task is usually greater. For example, the weight of an emergency supplies delivery task can be set higher than that of a routine inspection task. This weight can be determined based on task type, payload value, or user input; it can be a fixed scalar or a variable that changes with the environment. The remaining execution time refers to the time from the current time to the task's deadline, which may be equal to the planned latest completion time minus the current time. It is understood that when the drone deviates from the predetermined plan due to avoidance or other reasons, the remaining execution time will be shortened accordingly, thus dynamically reflecting the margin for task delay.

[0058] In S22, the basic task urgency is an intermediate variable that integrates task value and remaining time. By using the task value weight as the numerator and the remaining execution time as the denominator to form a ratio, this indicator simultaneously reflects the relative importance of the task and time pressure. For example, a high-value task nearing its deadline will have a significantly higher basic task urgency; while a low-value task with ample time will have a relatively lower value. When the remaining execution time approaches zero, the basic task urgency increases dramatically, creating a superlinear growth in the sense of urgency.

[0059] In S23, the preset monotonic nonlinear mapping function is a pre-designed mathematical transformation used to convert the basic task urgency into a value within a standard range, i.e., the normalized negotiation priority weight. Monotonicity ensures that the greater the input's basic urgency, the greater the output's priority weight, without reversing the magnitude relationship. Nonlinearity means that the relationship between input and output is not a simple direct proportionality. It is particularly noteworthy that the mapping function makes the rate of change of the negotiation priority weight greater as the remaining execution time shortens. This means that in the later stages of task execution, even if the remaining time decreases by only a small amount, the priority weight will show a more significant increase than in the earlier stages. For example, this function can be implemented using an exponential function, a power function, or a piecewise linear function. The normalized negotiation priority weight can be a number between 0 and 1, facilitating standardized comparisons between different drones.

[0060] S21, S22, and S23 together form a complete quantitative transformation chain from the original attributes of the task to the standardized negotiation priority. Through progressive calculation, the three gradually realize a refined expression of the true urgency of the task.

[0061] S21 obtains task value weights from the upper-level task management system and calculates the physical world's time urgency—remaining execution time—in real time. The former reflects the task's importance in a static dimension, while the latter captures the task's pressure in a dynamic time dimension. These two parameters describe the task's characteristics from different perspectives, providing the basic material for subsequent urgency assessment. Without either one, urgency evaluation will become one-sided; for example, considering only value will lead to permanently high priority without time constraints, while considering only time will fail to distinguish the relative importance of tasks due in the same batch.

[0062] S22 combines the two seemingly unrelated dimensional parameters into a single scalar quantity—the basic task urgency—through a ratio operation. The core contribution of this step lies in constructing a unified and comparable urgency metric, allowing the combined effect of value and time to be simultaneously reflected in a single numerical value. The ratio form inherently possesses a gain effect: with the same value weight, the shorter the remaining time, the higher the urgency; with the same remaining time, the higher the value of the task, the more pronounced the urgency. This choice of mathematical form allows urgency to naturally respond to changes in task status.

[0063] S23 establishes a non-linear mapping between the basic task urgency and the negotiated priority weights of the final output. Relying solely on the linear ratio of S22 would result in equal changes in urgency in the early and late stages of task execution due to the same amount of time elapsed, which may not meet the real-world decision-making requirements where the closer to the deadline, the more sensitive the decision-making becomes. By introducing the non-linear characteristic that the shorter the remaining execution time, the greater the rate of weight change, S23 allows the priority weights to exhibit an accelerated response to the passage of time in the later stages of the task. This design helps in multi-drone conflict scenarios where, when the remaining time for a task is extremely limited, its negotiation weight is quickly pushed to a higher level, thus gaining stronger pipeline expansion capabilities and less concession obligations in the upcoming negotiation process. Simultaneously, the normalized output maps the urgency of different types and scales of tasks to a standard range, making the weights of various drones directly comparable and computationally consistent. This provides standardized input for subsequent steps such as comparing weight differences with thresholds and calculating pipeline expansion magnitudes. Overall, these three sub-steps, from acquiring raw information to dimensional synthesis, and then to standardized expression and sensitivity optimization, are progressively advanced, ensuring that the final negotiated priority weights not only faithfully reflect the multi-dimensional urgency of the task, but also adapt to the actual needs of distributed negotiation for numerical sensitivity and comparability.

[0064] In one or more embodiments of this application, S3 generates a spatiotemporal flight pipeline with preemptive characteristics as intent information based on real-time motion state information and negotiation priority weights, specifically including: S31. Based on real-time motion status and planned mission path, generate a nominal safe flight corridor for the aircraft within a future preset time domain, defined by position boundaries and speed range; S32. Based on the value of the negotiation priority weight, the expansion margin is superimposed on the spatial dimension of the nominal safe flight corridor, and the available time period is extended in the time dimension. The expansion range increases with the increase of the negotiation priority weight, forming a spatiotemporal flight channel with preemption characteristics. S33. Write priority weight flags into the broadcast message of the spacetime flight pipeline and encapsulate them as intent information.

[0065] In this embodiment, the future preset time domain in S31 refers to a predefined time interval extending into the future from the current moment. Its length can be dynamically set according to the mission stage or flight speed, for example, a range of 10 to 30 seconds into the future. The position boundary defines the permissible range of movement of the UAV in space, which can be the outline of a three-dimensional volume, such as a channel composed of a series of continuous spatial prisms or ellipsoids. The speed range specifies the upper and lower limits of the speed magnitude and direction allowed for the UAV within this corridor. The area jointly defined by the position boundary and the speed range constitutes the nominal safe flight corridor, which is the basic flight space and time occupancy when priority preemption is not considered and only the mission trajectory and safety margin are satisfied.

[0066] In S32, the expansion margin refers to the additional radial or axial space added outward from the nominal safe flight corridor's spatial boundary, such as increasing the original lateral cross-sectional radius by an additional amount. Extending the available time period refers to extending the time occupancy interval of the flight corridor before the start point or after the end point, allowing high-priority drones to occupy a certain airspace earlier or delay their departure. The expansion magnitude increases with the negotiation priority weight. This correlation allows drones with higher weights to obtain a more spacious spatiotemporal occupancy area, thus forming a spatiotemporal flight corridor with preemptive characteristics. This characteristic manifests as high-weight corridors having stronger exclusivity in both space and time.

[0067] In S33, broadcast messages are wireless data packets carrying intent information, which may include data such as the geometric parameters, timing parameters, and priority weights of the spatiotemporal flight path; the priority weight flag is a data field in the message used to identify the priority weights negotiated by the sender, and the receiver can directly parse this field to obtain priority information.

[0068] S31, S32, and S33 gradually transform abstract task priorities into physically perceptible spatiotemporal occupancy intentions and complete the information transmission from individual planning to group broadcasting, forming a tight functional closed loop among the three steps.

[0069] First, the nominal safe flight corridor generated by S31, following the drone's own mission trajectory and mobility, constructs a basic spatiotemporal channel purely serving the drone's mission. This channel is entirely task-oriented and does not currently consider priority differences between clusters, essentially providing each drone with a standard, comparable spatiotemporal requirement expression. It forms the foundation for subsequent differentiated processing, providing a unified benchmark container for priority expansion.

[0070] Building upon this, S32 utilizes the negotiated priority weights derived from the preceding steps to asymmetricly expand the nominal safe flight corridor. The weight values ​​directly affect the expansion margin and available time slots, materializing priority information as additional spatial and temporal occupancy. This expansion is not a simple proportional enlargement, but rather a non-linear increase with increasing weight; drones with higher weights gain more margin in the corridor's outline. Thus, the originally equal baseline corridor for all drones is differentiated and assigned preemptive capabilities, allowing high-priority drones to occupy a more dominant position in the spatiotemporal landscape even before broadcasting. This step transforms priority from numerical symbols into a spatial and temporal priority allocation pattern, representing a crucial leap in the expression of intent.

[0071] Finally, S33 encapsulates this preemptively acquired spatiotemporal flight pipeline, along with its priority weight marker, into a broadcast message and sends it out, completing the transformation of individual intent into shared collective knowledge. The priority weight marker is transmitted with the message, allowing the receiver to directly obtain the preemption level of the pipeline without additional negotiation or handshake, thus understanding the strength of the other party's intent. In this way, the spatiotemporal needs of high-priority UAVs and their tolerance for conflict are proactively and transparently transmitted to neighboring UAVs that may interact, enabling conflict prediction and avoidance to occur at an earlier planning stage, rather than reacting passively as conflict approaches.

[0072] Overall, these three sub-steps—from baseline generation to differentiated expansion and broadcast transmission—integrate the logic of task priority throughout the entire intent expression channel. In this way, task urgency is ultimately transformed into spatiotemporal constraints that cluster members can directly perceive and respond to, achieving an asymmetric, ordered partition of the shared space without requiring additional negotiation communication. This mechanism helps reduce the probability of conflict and ensures that when necessary negotiation occurs, all parties are already making decisions based on an asymmetric situation, thereby reducing the frequency and magnitude of repeated concessions and adjustments.

[0073] In one or more embodiments of this application, in step S4, the intent information broadcast by a neighboring UAV is received, and combined with the local planned mission path and motion state, the spatial regions and time windows with potential conflicts are predicted, specifically including: S41. Parse the received intent information and construct the occupancy grid set of the corresponding neighboring UAV in the time-space dimension. The occupancy grid set includes a three-dimensional spatial grid index and the corresponding occupancy time slot. S42. Discretize the local planned task path into a sequence of track points carrying timestamps according to a preset time step; S43. Perform spatiotemporal intersection detection on the track point sequence and the occupied grid set, and take the overlapping spatial grids and their covered time intervals as potential conflict spatial regions and time windows.

[0074] In this embodiment, the occupancy grid set in S41 is a data structure used to quantize and map the spatiotemporal flight pipelines expressed by the intent information of neighboring UAVs onto discrete spatial grids and time slots. The three-dimensional spatial grid index is a numbered spatial unit after dividing the conflict-detection-related airspace. For example, the overall airspace is divided into several cubic grid units according to latitude, longitude, and altitude, each unit having a unique number. The occupancy time slot represents the specific time period during which the corresponding spatial grid is covered by the flight pipelines of neighboring UAVs. This can be a continuous start-end time pair or a set of discrete time slices. By parsing the spatiotemporal flight pipeline parameters carried in each intent message, it is possible to determine which spatial grids the pipeline covers and mark each grid with a corresponding time occupancy marker, thus forming a spatiotemporal occupancy map of the neighboring UAV group in future time periods.

[0075] The preset time step in S42 refers to the time interval used when discretizing continuous mission paths. Its value can be fixed according to the planning accuracy requirements, such as 0.1 seconds or 0.5 seconds, or it can be dynamically adjusted according to the speed of the UAV. The track point sequence with timestamps is a series of spatial location points sampled along the planned mission path of the UAV according to the above time step. Each spatial location point is bound to a future time value, which together describe when the UAV is expected to arrive at what location.

[0076] The spatiotemporal intersection detection in S43 involves traversing the sequence of local waypoints, querying whether the spatial grid containing each waypoint has been marked by the occupied grid set of neighboring UAVs, and whether the timestamp of the waypoint falls within the occupied time slot of that grid. Overlapping spatial grids refer to the grid containing the local waypoint being the same as the grid occupied by a neighboring UAV; the covered time interval refers to the specific time range within which the local timestamp and the occupied time slot of a neighboring UAV intersect on such overlapping grids. Through this detection, potential conflicts can be captured simultaneously in both spatiotemporal dimensions, locating them as specific spatial regions and their corresponding time windows.

[0077] The three sub-steps S41, S42, and S43 together constitute a complete processing link from multi-source intent information to specific conflict prediction. Through unified data structure and spatiotemporal alignment, the three steps achieve accurate quantification of conflicts.

[0078] S41 transforms the continuous spatiotemporal flight pipelines broadcast by neighboring UAVs into a discrete set of occupancy grids. This effectively unifies flight intentions from multiple sources, with different shapes and time spans, under a single spatial grid reference system. The advantage of this unified processing is that regardless of the geometry and temporal distribution of the neighboring UAVs' flight pipelines, they are ultimately standardized and expressed as whether or not the spatial grid is occupied, and the duration of occupation. This eliminates the need for subsequent conflict detection to perform geometric intersections for the complex parameters of each pipeline individually; instead, it only requires querying the grid's state.

[0079] On the local side, S42 discretizes the planned task path into a sequence of timestamped waypoints using the same time step. The planned path, originally a smooth trajectory in the spatial dimension, is discretized in time to create a one-to-one correspondence between space and time, ensuring that each waypoint has both spatial and temporal coordinates, aligning with the grid set constructed by S41 in the spatiotemporal dimension. The preset time step can be adjusted between detection accuracy and computational cost; a smaller step size results in denser waypoints and more detailed collision detection.

[0080] Building upon the two steps above, S43 performs direct and efficient spatiotemporal intersection detection. It uses the timestamped trackpoint sequence generated in S42 to poll the occupied grid set constructed in S41. Based on whether the grid containing the trackpoint is occupied and whether the corresponding time slot covers the trackpoint's timestamp, it determines whether a conflict exists and extracts overlapping spatial grids and covered time intervals. This method of transforming continuous spatiotemporal conflict problems into discrete grid queries maintains high detection accuracy while reducing computational complexity and avoiding real-time intersection calculations on complex geometries. Furthermore, the detection results not only provide a binary judgment of conflict-free or conflict-prone, but also output specific spatial regions and time windows, providing accurate and quantifiable input for subsequent steps such as determining conflict types, calculating negotiation adjustment step sizes, and applying flight path locking. These three sub-steps, from neighborhood information abstraction and local path discretization to spatiotemporal intersection matching, progressively complete conflict prediction, enabling the early and accurate identification of potential interactions between multiple aircraft.

[0081] In one or more embodiments of this application, step S5 utilizes the difference in negotiation priority weights between the local drone and neighboring drones, as well as the urgency of the time window, to determine whether a conflict is a negotiable conflict. Specifically, this includes: S51. Calculate the negotiation priority weight difference between the local machine and the neighboring UAVs that caused the conflict; S52, Get the duration of the time window; S53. When the difference in negotiation priority weights is less than the preset negotiation difference threshold and the duration is greater than the minimum negotiation window threshold, the conflict is determined to be a negotiable conflict; otherwise, it is determined to be an emergency conflict. S54. When an emergency conflict is identified, an emergency avoidance flag is generated and the distributed negotiation protocol is suspended.

[0082] In this embodiment, the negotiation priority weight difference in S51 refers to the absolute difference between the negotiation priority weights of the local machine and the specific conflicting neighboring UAV, reflecting the magnitude of the gap in task urgency between the two parties. This difference can be obtained by subtracting the negotiation priority weight previously calculated by the local machine from the priority weight flag parsed from the intent information message of the neighboring UAV and taking the absolute value. It can be understood that a smaller difference indicates that the task urgency of the two machines is relatively close; a larger difference indicates that one party has a significantly higher urgency than the other, possessing a stronger dominant basis in the negotiation.

[0083] The duration of the time window in S52 refers to the time span between the start and end times of the conflict time window in the S4 prediction output. It represents the available reaction time margin from the start to the end of the conflict. The duration is directly related to the urgency of the conflict in the time dimension. The shorter the time window, the smaller the operational margin for negotiation iteration, and the higher the requirement for rapid decision-making in the system.

[0084] The preset negotiation difference threshold in S53 is a pre-defined weight difference limit value used to distinguish whether the priorities of the conflicting parties can be negotiated gradually or whether the gap is so large that forced avoidance is necessary. The minimum negotiation window threshold is a pre-defined time span limit value used to determine whether the conflict time window is sufficient to support a complete iterative negotiation process. Only when the weight difference does not exceed the negotiation difference threshold and the duration of the time window is greater than the minimum negotiation window threshold, it indicates that the two parties are neither severely unequal and have enough time for information exchange and pipeline adjustment. In this case, the conflict is determined to be a negotiable conflict. Conversely, if the weight difference is equal to or exceeds the negotiation difference threshold, or the duration is equal to or shorter than the minimum negotiation window threshold, it means that either one party has an overwhelming advantage in priority, or the conflict is imminent and cannot wait for negotiation to converge. In this case, the conflict is determined to be an urgent conflict.

[0085] The emergency avoidance flag in S54 is a data identifier used in the system state machine to indicate that the current conflict has entered an emergency handling branch, requiring a pause in the standard distributed negotiation protocol. Pausing the distributed negotiation protocol means that in this conflict resolution, the iterative adjustment and verification process described in S6 will not be entered, but instead, the process will jump to other predetermined emergency response mechanisms for handling.

[0086] The four sub-steps S51 to S54 establish a clear dividing line between negotiable and urgent conflicts through a quantitative two-dimensional evaluation logic, and configure different follow-up processing channels for different categories.

[0087] S51 and S52 each extract two key decision variables from the output of the previous steps: one is the priority weight difference representing the disparity in the status of the parties, and the other is the duration of the time window representing the urgency of the event. These two variables independently describe the negotiable conditions of the conflict from one perspective, but neither can provide a comprehensive assessment in isolation. If only the weight difference is considered, ignoring whether the time window is sufficient to complete the negotiation, negotiation might be initiated even in a time-critical scenario, leading to conflict before the negotiation is completed. If only the time window is considered, ignoring the significant disparity in the status of the parties, a high-priority task might be forced into an unnecessary and lengthy negotiation with a low-priority task.

[0088] S53 combines the two variables mentioned above with two preset threshold conditions for judgment. The preset negotiation difference threshold and minimum negotiation window threshold serve as the decision boundary. For conflicts with small weight differences and ample time windows, it means that both parties have a relatively balanced negotiation basis and sufficient adjustment time. Such conflicts are suitable for using a complete distributed negotiation mechanism, where the low-priority party finds a coordination path acceptable to both parties through gradual concessions and verification by the high-priority party. For conflicts with excessively large weight differences or excessively short time windows, requiring both parties to negotiate is neither realistic nor necessary. The conflict is directly escalated to emergency conflict handling, switching to a command-style rapid avoidance mechanism. The high-priority party maintains the original channel, while the low-priority party immediately executes the maximum available maneuver.

[0089] S54 establishes a clear signal exit for emergency conflicts, generating an emergency avoidance flag and suspending the distributed negotiation protocol, logically switching the system state to the emergency handling branch. This flag does not affect the simultaneous negotiation and judgment of other potential conflicts; it only terminates the negotiation process for the current specific conflict pair. Through this classification and routing, computing resources and communication bandwidth are concentrated on conflicts that truly have a chance to achieve a win-win situation through negotiation, while conflicts with insufficient time and leeway no longer consume unnecessary negotiation iterations, contributing to more stable avoidance performance in complex multi-machine dynamic environments. Overall, these four sub-steps together construct a strategy fork based on current situation assessment, allowing the system to adaptively select the most appropriate processing mode according to the specific characteristics of the conflict.

[0090] In one or more embodiments of this application, when an urgent conflict is determined, the method further includes: S5a: Trigger a priority-based emergency avoidance protocol, commanding the drone with the lower priority weight to perform the maximum available maneuver to avoid the obstacle under its own constraints; S5b: UAVs with higher negotiation priority weights maintain their spatiotemporal flight pipeline unchanged within the time window until the time window ends.

[0091] In this embodiment, the priority-based emergency avoidance protocol in S5a is a set of pre-defined, direct-instruction avoidance rules that do not rely on iterative negotiation between the two parties. Its triggering condition is after the conflict is determined to be an emergency conflict in S53 and an emergency avoidance marker has been generated in S54. The core feature of this protocol is its one-way command attribute, meaning that the higher-priority party does not need to make any trajectory adjustments, and the avoidance obligation is entirely borne independently by the lower-priority party. Self-constraints refer to the physical and mission boundaries that the lower-priority UAV must adhere to when performing evasive maneuvers. These may include maximum flight speed, minimum turning radius, maximum climb or descent rate, remaining energy limits, and the requirement not to trigger new conflicts with third parties due to avoidance. It is understood that these constraints are determined in real-time based on the UAV's own flight envelope and current state, and are not pre-fixed parameters. Maximum available maneuver refers to the extreme flight maneuver that the UAV can perform to quickly change the conflict situation, provided that the above self-constraints are met. Examples include turning the heading at the maximum available angular rate or changing altitude at the maximum climb rate.

[0092] In S5b, the drone with the higher negotiation priority weight maintains its spatiotemporal flight pipeline unchanged. This means that the drone continues to fly according to the pipeline parameters generated in S3, which have preemptive expansion characteristics, and does not make any concession adjustments to its spatial boundaries, time intervals, or velocity profiles caused by this conflict. "Until the end of the time window" refers to the time range during which the high-priority drone maintains this unchanged state, starting at the beginning of the conflict time window and ending at its end. During this period, the drone does not respond to any adjustment requests from the lower-priority party.

[0093] The S5a and S5b steps work together in emergency conflict scenarios to form a clear-cut, rapid-response asymmetric avoidance mechanism, which is significantly complementary to the iterative negotiation path for negotiable conflicts in S5.

[0094] When a conflict is determined to be an urgent conflict by preliminary steps, it often signifies one of two typical situations: either one party's task is far more urgent than the other's, or the conflict is imminent with almost no time to spare. In this situation, continuing to use the negotiation model of information exchange, gradual adjustment, and iterative verification between the two parties may lead to two unfavorable outcomes: first, the conflict may actually occur before the negotiation has converged; second, the lower-priority party may still attempt minor adjustments as the conflict approaches, while the higher-priority party, fearing potential misjudgment, is also forced to take preparatory maneuvers, causing fluctuations in the trajectories of both parties. S5a and S5b address these issues by providing a simple and predictable emergency handling method through unidirectional avoidance obligations, maximizing the execution scale, and fixing high-priority behaviors.

[0095] S5a commands low-priority drones to perform maximum available maneuvers for evasive maneuvers under constraints. This arrangement means that the magnitude of the evasive maneuver is no longer limited by the negotiated step size and concession limit, but directly takes the maximum physically permissible value. This aggressive maneuver can create a safe distance between the two sides in the shortest possible time, addressing the challenges posed by a short conflict window. At the same time, the introduction of self-constraints ensures that the maneuver does not come at the expense of flight safety or trigger secondary conflicts, thus protecting the passage path of the high-priority drone while limiting the risk to the low-priority drone to an acceptable level.

[0096] S5b requires high-priority drones to maintain their trajectory unchanged within the time window. This fixed strategy eliminates the possibility of any responsive adjustments by the high-priority drone. For high-urgency tasks, a stable trajectory means that it can continue to focus on executing the time-sensitive task without expending attention resources to determine whether the adjustments made by lower-priority drones are appropriate or will cause new conflicts, nor worry about additional conflicts with a third drone due to its own adjustments. Throughout the time window, the high-priority drone essentially receives a commitment to a stable flight environment.

[0097] The coordination between the two steps forms a unidirectional command execution chain. Commands issued by S5a do not require feedback confirmation from S5b, and S5b's inaction does not require continuous coordination from S5a. This loosely coupled yet explicit relationship allows decisions to be made in a very short time and executed in parallel by both parties, without experiencing the latency of round-trip communication. Compared to the negotiable conflict path, the emergency avoidance branch sacrifices some optimality and fairness, but gains processing speed and stability of high-priority trajectories, making it suitable for scenarios with narrow time windows or significant priority differences. Overall, S5a and S5b provide a means of handling emergency conflicts that maintains functional continuity under extreme constraints. Logically, this is closely linked to the aforementioned negotiable conflict judgment steps, enabling the conflict response strategy to have the ability to classify and adapt to different situations.

[0098] In one or more embodiments of this application, in S6, the drone with the lower negotiation priority weight adjusts its own spatiotemporal flight path and sends the adjusted intent information, and the drone with the higher negotiation priority weight verifies whether the adjusted intent information eliminates the conflict, specifically including: S61. For UAVs with lower negotiation priority weights, the amount of spatiotemporal flight pipeline space contraction or time offset required to eliminate the conflict is calculated based on the conflict geometry. S62. Modify the spatial boundary parameters or time interval parameters of the space-time flight pipeline step by step according to the preset adjustment step size. After each step of modification, keep the pipeline continuous and do not cause new conflicts with its own mission path. S63. After each modification is completed, the updated intent information is broadcast, and the drone with the higher priority weight in the negotiation performs conflict detection and verification. S64. If the conflict disappears, the current parameters are fixed as part of the cooperative trajectory. If the conflict does not disappear and the cumulative adjustment steps have not reached the preset concession limit, return to S62 and continue to modify according to the adjustment step size. If the conflict still exists after reaching the preset concession limit, pause the adjustment and send a concession limit notification to the high-priority weight drone.

[0099] In this embodiment, the conflict geometry in S61 refers to the relative position and overlap pattern between the spatiotemporal flight paths of the local machine and neighboring UAVs within the potential conflict space region predicted by S4. This can include geometric quantities such as the minimum spatial distance, overlap volume, or intrusion depth of the two paths. Spatial contraction refers to the magnitude required for the lower-priority party to narrow or offset its spatiotemporal flight path in the spatial dimension, such as reducing the cross-sectional radius of the path by a certain value or shifting the path's centerline away from the conflict zone by a certain distance. Temporal offset refers to the duration by which the overall occupation period of the path is delayed or advanced, so that the time of passage of both parties through the same spatial grid is staggered. It is understood that eliminating the same conflict can be achieved by pure spatial contraction, pure temporal offset, or a combination of both, and the specific choice can be determined based on the characteristics and adjustment costs of the conflict geometry.

[0100] In S62, the preset adjustment step size is a pre-defined single modification magnitude value. This breaks down the total contraction or offset into multiple small operations, each step changing only a limited spatial dimension or time duration to avoid new conflicts or unpredictable chain reactions caused by a single large adjustment. Step-by-step modification of spatial boundary parameters or time interval parameters means that each adjustment step only changes the geometric boundaries or time start and end points of the spacetime flight pipeline by a small step. Maintaining pipeline continuity means that the adjusted pipeline must still be connected end-to-end in time without any breaks; avoiding new conflicts with its own mission path requires that the modified pipeline can still completely accommodate its planned mission path, without sacrificing the feasibility of its own mission to avoid other aircraft.

[0101] In S63, broadcasting updated intent information means that after each adjustment step, the low-priority drone re-encapsulates the modified spatiotemporal flight pipeline parameters into an intent information message and broadcasts it to neighboring drones within the communication range via the wireless link. The high-priority drone performs conflict detection and verification. After receiving the updated intent information, the high-priority drone runs a conflict detection process similar to S4 again to determine whether the modified pipeline no longer has a spatiotemporal intersection with its own pipeline.

[0102] In S64, fixing the current parameters as part of the cooperative trajectory signifies that both parties agree on the newly generated spatiotemporal flight path and will execute it as a component of the formal flight trajectory. The preset concession limit is a pre-defined maximum cumulative adjustment step count, reflecting the maximum acceptable concession level for the lower-priority party. When the cumulative adjustment steps reach this limit and the conflict remains unresolved, it indicates that negotiation cannot converge within the current concession space. At this point, the lower-priority party issues a concession limit notification, informing the higher-priority party that it cannot continue adjusting, thus signaling the need for alternative measures.

[0103] The four sub-steps S61 to S64 together construct a controlled iterative one-way concession negotiation mechanism. Through the progressive cooperation of four stages—quantitative calculation, step adjustment, successive verification, and boundary protection—convergence-oriented and clearly defined conflict resolution is achieved.

[0104] First, S61 uses conflict geometry to transform ambiguous avoidance requirements into concrete adjustment quantities. This step quantifies the starting point of conflict resolution from a directional intention into an actionable numerical objective. Spatial contraction and temporal offset can be directly correlated with modifications to pipeline parameters, providing clear direction and scale for subsequent adjustment operations. This quantification method ensures that adjustments are not blind trial and error, but rather targeted corrections based on the geometric characteristics of the conflicting entities.

[0105] Next, S62 introduced a dual mechanism of process control and self-protection by setting adjustment step sizes and implementing step-by-step modifications. Step-by-step modifications replace drastic, one-step changes; each step only produces minor pipeline changes. This helps maintain the situational awareness stability of other UAVs near the conflict space and reduces the possibility of new conflicts arising from unilateral pipeline abrupt changes. Simultaneously, after each modification, pipeline continuity and its own mission path constraints are checked. This ensures that lower-priority parties do not sacrifice their own flight physical feasibility to meet the needs of higher-priority parties during concessions; concessions are conditional and have boundaries.

[0106] S63 immediately broadcasts information and verifies the changes after each modification, forming an asynchronous collaborative closed loop where low-priority parties adjust and high-priority parties verify. This design eliminates the need for complex three-way handshakes or centralized arbitration, simplifying the process and enabling distributed execution. High-priority parties only need to perform conflict detection and verification, without needing to analyze the intentions of low-priority parties or calculate their own adjustments, thus reducing their burden in the negotiation and improving decision-making speed. Step-by-step verification also ensures that the system can immediately stop adjusting once the conflict is resolved, avoiding excessive concessions.

[0107] S64 defines two termination exits and one exception exit for the iterative process. When the conflict disappears, the convergence exit solidifies the current parameters, and the negotiation ends successfully. The concession limit condition serves as the exception exit, preventing low-priority parties from getting bogged down in endless adjustments or even conceding to the point of affecting their own task completion if the other party consistently fails to meet it. The concession limit notification explicitly transmits the state of unconverged negotiation to the high-priority party, informing it of the low-priority party's predicament and providing a trigger signal for higher-level task replanning or priority reallocation at the scheduling level. This concession mechanism with boundary protection ensures the efficiency of high-priority tasks while maintaining the bottom-line space for low-priority tasks, helping to improve the overall operational coordination of the cluster and the task completion rate of each machine in dynamic scenarios. The four sub-steps, from target quantification to step-by-step execution, from successive verification to boundary convergence, form a complete, self-limiting distributed negotiation closed loop.

[0108] In one or more embodiments of this application, before S5 triggers the distributed negotiation protocol, the following is also included: S50. Detect the quality of the current communication link with neighboring drones; if the quality of the communication link is lower than the preset reliability threshold, determine the drone with the highest weight value from the drones that are accessible to the communication as a temporary coordination node according to the negotiation priority weight. The temporary coordination node collects the intention information of each drone, generates a safe trajectory using a centralized conflict resolution algorithm, and broadcasts it to each drone.

[0109] In this embodiment, the communication link quality in S50 refers to the reliability of the wireless communication channel between UAVs, which can be measured using conventional wireless channel evaluation metrics in the art, such as Received Signal Strength Indication (RSSI), packet error rate, or signal-to-noise ratio. It is understood that the communication link quality detection can be continuous, with a current value query performed each time before the distributed negotiation protocol is triggered, or it can be continuously collected at a preset period and the recent average value taken. The preset reliability threshold is a pre-set communication quality threshold. When the communication link quality is higher than or equal to this threshold, the link is considered to have the information exchange capability required to support the distributed negotiation protocol; if it is lower than this threshold, it means that messages may be lost, the delay is too large, or the bandwidth is insufficient to support the reliable transmission of negotiation messages. A temporary coordination node refers to a UAV designated to undertake conflict resolution calculation tasks during the communication-restricted period. Its role is not fixed and only takes effect during the current conflict resolution period; the role is released after the conflict resolution is completed. Determining the drone with the highest weight among those within communication reach refers to selecting the drone with the highest negotiation priority weight from among those neighboring drones that can still maintain effective communication with the local drone under the current communication link quality conditions, as a temporary coordination node. If there are multiple drones with equally high weights within communication reach, a selection can be made according to preset decision rules, such as selecting the one with the smallest identifier value to break a tie. A centralized conflict resolution algorithm refers to an algorithm where a single computing node collects the intent information of all relevant drones and performs global optimization or heuristic solution within a unified problem space to generate multiple conflict-free trajectories at once. For example, trajectory optimization methods based on mixed-integer linear programming or sequential convex programming can be used. The generated safe trajectories are sets of motion trajectories or flight path parameters that can be distributed and directly executed by each drone.

[0110] S50, as a pre-detection and mode switching step, inserts a communication capability assessment step before the distributed negotiation process starts, enabling the conflict avoidance system to adaptively switch to the backup mode when communication conditions deteriorate.

[0111] When multiple UAVs are in a good communication environment, the system operates according to the distributed negotiation protocol described in S5 and S6. Each UAV exchanges intent information point-to-point, with lower-priority parties conceding and higher-priority parties verifying. The negotiation process does not require intervention from a central node, avoiding single-point computation bottlenecks and single-point failure risks. However, the effective operation of distributed negotiation relies on a fundamental premise: that both negotiating parties can stably send and receive multiple rounds of messages, that the intent information adjusted by the conceding party at each step can be received by the verifying party in a timely manner, and that the verification party's detection results can be successfully fed back. Once the communication link quality deteriorates below the reliability threshold, the above assumptions no longer hold, and continuing to force the use of distributed negotiation may lead to negotiation process stagnation, message conflicts, or inconsistent decisions.

[0112] S50 performs conditional checks before entering the negotiation process by real-time monitoring of the communication link quality. If the link quality is good, the process naturally proceeds to the original logic of S5. If the link quality does not meet the requirements, S50 avoids the potentially unreliable distributed negotiation and immediately activates a communication degradation response plan. This plan utilizes drones within the currently communicable range and selects the one with the highest priority as the temporary coordinating node. The selection of the highest-weighted drone for the temporary coordinating role is not due to superior communication conditions, but rather leverages its high task urgency and greater sensitivity to conflict resolution quality, motivating it to complete reasonable trajectory calculations within a short time. Simultaneously, high-weighted drones are often the higher-priority party in a conflict, and having them perform the calculations also helps reduce the number of hops in information transmission to some extent.

[0113] After collecting the intent information from each drone, the temporary coordination node uses a centralized conflict resolution algorithm to uniformly generate a safe trajectory and broadcast it to other drones. This transformation compresses the original distributed iterative process, which required multiple round trips, into a single unidirectional flow of collection, computation, and distribution, significantly reducing the requirements for communication reliability and round-trip latency. Each drone only needs to receive the computation result at the receiving end to execute the algorithm, eliminating the need for back-and-forth verification.

[0114] This step complements, rather than replaces, the subsequent S5 and S6 steps. S50's communication quality check is performed only when negotiation is about to be triggered, meaning that for conflicts with normal communication conditions, the system still prioritizes the lightweight distributed negotiation mechanism; only when local communication deteriorates is a centralized backup mode temporarily activated for a specific conflict pair or airspace. This on-demand switching approach avoids the scalability issues of switching the entire system to a centralized architecture and provides a recoverable path for conflict avoidance under communication disturbances. The temporary coordinating node releases its role after the conflict is resolved, without changing the fundamental attributes of the system's distributed architecture. The entire method achieves a balance between communication robustness and system simplicity.

[0115] In one or more embodiments of this application, before the UAV with the lower negotiation priority weight adjusts its own spatiotemporal flight path in step S6, the following steps are also included: Query the historical concession records between this machine and the high-priority drone involved in the current conflict; if the historical concession records show that the number of concessions made by this machine exceeds the preset value, reduce the upper limit of the adjustment range, or request temporary compensation for the priority weight from the high-priority drone.

[0116] In this embodiment, the historical concession record refers to the cumulative record of concessions made by the UAV to a specific high-priority weight UAV in previous rounds of distributed negotiation. This record may include fields such as the number of concessions, the adjustment range of each concession, and a timestamp. This record can be maintained locally by each UAV. Each time an S6 negotiation is completed and the adjusted pipeline parameters are fixed as a cooperative trajectory, the low-priority party writes the concession information to its local storage. It is understood that the historical concession record can have a valid time window, for example, only counting concession events within a recent period. Expired records can be cleared or have their weight reduced to reflect the dynamic changes in the task progress. The preset value is a pre-set upper limit for the cumulative number of concessions or the cumulative upper limit for the concession range, used to define whether the low-priority party has entered a state of excessive concession. Reducing the upper limit of the current adjustment range means multiplying the preset adjustment step size in S62 by a coefficient less than one, or directly reducing the value of the preset concession upper limit, to limit the maximum concession space that the low-priority party can make in the current conflict. Temporary compensation for negotiation priority weight refers to a request made by a lower-priority party to a higher-priority party during the current conflict resolution process. The request seeks to temporarily increase the lower-priority party's negotiation priority weight, narrowing the gap and thus securing a more favorable negotiating position in future conflicts that may involve both parties. This compensation is temporary, effective only within the current task cycle or a specified period, and does not permanently alter the task's value weight.

[0117] This step inserts a review and adjustment process based on historical fairness before the low-priority party formally enters the S61 calculation of adjustment amounts and begins to make concessions. Its purpose is to incorporate the burden of long-term concessions from multiple conflicts into the handling of the current conflict, so as to avoid a small number of drones being continuously at a disadvantage in continuous conflicts.

[0118] In dynamic swarm operations, a single drone may collide with multiple high-priority drones in a short period. If it is on the lower priority side in every conflict and yields to the full concession limit each time, then after several rounds of negotiation, the drone's cumulative trajectory deviation will gradually increase, potentially deviating significantly from its original mission path, leading to extended mission execution time or even mission failure. This cumulative effect is difficult to detect and correct in a distributed negotiation mechanism that only considers the optimal outcome of a single conflict, because each conflict is treated as an independent event, lacking cross-event memory and compensation mechanisms.

[0119] This step involves retrospectively evaluating historical concession records before each concession begins. If it's detected that the drone has already made multiple concessions to the same high-priority party, it indicates that the drone is providing unconventional cooperation for the other party's mission. In this case, two alternative or parallel adjustment methods are triggered. Lowering the upper limit of this adjustment directly limits the maximum concession amount through hard constraints, preventing low-priority parties from retreating indefinitely, even if this increases the probability of a round of negotiation failing to converge and requiring a concession limit notification. This approach introduces a dynamic upper bound between S62 and S64 of S6, making the concession step size or upper limit no longer a fixed parameter, but adaptively decreasing based on historical data. The practical effect is to make low-priority parties more determined to maintain their own mission bottom line.

[0120] Another approach is to request temporary priority compensation from higher-priority drones. This doesn't require the drone to substantially change its flight path, but rather requests a temporary increase in the lower-priority drone's weight in its priority label. This reduces the weight difference in potential future conflicts, making it easier for the lower-priority drone to fall into the negotiable conflict category and illegally enter emergency avoidance during S5 assessments, or making the concession threshold for the lower-priority drone more favorable during S6 negotiations. This compensation mechanism utilizes historical concession records as evidence of the lower-priority drone's accumulated contributions, transforming concessions in cooperation into an intangible priority incentive.

[0121] This step is a pre-filtering step within the S6 main process. It doesn't change the specific mechanisms for calculating conflict geometry, step-by-step adjustments, and progressive verification. Instead, it determines the adjustment strategy parameters—how far the concession can retreat and its weight in subsequent interactions—before entering S61. By embedding this review and adjustment mechanism before concession initiation, the entire distributed negotiation not only addresses conflicts in the current spatial and temporal dimensions but also considers fairness and sustainability across conflict rounds in the temporal dimension. This helps improve the overall task completion balance of the drone swarm in multi-tasking, long-endurance operations.

[0122] In one or more embodiments of this application, when a drone simultaneously encounters multiple conflicts with multiple neighboring drones, the method further includes: Obtain the start time of the time window corresponding to each conflict; Multiple conflicts are sorted according to their starting time; for conflicts with the same starting time, they are further sorted from largest to smallest according to the difference in the negotiation priority weight involved. Based on the sorting results, a distributed negotiation protocol is initiated for each conflict in turn, and during the processing of the current conflict, the spatiotemporal flight pipeline associated with the conflict later in the time window is locked.

[0123] In this embodiment, multiple conflicts generated by multiple neighboring UAVs refer to a single UAV independently predicting potential conflict areas and time windows with two or more neighboring UAVs within the same time period. Each conflict has its own conflict time window, and its start time is the expected start time of the conflict. The order of start times refers to arranging the conflicts according to their chronological order on the timeline, with conflicts occurring earlier in time ranking higher. The negotiation priority weight difference, defined in previous steps, is used here for secondary sorting among conflicts with the same start time. A larger difference indicates a more significant difference in priority between the conflicting parties, resulting in a higher ranking. Sequentially initiating the distributed negotiation protocol means processing the first-ranked conflict completely according to the sorting results. After its negotiation converges and generates a conflict-free cooperative trajectory, the negotiation process for the second-ranked conflict is initiated, and so on, rather than initiating parallel negotiation for multiple conflicts simultaneously. Applying a lock refers to temporarily marking the spatiotemporal flight paths involved in conflicts that occur later in the time window at the start of the current time window as unmodifiable. The spatial boundaries and time intervals of these paths will not be altered by other negotiation processes until the current conflict is resolved. It is understood that the locked objects can include the corresponding paths of the local drone and neighboring drones involved in subsequent conflicts. The specific implementation of the lock can be achieved by setting status flags in the intent management unit. Once the current conflict is resolved and the cooperative trajectory is fixed, the locked paths are released and made available for adjustment in the next round of conflict negotiation.

[0124] This multi-conflict handling mechanism introduces sorting logic based on time windows and priority differences, as well as locking protection for unprocessed conflict pipelines, to build an ordered and non-overlapping serialized processing framework for complex scenarios with multiple concurrent conflicts.

[0125] When a drone faces multiple conflicts simultaneously, the most direct challenge lies in the potential temporal and spatial coupling of these conflicts. If distributed negotiation is initiated in parallel for different conflicts, lower-priority parties may need to make concessions to multiple higher-priority parties at the same time. Each adjustment may alter the geometric relationships with other conflicting parties, causing changes in the spatial regions and time windows of previously unresolved conflicts, and potentially even reactivating already resolved conflicts. This cross-influence can easily cause the entire conflict resolution process to fall into cyclical adjustments or oscillating divergence, making it difficult to converge to a stable, globally conflict-free solution within a finite number of steps.

[0126] By first obtaining and sorting the start times of each conflict time window, the system establishes the order of processing, with the basic principle being that conflicts occurring earlier are processed first. This aligns with the intuitive logic of time urgency, as conflicts earlier in the time window have less room for waiting; if they are processed later, danger may arise due to the actual approach of the aircraft during the waiting period. For conflicts with overlapping start times, they are further sorted by the difference in priority weights. Conflicts with larger differences have more distinct priorities, often indicating clearer avoidance obligations and faster negotiation convergence. Prioritizing these conflicts can quickly alleviate some of the conflict pressure, freeing up adjustment space for subsequent processing.

[0127] Negotiations are initiated sequentially based on the ranking results. While addressing the current conflict, the relevant channels for subsequent conflicts are locked. This combination of serialization and locking fundamentally severs the coupling and interference paths between conflicts. When addressing the current conflict, the channels for subsequent conflicts are frozen, meaning that adjustments made by lower-priority parties in the current conflict will not inadvertently involve or alter the states of channels for subsequent conflict parties. Simultaneously, subsequent conflict parties will not react prematurely to perceived adjustments in the environment, avoiding decision-making conflicts and uncontrollable superposition caused by parallel adjustments in multiple directions. The locking effect maintains the original state of subsequent conflicts, ensuring that previously predicted spatial areas and time windows for subsequent conflicts remain effective, eliminating the need to re-predict the entire multi-conflict situation in each round of negotiation.

[0128] After each round of negotiation to fix the result, the corresponding pipeline is unlocked. The system can then reconfirm whether the spatiotemporal geometry of the next ordering conflict still holds based on the updated cooperative trajectory. If it does, negotiation is initiated; otherwise, it can be skipped if the current adjustment has already caused the conflict to resolve itself naturally. This forms a processing rhythm of tackling problems one by one and verifying progress. This mechanism is fully compatible with the single-conflict negotiation process of S5 and S6 mentioned above. The negotiation of each conflict still follows the complete process of judging negotiability, yielding to lower priorities, verifying higher priorities, and iterative convergence. However, in multi-conflict scenarios, two additional management measures, ordering scheduling and pipeline locking, are added. The logic of these management measures does not intervene in the negotiation decision within a single conflict, but rather operates at the scheduling level between multiple negotiation processes. This allows the entire system to maintain the determinism of order when dealing with complex multi-machine interactions, helping to reduce the fluctuation amplitude of trajectory adjustments and the total number of iterations required for convergence.

[0129] In one or more embodiments of this application, when returning to S2 in S7 to re-determine the task urgency and negotiate the priority weight, if it is detected that the current task deadline has expired or the task has been completed, the determination of task urgency is stopped, the machine's spacetime flight pipeline is released, and the cooperative conflict avoidance process is exited.

[0130] In this embodiment, "task deadline expired" means that the current time has exceeded the final completion time specified by the task, meaning that the task has lost its meaning in the time dimension and there is no longer a need to reserve spatiotemporal resources for it, even if the UAV has not yet reached the predetermined target location. "Task completed" means that the UAV has arrived at the target area or completed the predetermined operation according to the task requirements, such as completing the inspection and photography of the designated area or completing the delivery of materials, and its task objective has been achieved. Stopping the determination of task urgency means that the process of calculating urgency based on the task deadline parameter and converting it into negotiation priority weight in S2 is no longer executed, because the urgency is no longer meaningful at this time. Releasing the spatiotemporal flight pipeline of the local machine means removing or marking the spatiotemporal flight pipeline generated by the local machine in S3, and possibly modified in subsequent negotiations, from the current cluster-shared situation map, no longer broadcasting it to neighboring UAVs, and no longer using it as a constraint condition for conflict detection of other UAVs. It can be understood that the process of releasing the pipeline can be achieved by sending a pipeline cancellation message, or by directly stopping the periodic broadcast of the pipeline, so that neighboring UAVs will automatically remove the pipeline from the occupied grid set because they have not received an update due to timeout. Exiting the collaborative conflict avoidance process means that the UAV stops executing the loop steps S2 to S7 in this method, no longer participates in subsequent intent broadcasting, conflict prediction, negotiation judgment and collaborative trajectory generation, and returns to relying solely on its own planned flight or entering other flight phases such as return home or landing.

[0131] This step embeds a mission lifecycle termination judgment into the S7 flight deviation feedback loop, enabling the renegotiation closed loop to have the ability to self-clean up and exit in an orderly manner.

[0132] In the main process defined in S7, when flight deviations exceed the preset tolerance, the system returns to S2 to reassess the mission urgency and triggers a new round of intent broadcasting, conflict prediction, and negotiation processes. This closed-loop design aims to address dynamic interference during flight, enabling the cooperative trajectory to continuously adapt to environmental changes. However, without any exit conditions, this closed loop could theoretically run indefinitely. During normal mission execution, this loop is necessary and beneficial; however, when the mission itself has ended or is no longer possible, continuing the loop not only wastes onboard computing resources and communication bandwidth but also imposes unnecessary constraints on other UAVs in the cluster that are still performing missions, as their spatiotemporal flight pipelines continue to occupy airspace resources.

[0133] This step introduces a clear termination condition for the closed loop by adding a task status check before returning to S2. When the task deadline is detected, it means that the task has expired regardless of whether it was completed, and it is no longer worthwhile to retain or secure flight paths for it. Continuing to occupy airspace resources may hinder other drones with tasks. When the task is detected as completed, it means that the drone has withdrawn from the swarm's collaborative operation role and should promptly relinquish the spacetime resources it occupied. In both cases, stopping the calculation of urgency and releasing the path is equivalent to the drone voluntarily deregistering from the collaborative conflict avoidance system, leaving no outdated or invalid constraints for the swarm.

[0134] The release of the pipeline allows other drones to instantly update the occupancy grid set built by S4. Potential conflicts previously predicted due to the drone's own pipeline may automatically disappear if they have not yet entered negotiation; if they are already under negotiation, the parties involved can detect that the conflict no longer exists through the updated occupancy grid set, thus terminating the negotiation in advance. This effect allows drones that have completed their missions to quickly and non-forcefully relinquish airspace rights, improving the efficiency of airspace resource utilization.

[0135] After exiting the collaboration process, the drone will no longer participate in any of steps S2 to S7, marking the termination of its status as a member of the cluster collaboration. This exit mechanism is symmetrical to the entry mechanism: after being assigned a task, the drone enters S1 to begin collaboration; after the task is completed or timed out, it leaves the collaboration process via this exit judgment. This design of entry and exit ensures that the entire conflict avoidance method is not a static, infinite execution framework, but a service process that dynamically starts and stops following the task lifecycle. This helps the cluster maintain the timeliness and rationality of the collaboration member set in long-endurance, multi-wave task scenarios.

[0136] In one or more embodiments of this application, when determining the task urgency of the UAV in S2, the remaining energy state of the UAV is also incorporated; if the remaining energy state is lower than a preset energy threshold, the determined task urgency is adjusted to reduce the negotiation priority weight accordingly.

[0137] In this embodiment, the remaining energy state refers to the current available energy reserve level of the UAV during mission execution, which can be characterized by indicators such as remaining power (State of Charge, SOC), remaining fuel, or remaining flight time. It is understood that the remaining energy state can be acquired in real time using conventional monitoring methods in the art, such as an onboard battery management system (BMS) or fuel level sensors. The preset energy threshold is a pre-set energy warning limit. When the remaining energy state is lower than this threshold, it indicates that the UAV's energy reserves have entered a critical range, and continuing to perform high-energy maneuvers or detours may pose a risk of difficulty in returning to base. This threshold can be set based on the minimum energy required for the UAV's return plus a certain safety margin, or it can be dynamically adjusted according to the mission type. Adjusting the determined mission urgency means multiplying the base mission urgency or negotiated priority weight originally calculated in S2 by a decay coefficient less than one, or subtracting a preset offset, so that the adjusted mission urgency value is lower than the original calculation result. Understandably, the adjustment operation node can be located after S22 calculates the basic task urgency and before S23 performs nonlinear mapping, or it can be located after S23 normalizes the output negotiation priority weight. Both can achieve a corresponding reduction in the final negotiation priority weight. A corresponding reduction in the negotiation priority weight means that the UAV's negotiation priority weight is lower than the result calculated solely based on the task time limit parameters. Therefore, it is more likely to be identified as the party that should proactively concede in the negotiation during the subsequent S5 judgment, and its spatiotemporal flight pipeline expansion margin will also be correspondingly reduced during the S6 negotiation.

[0138] This step introduces an energy dimension correction factor on the basis of urgency calculation driven by task time limit, and incorporates the consideration of the physical capabilities of the UAV into the priority generation process. This makes the negotiated priority weight not only reflect the urgency at the task level, but also take into account the actual feasibility of the UAV to perform concession maneuvers.

[0139] In the original S2 calculation framework, mission urgency primarily depends on two mission-level factors: mission value weight and remaining execution time. This calculation method implicitly assumes that as long as the mission urgency is high, the UAV is capable of holding its position in a conflict, refusing to concede, or performing complex evasive maneuvers. However, in actual flight, UAVs have limited energy reserves. A UAV with high mission urgency but severely depleted remaining energy, even if given a high negotiation priority weight, may not actually possess the capability to perform extensive detours or loiter outside the conflict zone for extended periods. If it is forced to maintain high-priority behavior in a conflict, such as insisting on preserving ample spacetime and expecting lower-priority parties to make significant concessions, the high-priority UAV may face a situation where it cannot complete its mission before its energy runs out if the lower-priority party is unable to completely evade due to concession limits or other reasons.

[0140] This step proactively reduces the mission urgency when energy is scarce by querying the remaining energy status and comparing it with a preset energy threshold. This reduction does not change the objective time limit and value of the mission itself, but rather applies an energy constraint discount when translating it into negotiation priority weight. After the reduction, the drone's negotiation priority weight will be lower than the result calculated solely based on the mission time limit parameter. This numerical change affects the behavior of several subsequent steps in a chain-like manner: in S3, the expansion range of the spatiotemporal flight pipeline shrinks due to the reduced weight, and the drone's preemptive posture in space and time becomes more convergent; in S5, the weight difference with other drones may therefore narrow or even reverse, making conflicts easier to determine as negotiable conflicts; in S6, if the drone is still in a low-priority position, its concession limit may be further limited due to its own energy shortage. However, in scenarios where energy shortage is detected, the drone may be more inclined to make limited concessions as early as possible and lock in an executable path, rather than insisting on a high-priority posture in negotiations in the hope that the other party will make significant concessions.

[0141] Overall, this energy-aware correction mechanism adds a dimension reflecting the physical feasibility boundary to the negotiation priority weights, ensuring that priorities respond not only to what the task requires but also to what the drone can still do. This multi-source fusion approach to expressing urgency helps reduce mission failures caused by energy depletion and also mitigates the risk of collaborative deadlock that might result from assigning high priority to drones lacking execution capabilities. This step is logically linked to S2's original task-time-driven computation framework, first deriving the task-side urgency and then correcting it with energy status. The structure is clear, with minimal intrusion into the original computation steps, facilitating its overlay and deployment on existing solutions.

[0142] In one or more embodiments of this application, before the potential conflict is predicted in S4, the predicted intent information of the neighboring UAVs in the future is generated using a preset intent inference model based on the received historical intent information sequence of the neighboring UAVs and their task time limit parameters; the consistency of the predicted intent information with the actual intent information broadcast by the neighboring UAVs is checked; when the deviation between the predicted intent information and the actual intent information is lower than a preset confidence threshold, the neighboring UAV is marked as a predictable object; based on the predicted intent information of the predictable object, the time starting point of the conflict prediction in S4 is extended to the future by a preset prediction duration, the potential conflicts within the extended time period are pre-identified, and the negotiable conflict judgment process in S5 is initiated in advance.

[0143] This solution introduces a neighborhood drone behavior prediction mechanism, which shifts the time benchmark for conflict detection and negotiation judgment from passively responding to actual broadcast information to proactively predicting based on prediction information, forming a three-step collaborative processing logic of prediction generation, reliability verification, and prediction expansion.

[0144] First, the scheme utilizes received historical intent information sequences and task time limit parameters from neighboring UAVs to generate predicted intent information through a pre-defined intent inference model. The historical intent information sequence records the spatiotemporal flight path changes broadcast by neighboring UAVs over a past period, containing their movement habits, path preferences, and response patterns to task time limits. The task time limit parameters provide the deadline constraints for the current task of the neighboring UAV, enabling the inference model to infer the flight path parameters the UAV might adopt in the future, based on historical behavioral patterns and factors such as its remaining execution time. This step expands the input information for conflict prediction from a single real-time broadcast snapshot to a multi-dimensional information set including historical trends and task constraints, providing an information foundation for early detection of potential conflicts.

[0145] Secondly, the solution does not blindly trust the prediction results, but instead includes a consistency verification step between the predicted intent information and the actual broadcast intent information. When the actual broadcast intent information from a neighboring drone arrives, it is compared with the previously generated predicted intent information. If the deviation is below a preset confidence threshold, it indicates that the drone's actual behavior closely matches the predicted behavior driven by historical patterns and task constraints, and its future behavior is highly predictable. At this point, it is marked as a predictable object. This verification step is essentially an online assessment of the stability of the neighboring drone's behavior. Only those that pass historical verification and are determined to have obvious behavioral patterns will trigger subsequent prediction extensions. This approach leverages the lead time provided by prediction while limiting the risk of erroneous negotiated judgments due to prediction deviations.

[0146] Finally, for neighboring drones marked as predictable targets, the solution uses their predicted intent information as supplementary input, extending the conflict prediction timeframe forward by a preset prediction duration. This means that during the conflict detection phase, not only are conflicts within the currently broadcast coverage period checked, but also potential conflicts within the prediction period are additionally examined. For conflicts pre-identified within the extended timeframe, the negotiated conflict assessment process is initiated directly in advance. This pre-initiation allows lower-priority parties to begin assessing the priority differences and time urgency of the conflict before it is actually broadcast and confirmed, but before it has a high probability of occurrence, thus allowing for more preparation and processing time for subsequent adjustments. This prediction-driven, forward-shifted negotiation window, compared to the conventional approach where negotiation can only begin after both parties have actually broadcast, effectively mitigates situations where a short time window leads to a judgment of an emergency conflict and triggers forced avoidance, increasing the proportion of conflicts that achieve trajectory coordination through peaceful negotiation. The entire solution, from prediction generation and reliable screening to pre-judgment initiation, relies on each other and converges step by step. Without disrupting the original distributed negotiation architecture, it extends the system's conflict perception and response capabilities forward by a configurable pre-judgment time window.

[0147] In one or more embodiments of this application, when the drone with the lower priority weight calculates the amount of spatial contraction or time offset required to eliminate the conflict in S6, the delay loss caused by the adjustment to its own task completion time and the incremental collision risk introduced by the pipeline contraction are evaluated simultaneously. Based on the task delay loss and the incremental collision risk, an adjustment cost function is constructed, and the value of the adjustment cost function is compared with the estimated value of the task delay cost that the high priority party may bear when the conflict is not resolved. When the value of the adjustment cost function exceeds the product of the estimated value of the task delay cost of the high priority party and the preset tolerance coefficient, the concession adjustment is suspended and a cost unacceptable notification is sent to the high priority weight drone. After the two parties exchange cost information, they re-evaluate whether to continue the negotiation or switch to other conflict handling branches.

[0148] Before the lower priority party makes a concession, the scheme constructs a complete link from cost self-assessment and cross-party comparison to cost information exchange and decision reassessment, so that the concession behavior in distributed negotiation is no longer driven solely by the conflict geometry, but is also constrained by task loss and secondary risks.

[0149] First, the scheme requires simultaneously evaluating two cost indicators while calculating the spatial contraction or time offset: the time delay loss caused by the adjustment to the task's completion time, and the incremental collision risk introduced by the pipeline contraction. This simultaneous evaluation transforms the concession behavior from a simple geometric problem into a self-aware decision-making process. The time delay loss reflects the erosion of task timeliness by the concession, while the incremental collision risk reflects the potential cascading safety impacts of the concession. These two indicators describe the negative consequences of the concession from both the task and safety perspectives, allowing the lower-priority party to obtain a quantitative estimate of the potential cost of the concession before taking any adjustment steps.

[0150] Next, the solution synthesizes the delay loss and the incremental risk of new collisions into an adjusted cost function value, and compares it with the estimated task delay cost that the high-priority party might bear if the conflict remains unresolved. The purpose of constructing this adjusted cost function is to compress the multi-dimensional concession costs into a comparable scalar, while the estimated task delay cost for the high-priority party represents the gains or losses the other party might gain or avoid from this negotiation. The comparison transforms the originally independently assessed unilateral gains and losses into a cross-party cost-benefit trade-off. This trade-off expands the basis for concession decisions from unilateral capability boundaries to the relative gains and losses of both parties.

[0151] This application further introduces a preset tolerance coefficient as a comparison buffer. Only when the adjustment cost function value of the low-priority party exceeds the product of the delay cost and the tolerance coefficient of the high-priority party will the pause in concessions and cost information exchange be triggered. The existence of the tolerance coefficient ensures that negotiation is not interrupted at the slightest disparity in costs between the two parties, but rather a certain priority passage margin is reserved for the high-priority party, with intervention only occurring when the cost of the low-priority party is significantly disproportionate. This mechanism finds an adjustable balance between protecting the passage efficiency of high-priority tasks and preventing the low-priority party from suffering excessive sacrifices.

[0152] When the cost to the lower-priority party does indeed exceed a reasonable range, the solution does not directly terminate the negotiation or force one party to concede. Instead, it suspends the concession adjustment and issues a notification to the higher-priority party that the cost is unacceptable. Subsequently, both parties exchange cost information and reassess. This design temporarily pulls the negotiation from a one-way concession model back to an assessment node with symmetrical information. After learning the specific cost of the other party, the higher-priority party can reassess whether it is still necessary to require the other party to continue adjusting, whether it can make minor adjustments itself to replace the other party's huge cost, or whether it should move to other conflict handling branches. This cost-transparent reassessment mechanism helps avoid unfair concessions caused by information asymmetry and also provides a cost-aware negotiation backtracking channel for complex multi-aircraft conflict scenarios. Overall, by quantifying the cost of concession behavior, comparing across parties, and exchanging cost information, this solution, while maintaining the original iterative framework, adds consideration to the fairness of task loss and the overall efficiency of the cluster, and is expected to improve the balance of task completion rates among different aircraft in long-endurance, multi-task alternating cluster operations.

[0153] A second aspect of this application provides a conflict avoidance system for collaborative operation of unmanned aerial vehicle (UAV) swarms, comprising a task management unit, a data acquisition unit, a urgency calculation unit, an intent management unit, a conflict detection unit, a negotiation control unit, and a flight monitoring unit. The task management unit is used to store and provide task time limit parameters for each UAV; The data acquisition unit, connected to the task management unit, is used to collect real-time motion status information and obtain task time limit parameters from the task management unit. The urgency calculation unit, connected to the data acquisition unit, is used to determine the urgency of the task based on the acquired task time limit parameters and the current time, and convert it into a negotiated priority weight. The intent management unit, connected to the data acquisition unit and the urgency calculation unit, is used to generate a spatiotemporal flight pipeline with preemption characteristics as intent information based on real-time motion state information and negotiated priority weights, and broadcast it to neighboring UAVs within the communication range. The conflict detection unit, connected to the intent management unit, is used to receive intent information broadcast by neighboring UAVs, and, in combination with the local planned mission path and motion status, predict the spatial areas and time windows where potential conflicts exist. The negotiation control unit, connected to the conflict detection unit and the intent management unit, is used to determine whether a conflict is negotiable by using the difference in negotiation priority weights between the local and neighboring UAVs and the urgency of the time window. When a conflict is determined to be negotiable, the distributed negotiation protocol is triggered, and the UAV with lower priority weights is controlled to adjust its spatiotemporal flight path, while the UAV with higher priority weights verifies the adjustment results. Through iteration, a conflict-free cooperative flight trajectory is generated. The flight monitoring unit, connected to the negotiation control unit and the data acquisition unit, is used to control the UAV to fly along a conflict-free cooperative flight trajectory and to monitor flight deviations using real-time motion status. When the flight deviation exceeds the preset tolerance, the urgency calculation unit is triggered to re-determine the mission urgency and negotiation priority weights.

[0154] The system provided in this application integrates task urgency calculation, preemptive intent broadcasting, conflict spatiotemporal prediction, hierarchical negotiation iteration, and flight deviation closed-loop monitoring into a complete distributed collaborative architecture through orderly connection and functional division between units. The functions of each unit correspond one-to-one with the independent method steps, enabling the system's information flow and processing logic to fully support the execution of conflict avoidance methods and possessing engineering feasibility for dynamic tasks and restricted communication scenarios.

[0155] A third aspect of this application provides an electronic device, including: one or more processors, one or more input devices, one or more output devices, and one or more memories. The processors, input devices, output devices, and memories communicate with each other via a communication bus. The memories store computer programs, including program instructions. The processors execute the program instructions stored in the memories. The processors are configured to invoke the program instructions to execute the aforementioned conflict avoidance method for collaborative operation of unmanned aerial vehicle (UAV) swarms.

[0156] It should be understood that in one or more embodiments of this application, the processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0157] Input devices may include touchpads, fingerprint sensors (for collecting the user's fingerprint information and fingerprint orientation information), microphones, etc., while output devices may include displays (LCDs, etc.), speakers, etc.

[0158] The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store information about the device type.

[0159] In specific implementations, the processor, input device, and output device described in one or more embodiments of this application may execute the implementation method described in any embodiment of the conflict avoidance method for collaborative operation of UAV swarm provided in the embodiments of this application, or may execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0160] In another embodiment of this application, an electronic device is provided. The electronic device stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the conflict avoidance method for collaborative operation of UAV swarms described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in an electronic device, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0161] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0162] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0163] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0164] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0165] 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 units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0166] 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.

[0167] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for conflict avoidance of cooperative work of a UAV cluster, characterized in that, include: S1. Obtain the task time limit parameters and real-time motion status information of each drone in the drone cluster; S2. Based on the task time limit parameters and the current time, determine the task urgency of the UAV and convert it into a negotiated priority weight; S3. Based on the real-time motion state information and negotiation priority weight, generate a spatiotemporal flight pipeline with preemption characteristics as intent information, and broadcast it to neighboring UAVs within the communication range. S4. Receive the intent information broadcast by the neighboring UAV, and combine it with the local planned mission path and motion status to predict the spatial areas and time windows where potential conflicts exist. S5. Utilize the difference in negotiation priority weights between the local and neighboring UAVs and the urgency of the time window to determine whether the conflict is a negotiable conflict, and trigger the distributed negotiation protocol when it is determined to be a negotiable conflict. S6. In the distributed negotiation protocol, the drone with the lower negotiation priority weight adjusts its own spatiotemporal flight path and sends the adjusted intention information. The drone with the higher negotiation priority weight verifies whether the adjusted intention information eliminates the conflict. The process is iterated until a conflict-free cooperative flight trajectory is generated. S7. The UAV flies according to the conflict-free cooperative flight trajectory and monitors flight deviations using real-time motion status. When the flight deviation exceeds the preset tolerance, it returns to S2 to redetermine the mission urgency and negotiate priority weights, and then executes subsequent steps.

2. The method of claim 1, wherein, In S2, the mission urgency of the UAV is determined based on the mission time limit parameters and the current time, and converted into a negotiated priority weight, including: S21. Obtain the task value weight corresponding to the task time limit parameter, and calculate the remaining execution time from the current time to the task deadline. S22. The ratio of the task value weight to the remaining execution time is used as the basic task urgency. S23. Using a preset monotonic nonlinear mapping function, the basic task urgency is converted into a normalized negotiation priority weight. The mapping function makes the rate of change of the negotiation priority weight greater when the remaining execution time is shorter.

3. The method of claim 1, wherein, In S3, based on the real-time motion state information and the negotiated priority weight, a spatiotemporal flight pipeline with preemptive characteristics is generated as intent information, including: S31. Based on real-time motion status and planned mission path, generate a nominal safe flight corridor for the aircraft within a future preset time domain, defined by position boundaries and speed range; S32. Based on the value of the negotiation priority weight, an expansion margin is superimposed on the spatial dimension of the nominal safe flight corridor, and the available time period is extended in the time dimension. The expansion range increases with the increase of the negotiation priority weight, forming a spatiotemporal flight channel with preemption characteristics. S33. Write a priority weight flag into the broadcast message of the spacetime flight pipeline and encapsulate it as the intent information.

4. The method of claim 1, wherein, S4 receives the intent information broadcast by the neighboring UAVs, and combines it with the local planned mission path and motion state to predict potential conflict areas and time windows, including: S41. Parse the received intent information and construct the occupancy grid set of the corresponding neighboring UAV in the time-space dimension. The occupancy grid set includes a three-dimensional spatial grid index and the corresponding occupancy time slot. S42. Discretize the local planned task path into a sequence of track points carrying timestamps according to a preset time step; S43. Perform spatiotemporal intersection detection on the track point sequence and the occupied grid set, and take the overlapping spatial grids and their covered time intervals as the potential conflict spatial regions and time windows.

5. The method of claim 4, wherein, S5 utilizes the difference in negotiation priority weights between the local and neighboring UAVs and the urgency of the time window to determine whether a conflict is a negotiable conflict, including: S51. Calculate the negotiation priority weight difference between the local machine and the neighboring UAVs that caused the conflict; S52. Obtain the duration of the time window; S53. When the difference in negotiation priority weights is less than a preset negotiation difference threshold and the duration is greater than the minimum negotiation window threshold, the conflict is determined to be a negotiable conflict; otherwise, it is determined to be an emergency conflict. S54. When an emergency conflict is identified, an emergency avoidance flag is generated and the distributed negotiation protocol is suspended.

6. The method according to claim 1, characterized in that, In S6, drones with lower negotiation priority weights adjust their spatiotemporal flight paths and send the adjusted intent information. Drones with higher negotiation priority weights verify whether the adjusted intent information eliminates conflicts, including: S61. For UAVs with lower negotiation priority weights, the amount of spatiotemporal flight pipeline space contraction or time offset required to eliminate the conflict is calculated based on the conflict geometry. S62. Modify the spatial boundary parameters or time interval parameters of the space-time flight pipeline step by step according to the preset adjustment step size. After each step of modification, keep the pipeline continuous and do not cause new conflicts with its own mission path. S63. After each modification is completed, the updated intent information is broadcast, and the drone with the higher priority weight in the negotiation performs conflict detection and verification. S64. If the conflict disappears, the current parameters are fixed as part of the cooperative trajectory. If the conflict does not disappear and the cumulative adjustment steps have not reached the preset concession limit, return to S62 and continue to modify according to the adjustment step size. If the conflict still exists after reaching the preset concession limit, pause the adjustment and send a concession limit notification to the high-priority weighted drone.

7. The method according to claim 5, characterized in that, When a conflict is determined to be an emergency, the method further includes: S5a: Trigger a priority-based emergency avoidance protocol, commanding the drone with the lower priority weight to perform the maximum available maneuver to avoid the obstacle under its own constraints; S5b: The UAV with the higher negotiation priority weight maintains its spatiotemporal flight path unchanged within the time window until the time window ends.

8. The method according to claim 1, characterized in that, Before S5 triggers the distributed negotiation protocol, it also includes: S50. Detect the quality of the current communication link with neighboring drones; if the quality of the communication link is lower than a preset reliability threshold, then determine the drone with the highest weight value from the drones that are accessible to the communication as a temporary coordination node according to the negotiation priority weight. The temporary coordination node collects the intention information of each drone, generates a safe trajectory using a centralized conflict resolution algorithm, and broadcasts it to each drone.

9. The method according to claim 1, characterized in that, Before drones with lower negotiation priority adjust their spatiotemporal flight paths in S6, the following are also included: Query the historical concession records between the local machine and the high-priority drone involved in the current conflict; if the historical concession records show that the local machine's cumulative number of concessions exceeds a preset value, then reduce the upper limit of the adjustment range, or request the high-priority drone to negotiate temporary compensation for priority weight.

10. A conflict avoidance system for collaborative operation of unmanned aerial vehicle (UAV) swarms, used to execute the method according to any one of claims 1-9, characterized in that, It includes a mission management unit, a data acquisition unit, a urgency calculation unit, an intent management unit, a conflict detection unit, a negotiation control unit, and a flight monitoring unit; The task management unit is used to store and provide task time limit parameters for each UAV; The data acquisition unit is connected to the task management unit and is used to collect real-time motion status information and obtain the task time limit parameters from the task management unit. The urgency calculation unit is connected to the data acquisition unit and is used to determine the urgency of the task based on the acquired task time limit parameters and the current time, and convert it into a negotiated priority weight. The intent management unit, connected to the data acquisition unit and the urgency calculation unit, is used to generate a spatiotemporal flight pipeline with preemption characteristics as intent information based on real-time motion state information and negotiated priority weights, and broadcast it to neighboring UAVs within the communication range. The conflict detection unit, connected to the intent management unit, is used to receive intent information broadcast by neighboring UAVs, and, in conjunction with the local planned task path and motion state, predict the spatial areas and time windows where potential conflicts exist. The negotiation control unit is connected to the conflict detection unit and the intent management unit. It is used to determine whether a conflict is negotiable by using the difference in negotiation priority weight between the local drone and the neighboring drones and the urgency of the time window. When a conflict is determined to be negotiable, it triggers a distributed negotiation protocol and controls the drone with lower priority weight to adjust its spatiotemporal flight path, and the drone with higher priority weight to verify the adjustment result. The conflict-free cooperative flight trajectory is generated iteratively. The flight monitoring unit is connected to the negotiation control unit and the data acquisition unit. It is used to control the UAV to fly according to the conflict-free cooperative flight trajectory and to monitor flight deviations using real-time motion status. When the flight deviation exceeds the preset tolerance, the urgency calculation unit is triggered to re-determine the mission urgency and negotiation priority weight.

Images