A method for unmanned aerial vehicle cluster cooperative risk early warning based on a cloud native platform

Abstract

The application is suitable for the technical field of safety warning, and provides a UAV cluster cooperative risk warning method based on a cloud native platform, which comprises the following steps: collecting UAV cluster cooperative state data in real time based on the cloud native platform, and constructing a formation error index for quantifying the degree of cooperative abnormality; through monitoring the index and its change rate, presetting multi-stage trigger conditions, and triggering communication link optimization, local trajectory adjustment or global avoidance trajectory planning strategies respectively. The communication link optimization guarantees data transmission by switching to a relay communication topology, the local trajectory adjustment adjusts the formation by issuing flight parameter correction, and the global avoidance trajectory planning generates a smooth avoidance trajectory and issues an instruction via a backup link. The application realizes comprehensive perception of cluster cooperative risk and dynamic hierarchical response, so that the warning has higher accuracy and adaptability, and effectively improves the overall safety and task reliability of cluster flight.

Description

Technical Field

[0001] This invention relates to the field of security early warning technology, and in particular to a method for collaborative risk early warning of drone swarms based on a cloud-native platform. Background Technology

[0002] Unmanned aerial vehicle (UAV) swarm collaborative flight technology is widely used in reconnaissance, disaster monitoring, and logistics delivery. Existing UAV swarm monitoring systems typically employ a centralized ground station monitoring architecture. This involves deploying airborne sensors and communication modules on each UAV to collect real-time flight status parameters, including position coordinates, flight speed, remaining battery power, and attitude information. The ground station maintains periodic communication with each UAV in the swarm via wireless data links, receiving status data reported by each UAV and displaying the overall swarm situation on an electronic map in real time. When the flight parameters of a UAV exceed a preset safety threshold, the ground station monitoring system triggers an audible and visual alarm, prompting manual intervention from operators or automatically sending emergency commands such as return-to-home or hovering instructions to the malfunctioning UAV.

[0003] In existing technologies, monitoring systems only focus on the independent state parameters of individual drones, failing to detect real-time anomalies in the collaborative states between drones. These anomalies include cluster-level risks such as gradually widening formation consistency deviations, continuously increasing inter-drone communication delays, and successive loss of heartbeat signals from multiple drones. This results in blind spots and delays in risk identification. When anomalies are detected, the system triggers a single-level alarm based on a fixed threshold, lacking the ability to dynamically adjust warning strategies according to the degree of risk. Minor deviations and serious faults use the same alarm method, making it difficult for operators to quickly assess the severity of the risk and take differentiated countermeasures. When the main communication link deteriorates or is interrupted due to interference or obstruction, the system cannot ensure the reliable delivery of warning signals and emergency commands, leading to the inability to execute avoidance commands, cluster mission failures, or even collisions. Summary of the Invention

[0004] This invention provides a cloud-native platform-based method for early warning of risks in drone swarm collaboration, which addresses the problem of the inability to fully perceive abnormal collaborative states of drone swarms and to dynamically respond in a tiered manner.

[0005] This invention provides a method for collaborative risk warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform, comprising:

[0006] The monitoring agent is deployed in a containerized manner on a cloud-native platform, and the monitoring agent collects the collaborative status data of the drone cluster in real time.

[0007] Based on the cooperative state data, a formation error index is constructed to quantify the degree of abnormality in the cooperative state of the UAV swarm;

[0008] The system monitors the value and rate of change of the formation error index in real time, and presets multiple different trigger conditions, each corresponding to a warning response strategy; wherein:

[0009] When the formation error index exceeds a first preset threshold and the duration exceeds a first preset duration, and / or the rate of change of the formation error index exceeds a first preset rate of change threshold and the duration exceeds a first preset duration, a communication link optimization strategy is executed to ensure data transmission by dynamically switching to a relay communication topology composed of neighboring UAVs.

[0010] When the formation error index exceeds the second preset threshold and the duration exceeds the second preset duration, and / or the rate of change of the formation error index exceeds the second preset rate of change threshold and the duration exceeds the second preset duration, a local trajectory adjustment strategy is executed, and flight parameter adjustment instructions are issued to the affected UAVs to correct the formation.

[0011] When the formation error index exceeds the third preset threshold and the duration exceeds the third preset duration, and / or the rate of change of the formation error index exceeds the third preset rate of change threshold and the duration exceeds the third preset duration, a global avoidance trajectory planning strategy is executed, a trajectory planning algorithm is called to generate a smooth avoidance trajectory for the threatened UAV, and the trajectory command is sent to the corresponding UAV through a backup communication link.

[0012] Furthermore, the collaborative status data includes the round-trip communication delay between UAVs, the heartbeat signal reporting status of each UAV, the real-time position coordinates of each UAV, and the collaborative control command data between the UAVs.

[0013] Furthermore, the formation error index is constructed according to the following steps:

[0014] Based on the real-time position coordinates of each drone and the preset theoretical formation position coordinates, the position deviation of each drone is calculated, and the average position deviation of the cluster is obtained based on the statistical analysis of the position deviations of all drones in the cluster.

[0015] Based on the heartbeat signal reported by each drone, count the number of drones that are currently out of contact and calculate the proportion of the number of drones that are out of contact relative to the total number of drones in the cluster.

[0016] Based on the round-trip communication delay between UAVs, the deviation between the average communication delay of the cluster and the preset benchmark communication delay is calculated, and then normalized to obtain the communication delay deviation.

[0017] Calculate the rate of change of the cluster's average location deviation, disconnection rate, and communication delay deviation at the current moment;

[0018] The preliminary index is obtained by weighted summing of the cluster's average location deviation, disconnection rate, communication delay deviation, and their respective rates of change.

[0019] The preliminary index is transformed using a nonlinear mapping function to obtain the final formation error index.

[0020] Furthermore, the preset thresholds corresponding to the multiple different triggering conditions satisfy the following: first preset threshold < second preset threshold < third preset threshold; the corresponding preset durations are associated with each preset threshold; and the corresponding preset rate of change thresholds satisfy the following: first preset rate of change threshold < second preset rate of change threshold < third preset rate of change threshold.

[0021] Furthermore, the method of ensuring data transmission by dynamically switching to a relay communication topology composed of neighboring drones includes:

[0022] Identify target drones whose direct link quality to the ground station is degraded;

[0023] The current status parameters of multiple neighboring drones within a preset range around the target drone are obtained. The status parameters include location coordinates, remaining battery power, current communication load, and link quality with the target drone.

[0024] Calculate the path quality score for each candidate communication path with each neighboring UAV as a relay node based on the state parameters.

[0025] The candidate communication path with the highest path quality score is selected as the target relay communication topology, and the data transmission link of the target UAV is switched to the target relay communication topology.

[0026] Furthermore, the step of issuing flight parameter adjustment commands to the affected drones to correct their formation includes:

[0027] Determine the affected drones whose formation needs to be adjusted and their theoretical formation positions;

[0028] Obtain the position deviation between the current position of the affected UAV and the theoretical formation position, as well as the rate of change of the position deviation at the current moment;

[0029] Based on the position deviation and the rate of change of the position deviation, the expected flight parameter adjustment amount for the affected UAV is determined;

[0030] A flight parameter adjustment command is generated based on the desired flight parameter adjustment amount, and the flight parameter adjustment command is sent to the affected UAV.

[0031] Furthermore, the desired flight parameter adjustment amount includes at least one of the desired speed adjustment amount, desired heading adjustment amount, and desired attitude adjustment amount.

[0032] Furthermore, the invocation of the trajectory planning algorithm to generate a smooth avoidance trajectory for the threatened drone includes:

[0033] Obtain the current flight status parameters of the threatened drone and the location of the target avoidance point;

[0034] Based on the current flight state parameters, the target avoidance point position, and preset UAV dynamics constraints, an avoidance trajectory from the current position to the target avoidance point is generated. The avoidance trajectory satisfies the smoothness constraints of velocity, acceleration, and jerk.

[0035] Furthermore, when generating the avoidance trajectory, the acceleration parameters for the acceleration phase and the acceleration parameters for the deceleration phase are independently set according to the current speed of the threatened UAV, the preset maximum acceleration, the preset maximum deceleration, and the position of the target avoidance point, so that the avoidance trajectory has different acceleration change rates in the acceleration and deceleration phases.

[0036] Furthermore, the step of sending trajectory commands to the corresponding UAV via a backup communication link includes:

[0037] Monitor the communication quality of the main communication link. When the main communication link fails to meet the preset communication quality requirements, switch the transmission channel of the trajectory command to a pre-established backup communication link.

[0038] The trajectory command is sent to the corresponding UAV through the backup communication link, and the return confirmation information from the corresponding UAV is received. If the confirmation information is not received within a preset time, the trajectory command is resent.

[0039] As can be seen from the above technical solutions, the present invention has the following advantages:

[0040] This invention deploys a monitoring agent in a containerized manner on a cloud-native platform to collect real-time collaborative status data of a drone swarm and constructs a formation error index based on this data. By monitoring the value and rate of change of the formation error index in real time, multiple different trigger conditions are preset, each corresponding to a warning response strategy. When the formation error index exceeds a preset threshold and the duration exceeds a preset duration, and / or its rate of change exceeds a preset rate of change threshold and the duration exceeds a preset duration, a communication link optimization strategy, a local trajectory adjustment strategy, or a global avoidance trajectory planning strategy is triggered, respectively. The communication link optimization strategy ensures data transmission by dynamically switching to a relay communication topology composed of neighboring drones; the local trajectory adjustment strategy corrects the formation by issuing flight parameter adjustment commands to the affected drones; and the global avoidance trajectory planning strategy generates a smooth avoidance trajectory for the threatened drones by calling a trajectory planning algorithm and sends it through a backup communication link. This invention achieves comprehensive quantitative perception of the cluster's collaborative status by constructing a formation error index that integrates multi-source collaborative data. It can promptly detect cluster-level anomalies such as formation consistency deviation, increased communication latency, and heartbeat disconnection, thus compensating for the blind spots of traditional single-machine monitoring. By combining time and numerical dimensions to set multi-level triggering conditions, it effectively filters out false alarms caused by instantaneous interference. Furthermore, it dynamically matches differentiated response strategies according to different levels of collaborative risk, making the early warning more accurate and adaptive, and effectively improving the overall safety and mission reliability of cluster flight. Attached Figure Description

[0041] Figure 1 This is a schematic flowchart of an embodiment of a cloud-native platform-based drone swarm collaborative risk warning method according to the present invention.

[0042] Figure 2 This is a schematic diagram of the formation error index construction process in this invention;

[0043] Figure 3 This is a schematic diagram of the relay topology switching process in the communication link optimization strategy of this invention;

[0044] Figure 4 This is a schematic diagram of the formation correction process for the local trajectory adjustment strategy in this invention. Detailed Implementation

[0045] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0046] Example

[0047] The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The method in this application will be described below from the perspective of system implementation. Please refer to... Figure 1 The method provided in this application includes the following steps:

[0048] Step 1: Deploy the monitoring agent in a containerized manner on a cloud-native platform to collect real-time collaborative status data of the drone swarm;

[0049] In this embodiment, the collaborative status data includes the round-trip time delay between UAVs, the heartbeat signal reporting status of each UAV, the real-time position coordinates of each UAV, and the inter-UAV collaborative control command data. The round-trip time delay refers to the time interval between the ground station sending test data packets to each UAV and between the UAVs themselves via wireless data links, reflecting the real-time congestion level and transmission quality of the communication link. Here, a test data packet is sent to each UAV every second, and the round-trip time delay is recorded. The heartbeat signal reporting status refers to the survival signal sent by each UAV to the ground station once per second, along with the timestamp of its most recent successful reception, used to determine whether the UAV is in an online and controllable state. The real-time position coordinates refer to the current latitude, longitude, altitude, and speed information of the UAV collected through the airborne GPS system, used to determine the actual position of the UAV in three-dimensional space. The inter-UAV collaborative control command data refers to the formation-keeping commands, relative position deviation adjustment commands, and task collaborative control signaling transmitted within the cluster through the robot operating system network.

[0050] Specifically, the monitoring agent program and its dependent runtime environment are packaged into Docker container images. These images are then scheduled to run on various physical server nodes within the cluster via a Kubernetes container orchestration platform. Each monitoring agent container is allocated independent CPU and memory resources to ensure automatic horizontal scaling of the monitoring agent based on changes in the drone cluster size. The monitoring agent containers running on each server node capture communication data packets on the wireless data link in real time by calling underlying network interfaces, subscribing to robot operating system topics to obtain inter-drone coordination instructions, sending a heartbeat detection request to each drone every second and recording the response status, and simultaneously obtaining GPS location reports from each drone from the ground station via an application programming interface. All collected coordination status data is encapsulated in a unified timestamp format and aggregated to the central data processing module via a message queue service on a cloud-native platform.

[0051] Step 2: Construct a formation error index based on the collaborative state data to quantify the degree of abnormality in the collaborative state of the UAV swarm;

[0052] In this embodiment, the formation error index is a dimensionless quantitative indicator used to comprehensively reflect the degree of coordination anomalies in the UAV swarm across three dimensions: formation maintenance, communication status, and swarm integrity. A higher value indicates a worse swarm coordination state and a higher level of risk. (See also...) Figure 2 Specifically, construct it according to the following steps:

[0053] 1. Calculate the position deviation of each UAV based on its real-time position coordinates and the preset theoretical formation position coordinates. Calculate the average position deviation of the cluster based on the position deviation statistics of all UAVs in the cluster.

[0054] For the first in the cluster The drone's real-time position coordinates are , Indicates the first The longitude coordinates of the drone Indicates the first The latitude coordinates of the drone Indicates the first The altitude coordinates of the drone; the theoretical formation position coordinates are... , Indicates the first The theoretical longitude coordinates of the drone Indicates the first The theoretical latitudinal coordinates of the drone Indicates the first The theoretical altitude coordinates of the drone; then the first... Position deviation of the drone Calculated according to the Euclidean distance formula: In the formula This represents the positional deviation component along the longitude direction. This represents the positional deviation component in the latitudinal direction. This represents the positional deviation component in the height direction. Average positional deviation of the cluster. The arithmetic mean of the positional deviations of all UAVs is calculated using the following formula: In the formula This represents the total number of drones in the cluster. The theoretical formation position coordinates are determined based on the preset formation of the drone cluster and the current mission stage. The preset formation is a spatial arrangement of drones pre-defined according to mission requirements, including diamond formation, line formation, circular formation, etc. In this embodiment, a diamond formation is used as an example. In this formation mode, the theoretical position offset of each drone relative to the navigator is pre-calculated and stored in the ground station database. The specific offset is determined based on the formation geometry parameters.

[0055] 2. Based on the heartbeat signal reports of each drone, count the number of drones that are currently out of contact, and calculate the proportion of the number of drones that are out of contact relative to the total number of drones in the cluster;

[0056] The heartbeat signal reporting status here is determined by the timestamp of the most recent successful heartbeat signal reception. If the time interval between the current moment and the last successful heartbeat signal reception of a drone exceeds 3 seconds, the drone is considered to be out of contact. The system then counts the number of out-of-contact drones at the current moment. missing person rate according to calculate.

[0057] 3. Based on the round-trip communication delay between UAVs, calculate the deviation between the average communication delay of the cluster and the preset benchmark communication delay, and perform normalization processing to obtain the communication delay deviation;

[0058] First, calculate the average round-trip delay between all drones in the cluster and the ground station at the current moment. The calculation method is as follows: For each drone, record the number of drones... The most recent round-trip delay measurement between the drone and the ground station For all Sum and divide by the total number of drones ,Right now The preset baseline communication delay is here. This is the baseline value for the average communication delay of the cluster, measured under ideal, interference-free conditions; here, it is taken as 20 milliseconds. Communication delay deviation. according to Calculation, for Normalization is performed to obtain the communication delay deviation. The normalization method uses truncation normalization, that is... In the formula This is the preset maximum allowable communication delay deviation, and the value here is [value missing]. Milliseconds, the formula first... Limited to Within the interval, divide by Mapped to The interval ensures that the communication delay deviation is always within a reasonable range.

[0059] 4. Calculate the rate of change of the cluster's average location deviation, the proportion of disconnected devices, and the communication delay deviation at the current moment;

[0060] The rate of change is the difference between the current value and the previous value, divided by the sampling time interval. For the average location deviation of a cluster, the rate of change... The expression is In the formula , The average positional deviation of the cluster between the current and previous time points. The sampling time interval is 1 second. Similarly, the rate of change of the loss-of-connection ratio... The expression is In the formula , This represents the ratio of the current time point to the previous time point in terms of lost communication. It also represents the rate of change of the communication delay deviation. The expression is In the formula , This represents the communication delay deviation between the current moment and the previous moment.

[0061] 5. The cluster's average location deviation, loss of connection rate, communication delay deviation, and their respective rates of change are weighted and summed to obtain a preliminary index;

[0062] Preliminary index According to the formula Calculate, where , , , , , These are weighting coefficients. In this embodiment, the weighting coefficients are set according to the importance of each indicator to the cluster collaboration risk, and the preferred value is... , , , , , Furthermore, the sum of all weight coefficients is 1. The weight coefficients are dynamically adjusted based on different task scenarios and cluster sizes; for example, they can be appropriately increased in scenarios requiring high formation accuracy. The value of .

[0063] 6. The preliminary index is transformed through a nonlinear mapping function to obtain the final formation error index.

[0064] The nonlinear mapping function here uses a Sigmoid type function. ,in This is the curve steepness coefficient, which is set to 5 in this embodiment. This represents the offset of the midpoint; in this embodiment, it is set to 0.5. It is a natural constant. This is the preliminary exponent at the current moment. The function's purpose is to convert the preliminary exponent... Mapped to Within the range, and By amplifying its sensitivity to changes when approaching the median value, the formation error index becomes more responsive to minor changes in the initial index in high-risk areas, thus detecting trends of escalating risk earlier. It should be noted that this formation error index... The calculation is performed in real time, for each sampling time. The corresponding formation error index is calculated based on the coordinated state data collected at the current moment. This forms a sequence of formation error index changes over time, providing a data basis for subsequent monitoring of its value and rate of change.

[0065] Step 3: Monitor the value and rate of change of the formation error index in real time, and preset multiple different trigger conditions, each corresponding to a warning response strategy; among which:

[0066] The formation error index refers to the formation error index calculated based on the current moment. Its value ranges from 0 to 1, with a larger value indicating a higher degree of abnormality in the cluster's coordination state. The rate of change of the formation error index refers to how quickly the formation error index changes relative to the formation error index at the previous moment, according to the formula... Calculation, where The sampling time interval is 1 second. , The formation error index for the current time and the previous time; the rate of change. A positive value indicates that the coordination status is deteriorating, while a negative value indicates that the coordination status is improving. Presetting multiple different trigger conditions refers to setting three levels of early warning trigger conditions based on the different severity levels of the formation error index. Each level corresponds to a set of independent threshold parameters and an early warning response strategy, in order to achieve tiered early warning and differentiated response.

[0067] In this embodiment, the preset thresholds corresponding to multiple different triggering conditions satisfy the following: first preset threshold < second preset threshold < third preset threshold; the corresponding preset duration is associated with each preset threshold; and the corresponding preset rate of change threshold satisfies the following: first preset rate of change threshold < second preset rate of change threshold < third preset rate of change threshold.

[0068] Specifically, the first, second, and third preset thresholds are three critical values ​​used to classify the severity of the formation error index. The first preset threshold represents the critical point at which a slight abnormality occurs in the cluster coordination state, and in this embodiment, it is set to 0.3; the second preset threshold represents the critical point at which a significant abnormality occurs in the cluster coordination state, and in this embodiment, it is set to 0.6; the third preset threshold represents the critical point at which a severe abnormality occurs in the cluster coordination state, and in this embodiment, it is set to 0.8. The three thresholds increase sequentially, corresponding to a gradual increase in the risk level.

[0069] The first, second, and third preset durations are associated with the first, second, and third preset thresholds, respectively, to prevent false triggering due to instantaneous fluctuations. The first preset duration refers to the time required after the formation error index exceeds the first preset threshold; in this embodiment, it is set to 2 seconds. The second preset duration refers to the time required after the formation error index exceeds the second preset threshold; in this embodiment, it is set to 1.5 seconds. The third preset duration refers to the time required after the formation error index exceeds the third preset threshold; in this embodiment, it is set to 1 second. As the risk level increases, the required duration gradually decreases to reflect the requirement for rapid response to high-risk conditions.

[0070] The first, second, and third preset rate of change thresholds are three critical values ​​used to determine the rate of deterioration of the coordinated state. The first preset rate of change threshold represents the critical rate of slight deterioration of the coordinated state, which is set to 0.1 per second in this embodiment; the second preset rate of change threshold represents the critical rate of significant deterioration of the coordinated state, which is set to 0.2 per second in this embodiment; and the third preset rate of change threshold represents the critical rate of rapid deterioration of the coordinated state, which is set to 0.3 per second in this embodiment. The three rate of change thresholds increase sequentially, corresponding to a gradual increase in the rate of risk deterioration. When the rate of change of the formation error index exceeds the rate of change threshold of the corresponding level and the duration exceeds the preset duration of the corresponding level, a warning response of the corresponding level can also be triggered.

[0071] The values ​​of the aforementioned threshold parameters were determined statistically based on a large amount of UAV swarm simulation test data and actual flight test results. This ensures timely early warning while effectively filtering out false alarms caused by communication jitter and transient interference. Based on the triggering conditions set above, this embodiment further establishes a three-level early warning response strategy, corresponding to the following sub-steps.

[0072] First sub-step: When the formation error index exceeds the first preset threshold and the duration exceeds the first preset duration, and / or the rate of change of the formation error index exceeds the first preset rate of change threshold and the duration exceeds the first preset duration, execute the communication link optimization strategy and ensure data transmission by dynamically switching to a relay communication topology composed of neighboring UAVs.

[0073] In this embodiment, the first preset threshold is set to 0.3, the first preset duration is set to 2 seconds, and the first preset rate of change threshold is set to 0.1 per second. When the formation error index exceeds 0.3 for more than 2 seconds, or the rate of change of the formation error index exceeds 0.1 per second for more than 2 seconds, or both conditions are met simultaneously, it indicates a slight anomaly in the cluster coordination state. This anomaly is related to a decline in the quality of the communication link between some UAVs and the ground station, such as increased communication latency and higher packet loss rate, which may cause formation control commands to fail to be delivered in a timely manner. Therefore, a communication link optimization strategy is implemented. By dynamically adjusting the communication topology, multi-hop communication paths are constructed using surrounding UAVs as relay nodes to replace the degraded direct links, thereby ensuring the stability and reliability of data transmission and preventing further deterioration of the coordination state due to communication problems. (See also...) Figure 3 The communication link optimization strategy is implemented according to the following steps:

[0074] 1. Identify the target drone whose direct link quality to the ground station is degraded;

[0075] Real-time monitoring of the round-trip latency and packet loss rate between each drone and the ground station; setting conditions for determining link quality degradation: when the round-trip latency of a drone exceeds 100 milliseconds for three consecutive samples, or the packet loss rate exceeds 10% for three consecutive samples, the quality of the direct link between the drone and the ground station is determined to be degraded, and the drone is identified as the target drone.

[0076] 2. Obtain the current status parameters of multiple neighboring drones within a preset range around the target drone. The status parameters include location coordinates, remaining battery power, current communication load, and link quality with the target drone.

[0077] The preset range is a spherical area with a radius of 500 meters centered on the target drone. This range is set according to the typical communication distance and formation size of the drone swarm, ensuring that there are a sufficient number of candidate relay nodes within the range. Real-time status parameters of all other drones within this area are obtained through a ground station, including: position coordinates, remaining battery power, current communication load, and link quality with the target drone. The current communication load is the number of data streams that the drone is currently forwarding as a relay node, with each data stream representing one relay connection. The number of data streams is the average of three consecutive measurements taken by sending probe packets to the target drone to measure the round-trip delay and packet loss rate.

[0078] 3. Calculate the path quality score for each candidate communication path with each neighboring UAV as a relay node based on the state parameters;

[0079] For each neighboring drone Its candidate communication path quality score as a relay node Calculate using the following formula:

[0080]

[0081] in: Link quality scores are calculated based on the target drone and the relay drone. The communication round-trip delay and packet loss rate are comprehensively determined; specifically, the communication round-trip delay... Normalization to The normalization formula for the interval is: The maximum allowable latency is 200 milliseconds; the packet loss rate... Normalization Then take the average of the two. It should be noted that here... Indicates the target drone and the candidate relay drone The round-trip communication delay between them, and the aforementioned method used to construct the formation error index. The end-to-end delay between the drone and the ground station differs in physical meaning and application scenario. The former is used to evaluate the quality of the relay link, while the latter is used to evaluate the overall communication status of the cluster.

[0082] Score based on remaining battery power, according to the relay drone Remaining battery percentage Linear mapping to the 0-1 interval, i.e. The higher the remaining battery power, the higher the score;

[0083] For load scoring, based on relay drones Current communication load The lower the load, the higher the score. In this embodiment, the maximum allowable load is set to 5. When the load exceeds 5, the node is no longer suitable as a relay, and the score is directly set to 0; otherwise, the score calculation formula is as follows: ;

[0084] , , As the weighting coefficient, in this embodiment, it is set according to the degree of influence of each indicator on the relay path quality. , , This reflects the priority given to link quality.

[0085] 4. Select the candidate communication path with the highest path quality score as the target relay communication topology, and switch the data transmission link of the target UAV to the target relay communication topology.

[0086] Specifically, the scores of all neighboring drones are compared, and the drone with the highest score is selected as the relay node, establishing a communication path of "target drone - relay drone - ground station". The ground station sends a link switching command to the target drone through the currently available link, instructing it to send subsequent data to the selected relay drone. Simultaneously, it sends a relay forwarding command to the relay drone, configuring its data forwarding rules so that it can receive data from the target drone and forward it to the ground station. After the switchover is complete, data transmission between the target drone and the ground station is conducted through the relay link. The ground station continuously monitors the quality of the direct link. When the direct link quality returns to normal for 5 consecutive seconds, it can proactively switch back to the direct link to reduce the load on the relay node. If this strategy is triggered again during subsequent monitoring, the above steps are repeated for dynamic adjustment.

[0087] Second sub-step: When the formation error index exceeds the second preset threshold and the duration exceeds the second preset duration, and / or the rate of change of the formation error index exceeds the second preset rate of change threshold and the duration exceeds the second preset duration, execute the local trajectory adjustment strategy and issue flight parameter adjustment instructions to the affected UAVs to correct the formation.

[0088] In this embodiment, the second preset threshold is set to 0.6, the second preset duration is set to 1.5 seconds, and the second preset rate of change threshold is set to 0.2 per second. When the formation error index exceeds 0.6 and lasts for more than 1.5 seconds, or the rate of change of the formation error index exceeds 0.2 per second and lasts for more than 1.5 seconds, or both conditions are met simultaneously, it indicates that the cluster coordination state has become significantly abnormal. This is mainly manifested in the fact that the actual positions of some UAVs deviate significantly from the theoretical formation positions, but have not yet reached the level of serious loss of control. This abnormality is caused by the accumulation of local disturbances or slight control deviations. If not corrected in time, it may further develop into the risk of formation separation or collision. Therefore, a local trajectory adjustment strategy is implemented. By issuing refined flight parameter adjustment commands to individual or some UAVs that have deviated from the formation, they are made to actively return to the theoretical formation positions, restore formation consistency, and thus avoid escalation of risks. See also Figure 4 The local trajectory adjustment strategy is implemented according to the following steps:

[0089] 1. Identify the affected drones whose formation needs to be adjusted and their theoretical formation positions;

[0090] Iterate through all drones in the cluster, for each drone Calculate the Euclidean distance between its current position and the theoretical formation position. ,like If the preset adjustment trigger threshold is exceeded (in this case, 5 meters), the drone is identified as an affected drone. The theoretical formation position of an affected drone refers to its coordinates relative to the lead drone, based on the current formation and mission phase. These coordinates are calculated and stored in real time by the ground station using the formation control algorithm.

[0091] 2. Obtain the positional deviation between the current position of the affected UAV and the theoretical formation position, as well as the rate of change of the positional deviation at the current moment;

[0092] For each target drone Its positional deviation vector Defined as theoretical formation position coordinates Subtract current position coordinates ,Right now The magnitude of the positional deviation The direction is from the current position to the theoretical position. The rate of change of position deviation. It refers to the difference between the position deviation vector at the current moment and the position deviation vector at the previous moment, divided by the sampling time interval. ,in , This is the position deviation vector between the current time and the previous time. The value is taken as 1 second, and this rate of change reflects whether the deviation is increasing or decreasing.

[0093] 3. Based on the position deviation and the rate of change of the position deviation, determine the expected flight parameter adjustment amount for the affected UAV; wherein, the expected flight parameter adjustment amount includes at least one of the expected speed adjustment amount, expected heading adjustment amount, and expected attitude adjustment amount;

[0094] When the magnitude of the position deviation When the deviation is less than the first deviation threshold, speed adjustment is preferred. Expected speed adjustment amount. Calculated according to the proportional control law: The proportionality coefficient The value is 0.5, and the differential coefficient is... The value is 0.2. This is a sign function; the adjustment increases when the rate of change is in the same direction as the deviation direction, and decreases otherwise. The final target flight speed is the current speed plus the desired speed adjustment along the deviation direction.

[0095] When the magnitude of the position deviation When the deviation is greater than or equal to the first deviation threshold and less than the second deviation threshold, the heading adjustment method should be used first. Expected heading adjustment amount. Calculated based on the angle between the deviation direction and the current flight direction: In the formula: The deviation between the current longitude position and the theoretical longitude position of the affected drone. The deviation between the current latitude position and the theoretical latitude position of the affected drone. The affected drone's current heading angle, i.e., the angle between its current velocity direction and the due east direction, is obtained in real time by the onboard heading sensor; the result is then normalized. Within the range. Simultaneously, fine-tuning the speed allows the drone to maintain an appropriate forward speed while turning.

[0096] When the magnitude of the position deviation When the deviation is greater than or equal to the second deviation threshold, both speed and heading adjustments are applied simultaneously. The desired heading adjustment is the same as above, while the desired speed adjustment is based on proportional-derivative control with the addition of a basic regression speed term, set to 8 m / s, to ensure rapid return to the theoretical position.

[0097] It should be noted that the aforementioned first and second deviation thresholds are determined comprehensively based on the UAV's dynamic characteristics, formation accuracy requirements, and flight safety margins. In this embodiment, the first deviation threshold is set at 10 meters, corresponding to the allowable slight positional deviation of the UAVs during formation flight. Within this range, only minor speed adjustments are needed to restore formation, avoiding frequent heading adjustments that could affect formation stability. The second deviation threshold is set at 20 meters in this embodiment, corresponding to the critical value at which the UAVs significantly deviate from the formation but remain within a controllable range. Exceeding this threshold requires heading adjustments. These thresholds were calibrated through simulation experiments, and the values ​​given in this embodiment are preferred values ​​for a typical quadcopter UAV in diamond formation mode.

[0098] The desired attitude adjustment amount serves as an auxiliary adjustment parameter, simultaneously adjusting the UAV's roll and pitch angles during the aforementioned adjustment process to match them with the desired flight direction and speed, ensuring a smooth transition in flight attitude. The attitude adjustment amount is calculated inversely from the desired speed and heading using the UAV's dynamics model.

[0099] The proportional coefficient, differential coefficient, deviation threshold, and basic regression velocity are control parameters that are pre-calibrated based on the dynamic characteristics of the UAV, formation response time requirements, and flight safety constraints, enabling rapid formation correction while ensuring stable flight.

[0100] 4. Generate flight parameter adjustment instructions based on the desired flight parameter adjustment amount, and send the flight parameter adjustment instructions to the affected UAVs.

[0101] Specifically, the determined target speed, target heading angle, target attitude angle, and other desired flight parameters are encapsulated in a command format recognizable by the UAV flight control system to generate flight parameter adjustment commands. These commands are transmitted to the affected UAVs via the data link between the ground station and the UAV. To ensure reliable command delivery, a confirmation and retransmission mechanism is employed: after sending the command, the ground station starts a timer; if no confirmation is received from the affected UAV within one second, the command is retransmitted, up to a maximum of three times. Upon receiving the command, the affected UAV's flight control system parses and executes the corresponding flight parameter adjustments, gradually returning it to its theoretical formation position. During the adjustment process, the ground station continuously monitors the positional deviation of the affected UAVs. If the deviation continues to decrease, the current adjustment method is maintained; if the deviation continues to increase or the adjustment is ineffective, the adjustment strategy is reassessed, and if necessary, the risk level is escalated and a higher-level response is triggered.

[0102] The third sub-step: When the formation error index exceeds the third preset threshold and the duration exceeds the third preset duration, and / or the rate of change of the formation error index exceeds the third preset rate of change threshold and the duration exceeds the third preset duration, execute the global avoidance trajectory planning strategy, call the trajectory planning algorithm to generate a smooth avoidance trajectory for the threatened UAV, and send the trajectory command to the corresponding UAV through the backup communication link.

[0103] In this embodiment, the third preset threshold is set to 0.8, the third preset duration is set to 1 second, and the third preset rate of change threshold is set to 0.3 per second. When the formation error index exceeds 0.8 and lasts for more than 1 second, or the rate of change of the formation error index exceeds 0.3 per second and lasts for more than 1 second, or both conditions are met simultaneously, it indicates that the cluster coordination state has become seriously abnormal. This may manifest as multiple drones simultaneously deviating significantly from the formation, large-scale interruption of communication links, a surge in the number of lost drones, or a high risk of an impending collision. In this state, local fine-tuning is no longer effective in restoring formation safety, and global emergency avoidance measures must be taken. Therefore, a global avoidance trajectory planning strategy is executed to plan a complete avoidance path from the current position to a safe target point for the threatened drones, and to ensure that the instructions are delivered through the most reliable communication channel to maximize the safety of the drones and the continuity of the cluster mission. Specifically, as follows:

[0104] 1. Obtain the current flight status parameters of the threatened drone and the location of the target avoidance point;

[0105] 2. Based on the current flight status parameters, the target avoidance point position, and the preset UAV dynamics constraints, generate an avoidance trajectory from the current position to the target avoidance point. The avoidance trajectory satisfies the smoothness constraints of velocity, acceleration, and jerk.

[0106] A threatened UAV refers to a UAV that, after the formation error index exceeds the limit, is determined by the ground station to be at risk of collision or about to go out of control, based on the real-time position deviation, communication status, and motion trend of each UAV. Current flight status parameters include the UAV's real-time position coordinates, current velocity vector, current acceleration vector, and current heading angle and attitude angle. The target avoidance point location refers to a preset safe position for the threatened UAV. In this embodiment, it is dynamically determined based on the UAV's current flight direction and the overall situation of the cluster: when there is an obstacle in front of the UAV or an adjacent UAV, a safe airspace point 50 meters to the side or rear of the current heading is selected as the avoidance point; when a UAV needs to return due to communication loss, the midpoint of the return path or a point in the airspace directly above the ground station is selected as the avoidance point. UAV dynamic constraints refer to the physical limitations that the UAV must meet during flight, including maximum flight speed, maximum acceleration, maximum deceleration, and maximum jerk. In this embodiment, the following parameters are preset according to the UAV model and flight safety specifications: maximum flight speed 20 m / s, maximum acceleration 5 m / s², maximum deceleration 4 m / s², and maximum jerk 2 m / s³. Smoothness constraints refer to the continuous absence of abrupt changes in the velocity curve, acceleration curve, and jerk curve within the generated trajectory. This ensures stable flight attitude for the UAV during emergency avoidance maneuvers, preventing stall or loss of control due to violent maneuvers.

[0107] In this embodiment, when generating the avoidance trajectory, the acceleration parameters for the acceleration phase and the acceleration parameters for the deceleration phase are independently set according to the current speed of the threatened UAV, the preset maximum acceleration, the preset maximum deceleration, and the position of the target avoidance point, so that the avoidance trajectory has different acceleration change rates in the acceleration and deceleration phases.

[0108] Specifically, jerk is the rate of change of acceleration, that is, the amount of change in acceleration per unit time. Independently setting jerk parameters for the acceleration and deceleration phases means that the drone can gradually increase its acceleration at one jerk value during acceleration and gradually decrease its acceleration at another jerk value during deceleration, thus creating different slopes in the acceleration curve during the acceleration and deceleration phases. For example, when a drone needs to accelerate rapidly to avoid an approaching drone from behind, a larger jerk value can be set for the acceleration phase, allowing the acceleration to quickly reach its maximum value; while when approaching the target avoidance point, to avoid overshoot, a smaller jerk value can be set for the deceleration phase, allowing the deceleration to increase gradually and ensuring the drone lands smoothly at the target point. This asymmetrical jerk setting allows the avoidance trajectory to flexibly adjust acceleration and deceleration characteristics according to the actual urgency and spatial constraints, achieving better avoidance results while ensuring safety. The specific generation process adopts a fifth-order polynomial trajectory planning algorithm, with the current position, current velocity, and acceleration as the starting constraints, and the target avoidance point position, expected velocity, and expected acceleration as the ending constraints. It also introduces the maximum jerk constraint and independent acceleration and deceleration phase jerk parameters. By solving the optimization problem, smooth trajectory parameters that satisfy all constraints are obtained.

[0109] In this embodiment, the trajectory command is sent to the corresponding UAV through a backup communication link, including the following:

[0110] 1. Monitor the communication quality of the main communication link. When the main communication link cannot meet the preset communication quality requirements, switch the transmission channel of the trajectory command to the pre-established backup communication link.

[0111] 2. Send trajectory instructions to the corresponding UAV through the backup communication link and receive the return confirmation information from the corresponding UAV. If no confirmation information is received within the preset time, resend the trajectory instructions.

[0112] Communication quality refers to the reliability and real-time performance of data transmission, which is measured in this embodiment by round-trip latency and packet loss rate. The preset communication quality requirements are a round-trip latency of less than 200 milliseconds and a packet loss rate of less than 5%. The ground station continuously sends periodic probe packets to the UAV and tracks the response to assess the current communication quality of the main communication link in real time. If the above requirements are not met for three consecutive samplings, the main communication link is deemed unable to meet the preset communication quality requirements, and a link switching procedure is immediately initiated. The backup communication link refers to an independent communication channel different from the main communication link. In this embodiment, a data radio is used as the backup link. This link has strong anti-interference capabilities and stable transmission, but its bandwidth is low, and it is only used for transmitting critical commands. The backup link is pre-established and maintains a heartbeat connection during system initialization to ensure it can take over command transmission at any time.

[0113] Specifically, the ground station encapsulates the generated avoidance trajectory parameters into a trajectory command packet in a compact binary format and sends it to the target UAV via a backup communication link. A timer starts immediately after transmission, with a preset timeout retransmission interval of 0.5 seconds. If an acknowledgment is received from the UAV via the backup link within 0.5 seconds, the command is considered successfully delivered. If no acknowledgment is received within the timeout period, the command is considered likely lost, and the same command packet is immediately retransmitted, up to a maximum of 5 times. If no acknowledgment is received after 5 retransmissions, the UAV is deemed to have completely lost contact, the ground station records the event, and initiates other emergency procedures. Simultaneously, the ground station continuously monitors the quality of the primary communication link. Once the primary link recovers to meet communication quality requirements, it automatically switches subsequent commands back to the primary link to free up backup link resources. This dual-link redundancy plus acknowledgment retransmission mechanism ensures that critical avoidance commands can still be delivered to the target UAV with the highest probability, even in the worst communication environments, maximizing flight safety.

[0114] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for collaborative risk early warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform, characterized in that, include: The monitoring agent is deployed in a containerized manner on a cloud-native platform, and the monitoring agent collects the collaborative status data of the drone cluster in real time. Based on the cooperative state data, a formation error index is constructed to quantify the degree of abnormality in the cooperative state of the UAV swarm; the formation error index is constructed according to the following steps: Based on the real-time position coordinates of each drone and the preset theoretical formation position coordinates, the position deviation of each drone is calculated, and the average position deviation of the cluster is obtained based on the statistical analysis of the position deviations of all drones in the cluster. Based on the heartbeat signal reported by each drone, count the number of drones that are currently out of contact and calculate the proportion of the number of drones that are out of contact relative to the total number of drones in the cluster. Based on the round-trip communication delay between UAVs, the deviation between the average communication delay of the cluster and the preset benchmark communication delay is calculated, and then normalized to obtain the communication delay deviation. Calculate the rate of change of the cluster's average location deviation, disconnection rate, and communication delay deviation at the current moment; The preliminary index is obtained by weighted summing of the cluster's average location deviation, disconnection rate, communication delay deviation, and their respective rates of change. The preliminary index is transformed through a nonlinear mapping function to obtain the final formation error index; The system monitors the value and rate of change of the formation error index in real time, and presets multiple different trigger conditions, each corresponding to a warning response strategy; wherein: When the formation error index exceeds a first preset threshold and the duration exceeds a first preset duration, and / or the rate of change of the formation error index exceeds a first preset rate of change threshold and the duration exceeds a first preset duration, a communication link optimization strategy is executed to ensure data transmission by dynamically switching to a relay communication topology composed of neighboring UAVs; the method of ensuring data transmission by dynamically switching to a relay communication topology composed of neighboring UAVs includes: Identify target drones whose direct link quality to the ground station is degraded; The current status parameters of multiple neighboring drones within a preset range around the target drone are obtained. The status parameters include location coordinates, remaining battery power, current communication load, and link quality with the target drone. Calculate the path quality score for each candidate communication path with each neighboring UAV as a relay node based on the state parameters. The candidate communication path with the highest path quality score is selected as the target relay communication topology, and the data transmission link of the target UAV is switched to the target relay communication topology. When the formation error index exceeds the second preset threshold and the duration exceeds the second preset duration, and / or the rate of change of the formation error index exceeds the second preset rate of change threshold and the duration exceeds the second preset duration, a local trajectory adjustment strategy is executed, and flight parameter adjustment instructions are issued to the affected UAVs to correct the formation. When the formation error index exceeds the third preset threshold and the duration exceeds the third preset duration, and / or the rate of change of the formation error index exceeds the third preset rate of change threshold and the duration exceeds the third preset duration, a global avoidance trajectory planning strategy is executed, a trajectory planning algorithm is called to generate a smooth avoidance trajectory for the threatened UAV, and the trajectory command is sent to the corresponding UAV through a backup communication link.

2. The method for collaborative risk early warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 1, characterized in that, The collaborative status data includes the round-trip communication delay between UAVs, the heartbeat signal reporting status of each UAV, the real-time position coordinates of each UAV, and the collaborative control command data between UAVs.

3. The method for collaborative risk warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 1, characterized in that, The preset thresholds corresponding to the multiple different triggering conditions satisfy the following: first preset threshold < second preset threshold < third preset threshold; the corresponding preset durations are associated with each preset threshold; and the corresponding preset rate of change thresholds satisfy the following: first preset rate of change threshold < second preset rate of change threshold < third preset rate of change threshold.

4. The method for collaborative risk warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 1, characterized in that, The step of issuing flight parameter adjustment commands to the affected drones to correct their formation includes: Determine the affected drones whose formation needs to be adjusted and their theoretical formation positions; Obtain the position deviation between the current position of the affected UAV and the theoretical formation position, as well as the rate of change of the position deviation at the current moment; Based on the position deviation and the rate of change of the position deviation, the expected flight parameter adjustment amount for the affected UAV is determined; A flight parameter adjustment command is generated based on the desired flight parameter adjustment amount, and the flight parameter adjustment command is sent to the affected UAV.

5. The method for collaborative risk warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 4, characterized in that, The desired flight parameter adjustment includes at least one of the desired speed adjustment, desired heading adjustment, and desired attitude adjustment.

6. The method for collaborative risk early warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 1, characterized in that, The trajectory planning algorithm generates a smooth avoidance trajectory for the threatened drone, including: Obtain the current flight status parameters of the threatened drone and the location of the target avoidance point; Based on the current flight state parameters, the target avoidance point position, and preset UAV dynamics constraints, an avoidance trajectory from the current position to the target avoidance point is generated. The avoidance trajectory satisfies the smoothness constraints of velocity, acceleration, and jerk.

7. The method for collaborative risk warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 6, characterized in that, When generating the avoidance trajectory, the acceleration parameters for the acceleration phase and the acceleration parameters for the deceleration phase are independently set according to the current speed of the threatened UAV, the preset maximum acceleration, the preset maximum deceleration, and the position of the target avoidance point, so that the avoidance trajectory has different acceleration change rates in the acceleration and deceleration phases.

8. The method for collaborative risk warning of unmanned aerial vehicle (UAV) swarms based on a cloud-native platform according to claim 1, characterized in that, The step of sending trajectory commands to the corresponding UAV via a backup communication link includes: Monitor the communication quality of the main communication link. When the main communication link fails to meet the preset communication quality requirements, switch the transmission channel of the trajectory command to a pre-established backup communication link. The trajectory command is sent to the corresponding UAV through the backup communication link, and the return confirmation information from the corresponding UAV is received. If the confirmation information is not received within a preset time, the trajectory command is resent.

Images