Cooperative relative positioning method and system for a cluster of aerial vehicles

Abstract

This invention relates to the field of collaborative positioning technology for aircraft swarms, specifically a collaborative relative positioning method and system for aircraft swarms. The method includes: optimizing the scheduling of ranging and communication resources within the swarm by dynamically assigning appropriate roles and establishing a hierarchical ranging topology; collecting raw ranging and exchange data based on role assignment results, and constructing a relative position constraint network after consistency correction; iteratively solving this network to obtain a globally consistent relative position estimate, and using this estimate to fuse and correct the local navigation state of each aircraft, generating a local navigation correction value. This method achieves tight coupling from collaborative solution to individual navigation, effectively improving positioning accuracy and consistency of the swarm's internal state, while enhancing system scalability and robustness through a dynamic hierarchical structure.

Description

Technical Field

[0001] This invention relates to the field of collaborative positioning technology for aircraft swarms, and more particularly to a collaborative relative positioning method and system for aircraft swarms. Background Technology

[0002] In collaborative operations among aircraft swarms, accurate and reliable relative positioning is fundamental for achieving tight formation, collaborative sensing, and mission execution. Existing collaborative relative positioning technologies largely rely on pre-defined fixed communication topologies or fully peer-to-peer self-organizing networks. Fixed topologies lack adaptability to dynamic tasks and environments, while using fully peer-to-peer data exchange and ranging in large-scale swarms can lead to communication link congestion, ranging conflicts, and uneven consumption of computing resources. Current solutions typically loosely couple or fusion the relative positioning calculation with the aircraft's own inertial navigation and satellite navigation systems. The relative positioning results are primarily used for attitude output or trajectory evaluation, failing to form a closed loop that can directly and in real-time correct the local navigation states of each aircraft. This separate processing approach is prone to accumulating positioning errors and inconsistencies in relative states within the swarm as its size increases or in highly maneuverable scenarios, affecting the overall accuracy and robustness of collaborative operations.

[0003] A method is needed to intelligently organize collaborative sensing and data exchange based on the real-time status of the cluster, in order to optimize resource utilization and improve system scalability. Simultaneously, a mechanism is needed to deeply integrate the relative position information calculated collaboratively by the cluster into the local navigation filtering loops of each aircraft, achieving tight coupling from relative observation to absolute state correction. This effectively suppresses error divergence and ensures the consistency, accuracy, and real-time performance of relative positioning for large-scale clusters in highly dynamic environments. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose a cooperative relative positioning method and system for aircraft swarms.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a cooperative relative positioning method for an aircraft swarm, comprising:

[0006] Obtain the individual status information and environmental constraint information of all aircraft in the cluster, and assign appropriate roles to each aircraft to form role assignment results;

[0007] Based on the role assignment results, a hierarchical ranging and data exchange topology is established within the cluster.

[0008] By utilizing the hierarchical ranging and data exchange topology, the relative ranging and communication processes between aircraft are scheduled, and raw ranging observation data and information exchange data are collected.

[0009] The collected raw ranging observation data is corrected for consistency error, and a corrected set of relative distances between all pairs of aircraft in the cluster is generated.

[0010] Based on the modified set of relative distances and information exchange data, a relative position constraint network for the aircraft cluster is constructed.

[0011] By solving the relative position constraint network, the estimated relative position of each aircraft in the cluster in a common reference frame is calculated iteratively.

[0012] The aircraft's navigation state is fused and corrected based on the calculated relative position estimate to generate a local navigation correction.

[0013] The local navigation correction is used to update the aircraft's own positioning information and generate role and mission suggestions for the next round of cooperative positioning.

[0014] By integrating key data from this round of positioning, the role assignment results, the topology, and the final relative position estimate, a collaborative positioning process report is generated.

[0015] As a further aspect of the present invention, a suitable role is assigned to each aircraft to form a role assignment result, including:

[0016] Extract the positioning sensor accuracy level, communication link bandwidth, remaining energy level, and flight stability indicators of each aircraft from the individual status information;

[0017] The environmental constraint information is used to identify the area to be covered by the cluster, the location information of known landmarks, and the distribution of communication interference sources.

[0018] Based on the accuracy level of the positioning sensor, the aircraft is initially classified into core reference node candidates and ordinary nodes;

[0019] Based on the communication link bandwidth and the remaining energy level, aircraft with stable data relay capabilities are selected from the core reference node candidates and designated as core reference nodes.

[0020] Based on the flight stability index and the area to be covered by the cluster, a relatively fixed monitoring sub-area is assigned to each ordinary node;

[0021] Based on the location information of the known landmarks, designate several aircraft closest to the landmarks as landmark auxiliary nodes;

[0022] The assignment information of core reference nodes, landmark auxiliary nodes, and ordinary nodes assigned to monitoring sub-regions is integrated to form a structured role assignment result, which includes a unique role identifier for each aircraft and its associated mission parameters.

[0023] As a further aspect of the present invention, based on the role assignment results, a hierarchical ranging and data exchange topology is established within the cluster, including:

[0024] Set the core reference node as the top-level root node of the topology;

[0025] Centered on each core reference node, a batch of ordinary nodes and landmark auxiliary nodes are dynamically associated based on its communication coverage capability and the bandwidth of the communication link, forming multiple star-shaped subnets with the core reference node as the parent node;

[0026] Within each star subnet, auxiliary communication links are allowed to be established between associated ordinary nodes based on the proximity relationship of the monitored sub-regions;

[0027] Establish dedicated data links between landmark auxiliary nodes for synchronizing landmark observation data;

[0028] Information flow rules are set: ranging data and status information of ordinary nodes are first aggregated to the core reference node of their respective star subnet; global data exchange is carried out between core reference nodes; observation data of landmark auxiliary nodes are shared through dedicated data links and then distributed to the associated core reference nodes;

[0029] Based on the core reference node relationships and information flow rules, a complete topology description including hierarchical relationships, link connection relationships, and data flow directions is constructed, namely the layered ranging and data exchange topology structure.

[0030] As a further aspect of the present invention, the hierarchical ranging and data exchange topology is utilized to schedule the relative ranging and communication processes between aircraft, and to collect raw ranging observation data and information exchange data, including:

[0031] Based on the division of the star subnets in the topology, an independent ranging time slot is assigned to each star subnet;

[0032] Within the allocated ranging time slot, the core reference node of the star subnet sends ranging request signals to all its associated ordinary nodes and landmark auxiliary nodes.

[0033] Upon receiving a ranging request signal, the associated node replies with a response signal containing its own identity identifier and timestamp;

[0034] The core reference node calculates the original two-way ranging observation value with the core reference node based on the time difference between sending a request and receiving a response;

[0035] All star-shaped subnets sequentially complete their internal ranging processes, collecting raw ranging observation data between all core reference nodes and their associated nodes;

[0036] During the preset communication time slots, the core reference nodes exchange their collected ranging observation data and their own navigation status information to form global information exchange data;

[0037] Landmark auxiliary nodes exchange their observation information of known landmarks on a dedicated data link;

[0038] The raw ranging observation data, global information exchange data, and landmark observation information from all star-shaped subnets are aggregated to form a complete raw observation and information dataset.

[0039] As a further aspect of the present invention, the collected raw ranging observation data is corrected for consistency errors to generate a corrected set of relative distances between all pairs of aircraft within the cluster, including:

[0040] Identify redundant measurements in the original ranging observation data: due to multiple ranging paths, either directly or indirectly, between core reference nodes and some ordinary nodes.

[0041] Using the core reference node as the benchmark, an error state equation is established regarding the relative distance between all aircraft pairs, and the original ranging observations are used as noisy observations.

[0042] Aircraft velocity information from global information exchange data is introduced as a constraint on the process model, and the absolute position reference of some aircraft in the landmark observation information is used to provide anchor points for the error state equation.

[0043] An optimization estimation algorithm based on a sliding window is used to iteratively solve the error state equation. Under the conditions of satisfying all direct measurements, velocity constraints and absolute position references, the total error of the relative distance network of the entire cluster is minimized.

[0044] The systematic error correction for all original ranging observations obtained after optimization is applied to the corresponding original data.

[0045] The output is a set of theoretically reachable relative distances that, after applying corrections, are more consistent and cover all aircraft pairs within the cluster.

[0046] As a further aspect of the present invention, based on the modified relative distance set and information exchange data, a relative position constraint network for the aircraft cluster is constructed, including:

[0047] Each aircraft is abstracted as a node in the relative position constraint network;

[0048] The relative distance between each pair of aircraft in the modified relative distance set is abstracted as an edge connecting the corresponding two nodes, with the edge carrying a distance value as a constraint weight.

[0049] From the information exchange data, the heading angle difference information exchanged between the core reference nodes is extracted, and the heading angle difference information is converted into soft constraints on the relative azimuth angles between aircraft, which are then used as angle constraint weights attached to the corresponding edges.

[0050] From the landmark observation information, the absolute position information of the known landmark points is converted into position anchor point constraints for the aircraft, which serves as landmark auxiliary nodes. The position anchor point constraints are used as absolute position weights applied to the corresponding nodes.

[0051] By integrating all nodes, edges with distance weights, edges with angle weights, and absolute position weights on the nodes, a graph model containing multiple constraint types is constructed, namely the relative position constraint network.

[0052] As a further aspect of the present invention, by solving the relative position constraint network, the estimated relative position of each aircraft in the cluster in a common reference frame is iteratively calculated, including:

[0053] Initialization steps: Assign an initial relative position coordinate to each node in the relative position constraint network, wherein nodes with absolute position weights are initialized with their anchor point positions, and other nodes are roughly geometrically initialized according to their distance constraints with neighboring nodes;

[0054] Establish the objective function: The objective function consists of three parts. The first part is the sum of squares of the differences between the calculated distance constraint values ​​and the actual distance weights of all edges in the network. The second part is the sum of squares of the differences between the calculated azimuth values ​​and the angle weights of all edges with angular constraints. The third part is the sum of squares of the differences between the calculated position values ​​of nodes with absolute position weights and the anchor positions.

[0055] A nonlinear least squares optimization algorithm is used to iteratively minimize the objective function, with the position coordinates of all nodes as the optimization variables.

[0056] In each iteration, the gradient of the objective function with respect to each node coordinate is calculated, and the node coordinates are updated along the gradient descent direction.

[0057] Set an iteration convergence condition: stop iterating when the change in the objective function value is less than a preset threshold or when the maximum number of iterations is reached.

[0058] The coordinates of each node obtained after the final iterative optimization are used as the relative position estimate of the corresponding aircraft in the common reference frame.

[0059] As a further aspect of the present invention, the navigation state of the aircraft itself is fused and corrected based on the calculated relative position estimate to generate a local navigation correction, including:

[0060] Each aircraft extracts its own relative position estimate from the set of relative position estimates;

[0061] The relative position estimate is compared with the local position estimate provided by the aircraft's own onboard navigation system, and the position difference between the relative position estimate and the local position estimate in the horizontal and vertical directions is calculated.

[0062] Obtain the current velocity estimate and attitude information of the aircraft;

[0063] Based on the position difference, velocity estimation, and attitude information, a kinematic coupling model is used to deduce the potential navigation error sources that cause the position difference. The potential navigation error sources include position drift, velocity deviation, and heading deviation.

[0064] Based on the type and magnitude of the error sources, and in conjunction with the priority of the current flight mission, a set of correction values ​​is generated;

[0065] The correction amounts include position correction compensation values ​​for the inertial navigation system, correction coefficients for the velocity integral, and fine-tuning amounts for the heading reference.

[0066] The correction values ​​are packaged to form the local navigation correction values ​​applied to the aircraft's own navigation system.

[0067] As a further aspect of the present invention, the local navigation correction is used to update the aircraft's own positioning information, and role and task suggestions for the next round of cooperative positioning are generated, including:

[0068] The aircraft receives the local navigation correction and inputs it into the onboard navigation filter for fusion;

[0069] The navigation filter updates the system state vector based on the correction amount, thereby correcting the aircraft's current position, velocity, and attitude estimation, and obtaining updated positioning information.

[0070] The confidence level of the positioning information output by the aircraft's own navigation system was evaluated after the local navigation correction was applied.

[0071] Based on the updated positioning information, the change in confidence level, and other aircraft status changes obtained from the information exchange data, the positioning sensor accuracy level and flight stability index in the individual status information are reassessed.

[0072] Based on the reassessed individual state information and the dynamic changes in the environmental constraint information, the demand for the next round of cooperative localization is predicted.

[0073] Role suggestions: Based on the predicted needs, suggest whether it is necessary to replace the core reference node, adjust the landmark auxiliary node, or reallocate the monitoring sub-areas of ordinary nodes;

[0074] The task suggestions are generated as follows: for core reference nodes, suggestions are made regarding the list of associated nodes for the next round and the timing of ranging scheduling; for ordinary nodes, suggestions are made regarding the key monitoring areas and the priority of auxiliary ranging with neighboring nodes.

[0075] The proposed roles and tasks are encapsulated and included as part of the output.

[0076] Integrating key data from this round of positioning, the role assignment results, the topology, and the final relative position estimate, a cooperative positioning process report is generated, including:

[0077] Create a report header that includes a timestamp and cluster identifier;

[0078] The report body first records the initial conditions at the start of this round of cooperative localization, including the event that triggered cooperative localization, the initial individual state information summary, and the environmental constraint information summary;

[0079] Record the complete details of the role assignment results, including the initial role and mission parameters for each aircraft;

[0080] The established hierarchical ranging and data exchange topology is recorded in the form of a graphical description or an adjacency matrix.

[0081] Record an overview of the raw data collected, including the total amount of distance measurement observation data, the scale of information exchange data, and the amount of landmark observation information;

[0082] Record a summary of the construction of the relative position constraint network, including the number of nodes, the number of edges, and the distribution of constraint types;

[0083] Record the convergence status of the iterative optimization process for solving the relative position constraint network, including the initial objective function value, the final objective function value, and the number of iterations;

[0084] List the final relative position estimates in a table, including the number of each aircraft, its coordinate estimates in a common reference frame, and the covariance information of the relative position estimates;

[0085] Record an overview of the generated local navigation corrections, and compare the updated positioning information with the previous cycle.

[0086] Attached are the suggested roles and tasks for the next round of generation;

[0087] All recorded content is organized in logical order, and necessary indexes and annotations are added to form a structured collaborative localization process report.

[0088] As a further aspect of the present invention, the present invention also includes a cooperative relative positioning system for an aircraft swarm, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the computer program, implements the steps of the cooperative relative positioning method for an aircraft swarm as described above.

[0089] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0090] This method changes the organizational model of traditional fixed or fully peer-to-peer networks by dynamically assigning appropriate roles to aircraft within the cluster and constructing a hierarchical ranging and data exchange topology based on role assignment results. This dynamic hierarchical topology clarifies the ranging and communication scheduling relationships between nodes at different levels, enabling structurally optimized allocation of ranging tasks and data flows. It avoids signal interference and resource contention caused by fully interconnected ranging in large-scale clusters, reducing unnecessary communication overhead. The topology can adaptively change with role adjustments, thus more rationally allocating sensing and communication loads within the cluster in complex task scenarios, enhancing the system's robustness to node failures or task changes, and supporting collaborative localization in larger-scale clusters.

[0091] By integrating all corrected relative distance observations and exchanged navigation information, a unified relative position constraint network is constructed and iteratively solved, yielding a globally consistent relative position estimate. This estimate is then deeply fused and corrected with information from each aircraft's own inertial navigation and satellite navigation systems, directly generating state corrections for the local navigation system. This process achieves tight coupling between cooperative relative positioning and individual autonomous navigation. Injecting the results of the cluster's collaborative solution directly into the local navigation filtering update process effectively calibrates errors in individual navigation sensors, particularly inertial navigation drift. This improves the absolute accuracy of each aircraft's self-positioning, ensures the consistency of relative position estimates among all members within the cluster, and suppresses the propagation and amplification of errors within the cooperative network. Attached Figure Description

[0092] Figure 1This is a flowchart of a cooperative relative positioning method for an aircraft swarm as described in this invention;

[0093] Figure 2 Flowchart for assigning roles to aircraft;

[0094] Figure 3 A flowchart for the scheduling of ranging and communication processes and the collection of raw data;

[0095] Figure 4 A graph for evaluating the data exchange volume and routing efficiency of hierarchical topology subnets;

[0096] Figure 5 This is a topological diagram showing the relative positions of nodes and their roles. Detailed Implementation

[0097] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0098] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0099] See Figure 1The system acquires individual state information and environmental constraint information of all aircraft within the cluster, assigns appropriate roles to each aircraft, and forms role assignment results. Based on these role assignment results, a hierarchical ranging and data exchange topology is established within the cluster. Using this hierarchical ranging and data exchange topology, the system schedules the relative ranging and communication processes between aircraft, collecting raw ranging observation data and information exchange data. Consistency error correction is applied to the collected raw ranging observation data to generate a corrected set of relative distances between all pairs of aircraft within the cluster. Based on the corrected set of relative distances and information exchange data, a relative position constraint network for the aircraft cluster is constructed. By solving the relative position constraint network, the relative position estimate of each aircraft within the cluster in a common reference frame is iteratively calculated. The navigation state of each aircraft is fused and corrected based on the calculated relative position estimate to generate a local navigation correction. The local navigation correction is used to update the aircraft's own positioning information and generate role and task suggestions for the next round of cooperative positioning. Finally, the system integrates key data, role assignment results, topology, and the final relative position estimate from this round of positioning to generate a cooperative positioning process report.

[0100] See Figure 2 In one embodiment of the present invention, it is assumed that a flight swarm consisting of 12 small unmanned aerial vehicles (UAVs) performs a collaborative mapping task within a rectangular mission area. Each UAV in the swarm is equipped with a satellite navigation receiver, inertial navigation system (INS), ultra-wideband ranging module, and data link equipment. The master control node for collaborative positioning collects individual state information and environmental constraint information for all 12 UAVs. The individual state information includes the positioning sensor accuracy level of each UAV's navigation sensor, which is jointly evaluated by the signal-to-noise ratio of the satellite navigation signal and the geometric accuracy factor of the positioning solution. The individual state information also includes the communication link bandwidth, which is determined by the actual available transmission rate of the current data link and the channel quality index. The individual state information includes the remaining energy level, which is represented by the current battery percentage and the estimated duration of the mission. The individual state information includes flight stability indicators, which are calculated by combining the current flight speed variance, attitude angle fluctuation amplitude, and past maneuver history.

[0101] In some embodiments, environmental constraint information is jointly provided by the mission planning system and airborne environmental perception sensors. The area to be covered by the cluster is defined by preset mapping boundary coordinates. The location information of known landmarks comes from three pre-calibrated radio frequency beacons in a digital map loaded before the mission. The distribution of communication interference sources is constituted by the interference signal strength and azimuth information detected and identified in real time by airborne spectrum monitoring equipment. In a specific implementation, based on the positioning sensor accuracy level extracted from individual state information, the system performs a preliminary classification operation. The three UAVs with the highest positioning sensor accuracy levels are marked as core reference node candidates, and the remaining nine UAVs are classified as ordinary nodes. Combining communication link bandwidth and remaining energy level, a selection is made from the three core reference node candidates. The communication link bandwidth must meet the threshold value for continuous relay data, and the remaining energy level must be higher than the threshold for maintaining the current mission intensity for one hour. After one candidate UAV is eliminated because its remaining energy level is close to the threshold, the remaining two UAVs that meet the conditions are designated as core reference nodes for this round of cooperative positioning.

[0102] Optionally, based on flight stability indicators and the area the cluster needs to cover, nine ordinary nodes are assigned relatively fixed monitoring sub-regions, dividing the entire rectangular task area into nine equal grid sub-regions. Ordinary nodes with higher flight stability indicators are assigned to sub-regions at the edge of the region or near potential interference sources, while ordinary nodes with lower flight stability indicators are assigned to sub-regions at the center of the region. Based on the location information of known landmarks, the system calculates the geometric distance between each ordinary node and three known landmarks, designating the ordinary node closest to each landmark as a landmark auxiliary node. In one example, the drone with ordinary node number 5 is closest to the first landmark and is designated as the landmark auxiliary node for the first landmark. The drones with ordinary node number 2 and number 8 are designated as landmark auxiliary nodes for the second and third landmarks, respectively.

[0103] It is understandable that the assignment information of core reference nodes, landmark auxiliary nodes, and ordinary nodes assigned monitoring sub-areas is integrated to form a structured role assignment result. This role assignment result is stored and distributed in the form of a data structure table, which contains a unique identifier for each UAV, a role identifier field, and associated task parameter fields. The role identifier field can contain values ​​for "core reference node," "landmark auxiliary node," and "ordinary node." The task parameter field of the core reference node contains the initial list of ordinary nodes it is responsible for associating with; the task parameter field of the landmark auxiliary node contains the known landmark point numbers it is associated with and the preset observation instructions; the task parameter field of the ordinary node contains the boundary coordinates of its assigned monitoring sub-area. This role assignment result serves as the basis for scheduling aircraft behavior throughout the entire cooperative positioning cycle.

[0104] In practice, the role assignment process can be implemented using a quantified decision function, which is used for the final selection from core reference node candidates. The decision function is defined as follows:

[0105]

[0106] Where: characters Indicates the filter score; character , , These are preset weighting coefficients, corresponding to the weights of sensor accuracy, communication capability, and energy level, respectively; characters Represents the normalized accuracy level of the positioning sensor; character Represents a specific numerical value for the bandwidth of the communication link; character This represents the normalized residual energy level. The system calculates the screening score for each core reference node candidate and selects the top-scoring nodes as the final core reference nodes.

[0107] See Figure 3 In one embodiment of the present invention, it is assumed that the cluster comprises twelve UAVs, of which two are core reference nodes, three are landmark auxiliary nodes, and seven are ordinary nodes. In a specific implementation, the two core reference nodes are set as the top-level root nodes of a hierarchical ranging and data exchange topology, labeled as root node Alpha and root node Beta, respectively. Centered on root node Alpha, four ordinary nodes and one landmark auxiliary node are dynamically associated based on the communication coverage capability of its onboard communication equipment and the actual communication link bandwidth reported in its individual status information, forming the first star-shaped subnet. Centered on root node Beta, the remaining three ordinary nodes and two landmark auxiliary nodes are associated based on its communication coverage capability and communication link bandwidth, forming the second star-shaped subnet. Within each star-shaped subnet, the associated ordinary nodes are allowed to establish auxiliary communication links based on the proximity relationship of the monitoring sub-areas allocated in the embodiment. For example, in the first star-shaped subnet, a point-to-point communication link is established between two ordinary nodes in adjacent monitoring sub-areas for status synchronization. Between landmark auxiliary nodes, a dedicated data link is established for synchronizing landmark observation data according to mission requirements. This dedicated data link uses a different frequency band or time-division channel than the main communication link of the star subnet.

[0108] In some embodiments, information flow rules are defined for a hierarchical ranging and data exchange topology. These rules stipulate that ranging data and status information from ordinary nodes must first be aggregated to the core reference node of their respective star-shaped subnet. Core reference nodes exchange data globally during global synchronization time slots, exchanging information including their collected subnet ranging data, position, velocity, and attitude information. Landmark auxiliary nodes' observation data are first shared via dedicated data links between them, ensuring all auxiliary nodes obtain complete information on observations of multiple known landmarks. Subsequently, the auxiliary nodes distribute the integrated landmark observation information to their associated core reference nodes. Based on the core reference node relationships and information flow rules, a complete topology description is constructed, including hierarchical relationships, link connections, and data flow directions. This topology description can be represented as a directed weighted graph, where vertices represent aircraft, edges represent pre-defined communication and ranging links, and edge weights can include link type and data flow direction. This graph structure constitutes the hierarchical ranging and data exchange topology.

[0109] Optionally, based on the star-shaped subnets in the hierarchical ranging and data exchange topology, an independent ranging time slot is allocated to each star-shaped subnet. The system adopts a time-division multiple access (TDMA) approach, dividing one communication frame into two consecutive ranging time slots and one subsequent communication time slot. Within the ranging time slot allocated to the first star-shaped subnet, the root node Alpha broadcasts a ranging request signal to all its four associated ordinary nodes and one landmark auxiliary node. This ranging request signal contains the root node Alpha's identifier and a precise sending timestamp. Upon receiving the ranging request signal, the four associated ordinary nodes and one landmark auxiliary node record the receiving time locally and immediately reply with a response signal containing their own identifier and receiving timestamp. After receiving the response signals from each node, the root node Alpha records the receiving time and, based on the time difference between sending the request signal and receiving each response signal, and considering the signal propagation speed, calculates the original bidirectional ranging observation value with each associated node. Within the ranging time slot allocated to the second star subnet, the root node Beta performs the same operation, obtaining raw two-way ranging observations between itself and its three associated ordinary nodes and two landmark auxiliary nodes. After all star subnets complete their internal ranging processes in sequence, the cluster collects raw ranging observation data between all core reference nodes and their associated nodes. This data is stored in the form of triplets of "transmitter node ID - receiver node ID - observed distance value".

[0110] Understandably, within a pre-defined communication time slot, the two core reference nodes, root nodes Alpha and Beta, exchange their collected ranging observation data and navigation status information via a dedicated control channel. The navigation status information includes latitude, longitude, elevation coordinates, velocity vectors, and attitude angles. This exchanged data forms the global information exchange data. Simultaneously, within a specific sub-time slot within the communication time slot, the three landmark auxiliary nodes exchange their observation information on known landmarks over a previously established dedicated data link. This observation information includes the landmark's identifier, the observed relative azimuth and elevation angles, and the observation time. The raw ranging observation data from all star-shaped subnets, the global information exchange data between core reference nodes, and the landmark observation information exchanged between landmark auxiliary nodes constitute a complete raw observation and information dataset, providing input for subsequent consistency error correction and relative position network construction.

[0111] In practical implementation, the relationships between nodes within a star subnet can be dynamically adjusted using a connectivity decision function. This function evaluates the potential connection quality between ordinary nodes and multiple core reference nodes. The connectivity decision function is defined as follows:

[0112]

[0113] Where: characters Indicates the connection quality assessment value; character Represents the received power of the reference signal from the core reference node received from a normal node; character Represents the geometric distance between ordinary nodes and the core reference node estimated based on coarse location information; character For a very small positive number that prevents the denominator from being zero; character Represents the currently available communication link bandwidth of the core reference node; character The system sets a baseline bandwidth value to be used to complete a ranging and data transmission operation. Before each topology construction, the system calculates the connection quality assessment value between ordinary nodes and each core reference node, and associates ordinary nodes with the star subnet containing the core reference node with the highest connection quality assessment value. This dynamic association mechanism enhances the adaptability of the topology to changes in the communication environment.

[0114] In one embodiment of the invention, redundant measurements are identified in the original ranging observation data. These redundant measurements arise from multiple ranging paths, either directly or indirectly, between core reference nodes and between some ordinary nodes. In the example, there is a direct ranging data exchange between the root nodes Alpha and Beta of the two core reference nodes, forming a direct measurement path. Simultaneously, ranging observations exist between root node Alpha and ordinary node A, and between root node Beta and ordinary node A, creating an indirect ranging path from ordinary node A to root node Beta via root node Alpha. The length of this indirect path equals the distance between ordinary node A and root node Alpha plus the distance between root node Alpha and root node Beta, constituting a redundant comparison with the direct ranging observation between ordinary node A and root node Beta. Using the core reference nodes as the reference point, an error state equation is established regarding the relative distances between all aircraft pairs. The state variables include the clock drift error and ranging system bias of each aircraft at multiple moments within the sliding window. The original ranging observations are substituted into the error state equation as noisy observations.

[0115] In some embodiments, aircraft velocity information from global information exchange data is introduced as a constraint on the process model. This global information exchange data includes velocity vectors reported by root node Alpha and root node Beta. The aircraft velocity information is used to constrain the smooth change of aircraft position between adjacent time points within a sliding window. Using absolute position references for some aircraft from landmark observation information, anchor points are provided for the error state equation. In the example scenario, three landmark auxiliary nodes each observe a known landmark point. The precise geodetic coordinates of the known landmark point are known. Combining the attitude and observation angle of the landmark auxiliary node itself, an observation estimate of the absolute position of the landmark auxiliary node can be calculated. This absolute position observation estimate is used as the position anchor point for the corresponding aircraft node in the error state equation. A sliding window-based optimization estimation algorithm is used to iteratively solve the error state equation. The optimization objective is to minimize the total error of the relative distance network of the entire cluster while satisfying all direct measurements, velocity constraints, and absolute position references. The total error is defined as the weighted sum of squares of all ranging residuals, velocity constraint residuals, and anchor point position residuals. The systematic error corrections obtained after optimization for all original ranging observations are applied to the corresponding original data. These corrections include fixed offset compensation and scaling factors for each ranging path. The output is a set of theoretically reachable relative distances, with higher consistency and covering all aircraft pairs within the cluster, obtained after applying the corrections. This set of relative distances is a symmetric matrix, with matrix elements... This represents the corrected relative distance estimate between aircraft i and aircraft j.

[0116] Optionally, each aircraft can be abstracted as a node in a relative position constraint network. In the example of a swarm of twelve drones, the relative position constraint network contains twelve nodes, with node numbers corresponding one-to-one with aircraft identifiers. The relative distance between each pair of aircraft in the modified relative distance set is abstracted as an edge connecting the corresponding two nodes, with the edge carrying a distance value as a constraint weight. For example, the constraint weight of the edge between node 3 and node 7 is the distance recorded in the modified relative distance set. Value. From the information exchange data, the heading angle difference information exchanged between the core reference nodes is extracted. Root node Alpha and root node Beta exchange their respective heading angles during communication, and the heading angle difference information is the difference between these two heading angles. The heading angle difference information is converted into a soft constraint on the relative azimuth angle between the aircraft. This relative azimuth angle soft constraint represents the estimated azimuth angle of the vector from root node Alpha to root node Beta on the horizontal plane. The relative azimuth angle soft constraint is used as an angle constraint weight added to the edge connecting root node Alpha and root node Beta.

[0117] It can be understood that the absolute position information of known landmarks is converted from landmark observation information into position anchor point constraints for aircraft acting as landmark auxiliary nodes. The absolute position information of the known landmarks is their three-dimensional coordinates in the global coordinate system. The conversion process utilizes the azimuth and pitch observations of the known landmarks by the landmark auxiliary nodes and the attitude data of the landmark auxiliary nodes themselves. Through coordinate transformation and trigonometric relationships, a set of position estimates of the landmark auxiliary nodes in the global coordinate system is calculated. This set of position estimates constitutes the position anchor point constraints of the corresponding nodes in the relative position constraint network. The position anchor point constraints are applied as absolute position weights to the corresponding nodes. The absolute position weights include the position estimates and their estimated covariance matrices. Integrating all twelve nodes, edges with distance weights, edges with angle weights, and the absolute position weights on the nodes, a graph model containing multiple constraint types is constructed. This graph model is the relative position constraint network of the aircraft cluster, where edges represent relative geometric relationship constraints between aircraft, and the weights on the nodes represent the direct position constraints of the aircraft in the global coordinate system.

[0118] In practical implementation, establishing and solving the error state equation involves modeling the measurement error, which can include systematic biases and random noise. An error correction model is defined to describe the k-th distance measurement observation. Its actual distance The relationship between them:

[0119]

[0120] Where: characters This represents the raw ranging observation between aircraft i and aircraft j at time k; character The true geometric distance between aircraft i and aircraft j is an unknown quantity; character This represents the proportional error factor affecting the distance measurement between aircraft i and aircraft j; this is an error state variable to be estimated. This represents the fixed deviation in distance measurement between aircraft i and aircraft j, which is also an error state variable to be estimated; (Character) Let represent the random noise of the k-th observation. Based on observations of all aircraft pairs at all times within the sliding window, an optimization algorithm is used to simultaneously estimate all... and Then use the estimated and For the original observation After making corrections, we obtain the corrected relative distance estimate. ,all This constitutes a modified set of relative distances.

[0121] In one embodiment of the present invention, the relative position constraint network constructed in the above embodiment is used as input. The relative position constraint network includes twelve nodes, several edges with distance weights, one edge with angle constraint weights, and three absolute position weights for the nodes. In a specific implementation, each node in the relative position constraint network is assigned an initial relative position coordinate. Nodes with absolute position weights are initialized with their anchor point positions. For example, nodes 5, 2, and 8 are used as landmark auxiliary nodes, and their initial coordinates are respectively set to position estimates calculated based on their respective observed landmark points. These position estimates are known three-dimensional coordinates in the global northeast-northeast coordinate system. The other nine nodes are roughly geometrically initialized according to their distance constraints with neighboring nodes. For node 1, it has distance constraints with node 5 and with node 3. The initial coordinate estimate of node 1 can be calculated using the known coordinates of node 5 and these two distance constraints through a simple triangulation method. The initial coordinates of other nodes are calculated sequentially through a similar polygonal positioning method.

[0122] In some embodiments, an objective function is established to optimize the final coordinates of all nodes. The objective function consists of three parts. The first part is the sum of squares of the differences between the calculated distance constraint values ​​and the actual distance weights of all edges in the network. For the edge connecting node i and node j, the calculated distance constraint value is the Euclidean distance calculated based on the current coordinates of the node. The actual distance weight is the corresponding value in the relative distance set. The first contribution is the sum of the squares of the differences between the calculated distance and the actual distance for all such edges. The second contribution is the sum of the squares of the differences between the calculated azimuth angle and the angle weight for all edges with angular constraints. In the example, only one edge has an angular constraint, namely the edge connecting the root node Alpha and the root node Beta. The calculated azimuth angle is the azimuth angle calculated based on the projected coordinates of the two nodes on the horizontal plane. The angle weight is the relative azimuth soft constraint value obtained from the heading angle difference information. The second part contributes to the calculation of the squared difference between the azimuth and angle weights. The third part is the sum of the squared differences between the calculated position values ​​of nodes with absolute position weights and the anchor point positions. For nodes 5, 2, and 8, their calculated position values ​​are their current coordinates, and the anchor point positions are the coordinates calculated based on landmark observations. The third contribution is the weighted sum of the squared differences between the coordinates of these three nodes and their respective anchor points, with the weights determined by the covariance matrix of the anchor points. The final objective function is the weighted sum of these three contributions:

[0123]

[0124] Where: characters Represents the overall objective function value, which is the variable to be optimized; character A vector representing the coordinates of all nodes; a character Represents the set of all edges with distance constraints; character Represents the confidence weight of the distance constraint on edge (i,j); character Represents the Euclidean distance between the coordinates of node i and node j; character Represents the corrected distance between aircraft i and j in the relative distance set; character The confidence weights representing angular constraints; characters This represents the azimuth angle calculated based on the coordinates of node A (root node Alpha) and node B (root node Beta); character Represents the relative azimuth soft constraint value obtained from information exchange data; character This represents a set of nodes with absolute anchor points, specifically three landmark auxiliary nodes; characters. The vector representing the current coordinates of node k; character Anchor point coordinate vector representing node k; character The weight matrix representing the anchor position constraint of node k is obtained by inverting the covariance matrix of the anchor position.

[0125] Optionally, a nonlinear least squares optimization algorithm is employed, using the three-dimensional position coordinates of all twelve nodes as optimization variables to iteratively minimize the aforementioned objective function. In each iteration, the gradient of the objective function relative to the coordinates of each node is calculated, and the node coordinates are updated along the gradient descent direction. The update step size is determined by the selected optimization algorithm. An iteration convergence condition is set: iteration stops when the change in the objective function value calculated in two consecutive iterations is less than a preset threshold, or when the number of optimization iterations reaches a preset maximum number of iterations. The coordinate values ​​of each node obtained after the final iteration optimization are used as the relative position estimates of the corresponding aircraft in a common reference frame, which adopts the Northeast-Northeast Geographic Coordinate System. Refer to Table 1 for examples comparing the relative position estimates of some aircraft with their initial coordinates after one optimization solution.

[0126] Table 1: Example Table Comparison of Coordinates of Some Aircraft Before and After Optimization

[0127] Node 1 (Alpha) 102.3 45.7 50.1 101.8 46.2 50.0 Node 2 205.0 150.0 52.0 204.9 150.2 52.1 Node 3 0.0 120.5 48.3 -0.5 121.0 48.5 Node 5 155.0 0.0 51.5 155.0 0.0 51.5

[0128] It can be understood that each aircraft extracts its own relative position estimate from the set of relative position estimates; for example, the optimized coordinates extracted by node 1 are (101.8, 46.2, 50.0). It compares its own relative position estimate with the local position estimate provided by its onboard navigation system, calculating the position difference between the relative and local estimates in the horizontal and vertical directions. It obtains the aircraft's current velocity and attitude information, derived from the fusion output of the onboard inertial navigation system and satellite navigation. Based on the calculated position difference, velocity estimate, and attitude information, a kinematic coupling model is used to deduce the potential navigation error sources causing the position difference. The kinematic coupling model, based on the aircraft's kinematic equations, correlates the position difference, velocity observation, and attitude information to separate and estimate error sources such as position drift, velocity deviation, and heading deviation. Based on the estimated error source category and magnitude, and combined with the priority of the current flight mission, a set of specific correction values ​​is generated. For example, if an eastward velocity deviation is estimated, a correction factor for the eastward velocity integral is generated; if a heading deviation is estimated, a fine-tuning amount for the heading reference is generated. The corrections also include a position correction compensation value for the inertial navigation system, which can be directly the calculated position difference after low-pass filtering. The generated position correction compensation value, velocity integral correction factor, and heading reference fine-tuning amount are packaged into a structured data packet that can be directly interpreted and applied by the aircraft's own navigation system. This data packet is the local navigation correction amount applied to the aircraft's own navigation system. The local navigation correction amount is sent to the onboard navigation filter through a specific data interface for subsequent state updates.

[0129] See Figure 4In the performance evaluation of hierarchical subnets, the relationship between data exchange volume and routing efficiency can be quantitatively analyzed using a bi-axis bar and line graph. The graph uses the Alpha subnet (node ​​1), Beta subnet (node ​​7), and Gamma subnet (node ​​11) as evaluation objects. The left vertical axis represents the data exchange volume (KB) within the subnet, and the right vertical axis represents the routing efficiency (%) of the corresponding subnet. From the perspective of data exchange volume, the Alpha subnet (node ​​1) has the highest data exchange volume, approximately 1250KB; the Beta subnet (node ​​7) has the lowest, approximately 980KB; and the Gamma subnet (node ​​11) has a moderate data exchange volume, approximately 1120KB. This difference is directly related to the topological hierarchical role of each subnet: the Alpha subnet, as the top-level core reference node, undertakes the core functions of global data aggregation and relay, thus having the largest data interaction load; the Beta subnet, as an intermediate-layer subnet, only needs to complete data aggregation within its own region, resulting in a relatively low exchange volume; and the Gamma subnet, as a regional monitoring subnet, has a data exchange volume between the two. From a routing efficiency perspective, the Alpha subnet achieves approximately 92% routing efficiency, the Beta subnet approximately 88%, and the Gamma subnet approximately 90%. Fluctuations in routing efficiency are related to node density, link redundancy, and topology stability within the subnet: the Alpha subnet, due to the strong relay capabilities of its core nodes, has clear routing decision paths and high efficiency; the Beta subnet, limited by the load of auxiliary communication links between nodes, experiences increased routing overhead and a slight decrease in efficiency; the Gamma subnet maintains high routing efficiency by optimizing the scheduling of neighboring links in the monitored sub-regions. In summary, the hierarchical topology subnet design achieves a trade-off between data exchange volume and routing efficiency: the core subnet (Alpha) trades higher data exchange volume for high global routing efficiency, the intermediate layer subnet (Beta) balances the overall load by reducing exchange volume, and the regional subnet (Gamma) achieves a performance compromise between the two. This hierarchical design provides a quantitative basis for dynamic topology adjustment in collaborative relative positioning of aircraft clusters, allowing for dynamic optimization of role allocation and link scheduling strategies for each subnet based on task priority.

[0130] In one embodiment of the invention, the aircraft receives local navigation corrections and inputs them into an onboard navigation filter for fusion. The onboard navigation filter employs an extended Kalman filter structure, and the system state vector includes position, velocity, attitude, and sensor error states. The navigation filter updates the system state vector based on the position correction compensation value, velocity integral correction coefficient, and fine-tuning of the heading reference contained in the local navigation corrections, thereby correcting the aircraft's current position, velocity, and attitude estimates to obtain updated positioning information. For example, after fusing the local navigation corrections, the navigation filter of ordinary node A updates its output position estimate from (29.8, -20.1, 101.9) meters to (30.0, -20.3, 102.0) meters, velocity estimate from (5.2, 0.3, -0.1) meters / second to (5.18, 0.28, -0.08) meters / second, and heading angle from 45.0 degrees to 45.1 degrees.

[0131] In some embodiments, the confidence change of the positioning information output by the aircraft's own navigation system after applying local navigation corrections is evaluated. Confidence is typically measured by the trace of the covariance matrix or the position accuracy dilution factor. Before the update, the navigation filter of ordinary node A output a horizontal position error ellipsoid with a semi-major axis of 1.5 meters. After the update, the semi-major axis of the horizontal position error ellipsoid is reduced to 1.2 meters, indicating an improvement in confidence. Based on the updated positioning information, the confidence change, and other aircraft state changes obtained from global information exchange data, the positioning sensor accuracy level and flight stability indices in the individual state information are re-evaluated. The evaluation of the positioning sensor accuracy level combines the magnitude of the covariance of the updated positioning information with the signal quality reported by the satellite navigation receiver. The evaluation of the flight stability indices combines the updated velocity variance, attitude angle fluctuation amplitude, and the history of recent maneuver commands. Based on the re-evaluated individual state information, combined with the dynamic changes in environmental constraint information obtained in real time from airborne sensors, the requirements for the next round of cooperative positioning are predicted. The dynamic changes in environmental constraint information may include newly detected communication interference sources or adjustments to the mission area boundaries.

[0132] Optionally, role and task suggestions are generated for the next round of cooperative localization. Role suggestions propose whether to replace the core reference node, adjust landmark auxiliary nodes, or reallocate the monitoring sub-areas of ordinary nodes based on predicted needs. For example, if a reassessment reveals that the remaining energy level of the core reference node root node Alpha has significantly decreased and fallen below a threshold, the role suggestion would propose replacing root node Alpha with an ordinary node and recommending an aircraft from the current ordinary nodes with the best overall positioning sensor accuracy and communication link bandwidth as a new core reference node candidate. Task suggestions recommend the core reference node's next round of associated node list and ranging scheduling timing. The associated node list is generated based on the latest inter-node distance estimates and communication link quality predictions, while the ranging scheduling timing considers avoiding overlap with potential communication interference periods. Task suggestions recommend ordinary nodes' key monitoring areas and priorities for auxiliary ranging with neighboring nodes. The key monitoring area recommendations are based on the reallocated monitoring sub-areas, and the priority for auxiliary ranging with neighboring nodes is ranked according to the stability of the inter-node geometry and historical ranging quality. Role and task suggestions are encapsulated in a structured suggestion data package as an output of the cooperative localization process.

[0133] Understandably, this involves integrating key data, role assignment results, topology, and final relative position estimates from the current positioning process to generate a cooperative positioning process report. A report header is created containing a timestamp (2025-07-15 10:30:00 UTC) and a cluster identifier ("Swarm_Unit_07"). The report body first records the initial conditions at the start of this cooperative positioning cycle. These conditions include events that trigger cooperative positioning, such as periodic timed triggers or formation reconfiguration commands, and initial individual state and environmental constraint information summaries. The complete role assignment results are then recorded in detail, listing each aircraft's ID, initial role, and mission parameters in a table format. Finally, the established hierarchical ranging and data exchange topology is recorded in the form of an adjacency matrix. Rows and columns of the adjacency matrix represent aircraft IDs, and matrix elements represent connection relationships and data types.

[0134] The document records an overview of the collected raw data, including a total of 24 ranging observation data points, two sets of core node data packets for information exchange, and three sets of landmark observation information. It also records a summary of the relative position constraint network construction, including 12 nodes, 28 edges, and the constraint type distribution. The constraint type distribution indicates that 27 edges are distance constraints, 1 edge has an angle constraint, and 3 nodes have absolute position anchor point constraints. The document records the convergence of the iterative optimization process for solving the relative position constraint network, including the initial objective function value of 154.7, the final objective function value of 3.2, and the number of iterations of 35. Finally, it lists the final relative position estimates in a table, containing the number of each aircraft, its coordinate estimates in a common reference frame, and the covariance information of the relative position estimates, expressed as variance values ​​along three axes. The report summarizes the generated local navigation corrections, indicating that nine out of twelve aircraft generated non-zero corrections. It also compares the updated positioning information with the previous cycle, showing an average position update of 0.3 meters. Attached are role and mission suggestions for the next round, presented as a summary of the aforementioned packaged suggestion data. All recorded information is organized logically, with necessary indexes and annotations added, forming a structured cooperative positioning process report suitable for post-event analysis and mission retrospective review.

[0135] In practical implementation, the quantitative evaluation of the confidence level of positioning information can adopt a measurement method based on information entropy, and the uncertainty of the navigation system state can be characterized by the covariance matrix. Define a confidence improvement metric. To calculate the reduction in entropy of the state estimation information before and after incorporating local navigation corrections:

[0136]

[0137] Where: characters The measure of confidence improvement is expressed in bits; characters Represents the covariance matrix of the navigation filter state estimate before fusing local navigation corrections; character Represents the covariance matrix of the navigation filter state estimate after fusing local navigation corrections; character This represents the determinant operation of a matrix. This metric... This directly reflects the information gain obtained by the navigation system through the fusion of cooperative positioning results; a regularized gain value indicates an increase in confidence. For example, if the determinant of the covariance matrix of a regular node A changes before and after the application of the correction, the calculated... The value is 0.5 bits, which can be used as an objective quantitative input when re-evaluating the accuracy of the positioning sensor.

[0138] See Figure 5 This diagram presents the relative positional distribution of nodes within a cluster under a common reference frame. From a role and topology perspective, the diagram clearly identifies three types of core nodes: Core Reference Nodes (dots): As the root node at the top level of the topology, they play a crucial role in data relay and global benchmarking. Nodes 01 and 02 are located near the origin and at the far end of the X-axis, respectively, providing stable positional anchoring and data aggregation centers for the entire cluster. Landmark Auxiliary Nodes (triangles): As anchor points with known absolute positions, they are distributed in the middle and far ends of the X-axis (nodes 03, 04, and 05). Their positional constraints provide crucial absolute positional weights for the relative position constraint network, effectively reducing the risk of global positional drift. Ordinary Nodes (squares): As the main sensing and monitoring entities of the cluster, they are associated with the core reference nodes in the form of a star-shaped subnet, distributed between 10 and 30 meters along the Y-axis. Their positional estimation accuracy depends on the distance constraints with the core reference nodes and neighboring ordinary nodes. The solution results for the relative position constraint network show that the estimated coordinates of each node exhibit good geometric configuration and consistency: core reference nodes 01 and 02 are located at (0,0) and (50,0) respectively, forming a baseline in the X-axis direction; landmark auxiliary nodes 03, 04, and 05 are located at (10,20), (40,20), and (25,40) respectively, constructing a three-dimensional anchoring framework covering the cluster; and ordinary nodes form a hierarchical formation structure based on the distribution of monitoring sub-regions. This distribution satisfies the topological constraints of cooperative positioning and provides a reliable positional reference for subsequent navigation correction and role reassignment.

[0139] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A cooperative relative positioning method for an aircraft swarm, characterized in that, The method includes: Obtain the individual status information and environmental constraint information of all aircraft in the cluster, and assign appropriate roles to each aircraft to form role assignment results; Based on the role assignment results, a hierarchical ranging and data exchange topology is established within the cluster. By utilizing the hierarchical ranging and data exchange topology, the relative ranging and communication processes between aircraft are scheduled, and raw ranging observation data and information exchange data are collected. The collected raw ranging observation data is corrected for consistency error, and a corrected set of relative distances between all pairs of aircraft in the cluster is generated. Based on the modified set of relative distances and information exchange data, a relative position constraint network for the aircraft cluster is constructed. By solving the relative position constraint network, the estimated relative position of each aircraft in the cluster in a common reference frame is calculated iteratively. The aircraft's navigation state is fused and corrected based on the calculated relative position estimate to generate a local navigation correction. The local navigation correction is used to update the aircraft's own positioning information and generate role and mission suggestions for the next round of cooperative positioning. By integrating key data from this round of positioning, the role assignment results, the topology, and the final relative position estimate, a collaborative positioning process report is generated.

2. A method of cooperative relative positioning of a fleet of aerial vehicles as recited in claim 1, wherein, Assign appropriate roles to each aircraft to generate role assignment results, including: Extract the positioning sensor accuracy level, communication link bandwidth, remaining energy level, and flight stability indicators of each aircraft from the individual status information; The environmental constraint information is used to identify the area to be covered by the cluster, the location information of known landmarks, and the distribution of communication interference sources. Based on the accuracy level of the positioning sensor, the aircraft is initially classified into core reference node candidates and ordinary nodes; Based on the communication link bandwidth and the remaining energy level, aircraft with stable data relay capabilities are selected from the core reference node candidates and designated as core reference nodes. Based on the flight stability index and the area to be covered by the cluster, a relatively fixed monitoring sub-area is assigned to each ordinary node; Based on the location information of the known landmarks, designate several aircraft closest to the landmarks as landmark auxiliary nodes; The assignment information of core reference nodes, landmark auxiliary nodes, and ordinary nodes assigned to monitoring sub-regions is integrated to form a structured role assignment result, which includes a unique role identifier for each aircraft and its associated mission parameters.

3. A method of cooperative relative positioning of a cluster of aircraft as recited in claim 2, wherein, Based on the role assignment results, a hierarchical ranging and data exchange topology is established within the cluster, including: Set the core reference node as the top-level root node of the topology; Centered on each core reference node, a batch of ordinary nodes and landmark auxiliary nodes are dynamically associated based on its communication coverage capability and the bandwidth of the communication link, forming multiple star-shaped subnets with the core reference node as the parent node; Within each star subnet, auxiliary communication links are allowed to be established between associated ordinary nodes based on the proximity relationship of the monitored sub-regions; Establish dedicated data links between landmark auxiliary nodes for synchronizing landmark observation data; Information flow rules are set: ranging data and status information of ordinary nodes are first aggregated to the core reference node of their respective star subnet; global data exchange is carried out between core reference nodes; observation data of landmark auxiliary nodes are shared through dedicated data links and then distributed to the associated core reference nodes; Based on the core reference node relationships and information flow rules, a complete topology description including hierarchical relationships, link connection relationships, and data flow directions is constructed, namely the layered ranging and data exchange topology structure.

4. A method of cooperative relative positioning of a cluster of aircraft as recited in claim 3, wherein, Utilizing the hierarchical ranging and data exchange topology, the relative ranging and communication processes between aircraft are scheduled, and raw ranging observation data and information exchange data are collected, including: Based on the division of the star subnets in the topology, an independent ranging time slot is assigned to each star subnet; Within the allocated ranging time slot, the core reference node of the star subnet sends ranging request signals to all its associated ordinary nodes and landmark auxiliary nodes. Upon receiving a ranging request signal, the associated node replies with a response signal containing its own identity identifier and timestamp; The core reference node calculates the original two-way ranging observation value with the core reference node based on the time difference between sending a request and receiving a response; All star-shaped subnets sequentially complete their internal ranging processes, collecting raw ranging observation data between all core reference nodes and their associated nodes; During the preset communication time slots, the core reference nodes exchange their collected ranging observation data and their own navigation status information to form global information exchange data; Landmark auxiliary nodes exchange their observation information of known landmarks on a dedicated data link; The raw ranging observation data, global information exchange data, and landmark observation information from all star-shaped subnets are aggregated to form a complete raw observation and information dataset.

5. A method of cooperative relative positioning of a cluster of aircraft as recited in claim 4, wherein, The collected raw ranging observation data is corrected for consistency errors to generate a corrected set of relative distances between all pairs of aircraft within the cluster, including: Identify redundant measurements in the original ranging observation data: due to multiple ranging paths, either directly or indirectly, between core reference nodes and some ordinary nodes. Using the core reference node as the benchmark, an error state equation is established regarding the relative distance between all aircraft pairs, and the original ranging observations are used as noisy observations. Aircraft velocity information from global information exchange data is introduced as a constraint on the process model, and the absolute position reference of some aircraft in the landmark observation information is used to provide anchor points for the error state equation. An optimization estimation algorithm based on a sliding window is used to iteratively solve the error state equation. Under the conditions of satisfying all direct measurements, velocity constraints and absolute position references, the total error of the relative distance network of the entire cluster is minimized. The systematic error correction for all original ranging observations obtained after optimization is applied to the corresponding original data. The output is a set of theoretically reachable relative distances that, after applying corrections, are more consistent and cover all aircraft pairs within the cluster.

6. A method of cooperative relative positioning of a cluster of aircraft as recited in claim 5, wherein, Based on the modified relative distance set and information exchange data, a relative position constraint network for the aircraft cluster is constructed, including: Each aircraft is abstracted as a node in the relative position constraint network; The relative distance between each pair of aircraft in the modified relative distance set is abstracted as an edge connecting the corresponding two nodes, with the edge carrying a distance value as a constraint weight. From the information exchange data, the heading angle difference information exchanged between the core reference nodes is extracted, and the heading angle difference information is converted into soft constraints on the relative azimuth angles between aircraft, which are then used as angle constraint weights attached to the corresponding edges. From the landmark observation information, the absolute position information of the known landmark points is converted into position anchor point constraints for the aircraft, which serves as landmark auxiliary nodes. The position anchor point constraints are used as absolute position weights applied to the corresponding nodes. By integrating all nodes, edges with distance weights, edges with angle weights, and absolute position weights on the nodes, a graph model containing multiple constraint types is constructed, namely the relative position constraint network.

7. A method of cooperative relative positioning of a cluster of aircraft as recited in claim 6, wherein, By solving the relative position constraint network, the estimated relative position of each aircraft in the cluster in a common reference frame is calculated iteratively, including: Initialization steps: Assign an initial relative position coordinate to each node in the relative position constraint network, wherein nodes with absolute position weights are initialized with their anchor point positions, and other nodes are roughly geometrically initialized according to their distance constraints with neighboring nodes; Establish the objective function: The objective function consists of three parts. The first part is the sum of squares of the differences between the calculated distance constraint values ​​and the actual distance weights of all edges in the network. The second part is the sum of squares of the differences between the calculated azimuth values ​​and the angle weights of all edges with angular constraints. The third part is the sum of squares of the differences between the calculated position values ​​of nodes with absolute position weights and the anchor positions. A nonlinear least squares optimization algorithm is used to iteratively minimize the objective function, with the position coordinates of all nodes as the optimization variables. In each iteration, the gradient of the objective function with respect to each node coordinate is calculated, and the node coordinates are updated along the gradient descent direction. Set an iteration convergence condition: stop iterating when the change in the objective function value is less than a preset threshold or when the maximum number of iterations is reached. The coordinates of each node obtained after the final iterative optimization are used as the relative position estimate of the corresponding aircraft in the common reference frame.

8. A method of cooperative relative positioning of a fleet of aerial vehicles as in claim 7, wherein, The aircraft's navigation state is fused and corrected based on the calculated relative position estimate to generate local navigation corrections, including: Each aircraft extracts its own relative position estimate from the set of relative position estimates; The relative position estimate is compared with the local position estimate provided by the aircraft's own onboard navigation system, and the position difference between the relative position estimate and the local position estimate in the horizontal and vertical directions is calculated. Obtain the current velocity estimate and attitude information of the aircraft; Based on the position difference, velocity estimation, and attitude information, a kinematic coupling model is used to deduce the potential navigation error sources that cause the position difference. The potential navigation error sources include position drift, velocity deviation, and heading deviation. Based on the type and magnitude of the error sources, and in conjunction with the priority of the current flight mission, a set of correction values ​​is generated; The correction amounts include position correction compensation values ​​for the inertial navigation system, correction coefficients for the velocity integral, and fine-tuning amounts for the heading reference. The correction values ​​are packaged to form the local navigation correction values ​​applied to the aircraft's own navigation system.

9. A method of cooperative relative positioning of a fleet of aerial vehicles as in claim 8, wherein, The local navigation correction is used to update the aircraft's own positioning information, and role and task suggestions for the next round of cooperative positioning are generated, including: The aircraft receives the local navigation correction and inputs it into the onboard navigation filter for fusion; The navigation filter updates the system state vector based on the correction amount, thereby correcting the aircraft's current position, velocity, and attitude estimation, and obtaining updated positioning information. The confidence level of the positioning information output by the aircraft's own navigation system was evaluated after the local navigation correction was applied. Based on the updated positioning information, the change in confidence level, and other aircraft status changes obtained from the information exchange data, the positioning sensor accuracy level and flight stability index in the individual status information are reassessed. Based on the reassessed individual state information and the dynamic changes in the environmental constraint information, the demand for the next round of cooperative localization is predicted. Role suggestions: Based on the predicted needs, suggest whether it is necessary to replace the core reference node, adjust the landmark auxiliary node, or reallocate the monitoring sub-areas of ordinary nodes; The task suggestions are generated as follows: for core reference nodes, suggestions are made regarding the list of associated nodes for the next round and the timing of ranging scheduling; for ordinary nodes, suggestions are made regarding the key monitoring areas and the priority of auxiliary ranging with neighboring nodes. The proposed roles and tasks are encapsulated and included as part of the output. Integrating key data from this round of positioning, the role assignment results, the topology, and the final relative position estimate, a cooperative positioning process report is generated, including: Create a report header that includes a timestamp and cluster identifier; The report body first records the initial conditions at the start of this round of cooperative localization, including the event that triggered cooperative localization, the initial individual state information summary, and the environmental constraint information summary; Record the complete details of the role assignment results, including the initial role and mission parameters for each aircraft; The established hierarchical ranging and data exchange topology is recorded in the form of a graphical description or an adjacency matrix. Record an overview of the raw data collected, including the total amount of distance measurement observation data, the scale of information exchange data, and the amount of landmark observation information; Record a summary of the construction of the relative position constraint network, including the number of nodes, the number of edges, and the distribution of constraint types; Record the convergence status of the iterative optimization process for solving the relative position constraint network, including the initial objective function value, the final objective function value, and the number of iterations; List the final relative position estimates in a table, including the number of each aircraft, its coordinate estimates in a common reference frame, and the covariance information of the relative position estimates; Record an overview of the generated local navigation corrections, and compare the updated positioning information with the previous cycle. Attached are the suggested roles and tasks for the next round of generation; All recorded content is organized in logical order, and necessary indexes and annotations are added to form a structured collaborative localization process report.

10. A cooperative relative positioning system for a fleet of aircraft, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the cooperative relative positioning method for an aircraft cluster as described in any one of claims 1 to 9.

Images