A multi-unmanned aerial vehicle formation rapid generation method

Abstract

This invention discloses a method for rapidly generating multi-UAV formations, belonging to the field of UAV formation control. Specifically, it includes: setting the target formation and three-dimensional inter-UAV spacing according to mission requirements; determining the cluster size; converting UAV GPS coordinates to planar coordinates; calculating the number of rows and columns of the formation based on the number of nodes using a closed-form algorithm, and assigning row and column positions according to IDs; calculating the offset of each slave UAV relative to the lead UAV after secondary numbering; calculating the target position of each slave UAV at the next moment by combining the real-time trajectory of the lead UAV; and finally guiding the UAVs safely to the target point using a distributed artificial potential field method. This invention achieves rapid formation generation and transformation, and implements multiple collision avoidance strategies such as three-dimensional layering and wave-shaped arrangement, effectively improving the efficiency and safety of the cluster in tasks such as logistics, surveying, and emergency rescue.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) formation control, and specifically relates to a method for rapidly generating multi-UAV formations. Background Technology

[0002] With the rapid development of unmanned systems technology, the application of drone platforms is evolving towards generalization and intelligence. Compared to a single drone, multi-drone swarm collaborative operations demonstrate significant advantages in improving task execution efficiency, expanding operational coverage, and enhancing system robustness. However, in civilian fields such as logistics delivery and emergency rescue, the operating environment is typically characterized by high timeliness requirements, strong environmental dynamics, and diverse task modes. Multi-drone swarm operations have become an effective solution for tasks such as rapid package delivery, disaster relief material distribution, and search and rescue. Therefore, the ability to rapidly generate drone formations plays a crucial role in whether the swarm can quickly respond to task requirements, efficiently enter a working state, and successfully complete operational objectives.

[0003] When performing diverse tasks, drone swarms often need to adopt specific flight formations based on specific operational requirements to optimize overall performance. For example, in large-scale geographic mapping or agricultural and forestry status monitoring, a horizontal straight-line formation is often used to maximize coverage width; in area inspection (such as power line and pipeline inspection) or precision agriculture (such as variable-rate fertilization and spraying), a rectangular formation helps achieve comprehensive coverage; when accompanying and tracking specific moving targets (such as logistics convoys or disaster victims) or conducting surround monitoring of key areas (such as fire scenes), an arc or circular formation provides the best field of view; during long-distance transfers or convoy transport, a triangular or wedge formation can effectively reduce flight drag and save energy. In applications with complex and ever-changing mission environments, the speed and stability of formation generation and transformation directly impact mission cost control, operational safety, and the achievement of expected results.

[0004] The publication CN114326826A, titled "Multi-UAV Formation Transformation Method and System," establishes a Markov decision model applicable to multi-UAV formation transformation. It employs deep reinforcement learning to calculate the optimal solution. During the calculation process, it collects the state and action information of other obstacles and calculates their attention scores. The higher the score of an obstacle, the greater its influence on the UAV's path. In the next action selection, the UAV should pay attention to this obstacle and avoid it, thereby effectively reducing the probability of collisions between UAVs.

[0005] The patent application CN110865653A, entitled "A Distributed Cluster UAV Formation Transformation Method," first initializes the positions and velocities of the cluster UAVs and designs the cluster's flight formation. Then, it establishes a collaborative task allocation model for the UAVs, where each UAV independently selects tasks and constructs and updates task packages. After the task packages are updated, each UAV simultaneously shares its known information with other UAVs in its communication neighborhood and performs consistency negotiation. CBBA (Conformity-Based Allocation) is used to complete the position allocation from the initial random position to the target cluster formation, ensuring that the algorithm eventually converges to a conflict-free task allocation scheme. Finally, by designing position and velocity consistency control laws based on the relationships between UAVs and between UAVs and the virtual leader, the formation and maintenance of the cluster formation are achieved.

[0006] The patent application CN115586795A, entitled "A Method and System for Formation Transformation of Fixed-Wing UAVs," selects state variables and manipulation variables to establish a UAV motion state model; determines the coordinates of reference points for multi-UAV formations; calculates the coordinates of multiple target points based on the distances and angles between multiple target points and the reference points, realizing a formation matrix representation; constructs a distance matrix at the current moment; performs row reduction on the distance matrix to achieve the allocation of multiple UAVs and multiple target points; determines a consistency control law; generates an inter-UAV collision avoidance strategy; and completes the multi-UAV formation transformation through consistency control. This method improves the flexibility, coordination capability, and efficiency of the formation control and transformation task for fixed-wing UAVs.

[0007] The patent application CN109254592A, entitled "A Method and System for Controlling UAV Formation Changes," establishes an optimal efficiency model for UAV formation changes using UAV position coordinates as constraints and energy consumption and / or completion time as efficiency metrics. Based on the actual changes in formation assembly and formation maintenance, a minimum energy consumption model and / or a shortest time model are obtained. Solving the minimum energy consumption model and / or shortest time model yields the UAV's implementation position, thus achieving formation change control. This solution addresses the problems of complex calculations and high energy consumption associated with multiple constraint models in existing technologies, simplifying the computational complexity of the models, saving resources, reducing costs, and improving computational efficiency.

[0008] However, the aforementioned existing technologies typically introduce multiple constraints and complex models when implementing drone formation control. As the formation size expands, or when multiple sub-formations need to be merged and split, the computational complexity increases significantly, placing excessive demands on the real-time computing capabilities of the onboard computing unit, which is not conducive to deployment on low-cost, lightweight commercial drone platforms. Furthermore, in application scenarios requiring the integration of multiple formations for rapid switching, existing solutions struggle to simultaneously balance generation efficiency and inter-drone collision avoidance reliability.

[0009] Therefore, there is an urgent need for a computationally efficient drone formation control method that can quickly generate various formations and incorporate effective collision avoidance strategies to meet the growing demand for civilian drone swarm applications. Summary of the Invention

[0010] To overcome the shortcomings of existing technologies, such as complex models, large computational load, and difficulty in achieving real-time collision avoidance between drones on resource-constrained embedded platforms, this invention proposes a method for rapid generation of multi-drone formations. By establishing a unified mathematical model for formation description and combining it with an optimized artificial potential field control law, this method enables rapid, stable, and safe generation of formations for large-scale drone swarms in scenarios such as logistics warehousing, aerial photography performances, and agricultural plant protection.

[0011] The specific steps of the method for rapidly generating multi-UAV formations are as follows:

[0012] Step 1: Determine the target formation of the UAVs according to the mission requirements, and set the three-dimensional interval reference parameters between the UAVs;

[0013] Target formations include: triangles, rectangles, or wedges, etc.

[0014] Interval reference parameters include: horizontal (between columns) safety interval reference. Vertical (between rows) safety interval benchmark Safety spacing benchmark in the vertical direction (between floors) .

[0015] Step 2: Determine the total number of drone nodes participating in the formation. Furthermore, a hierarchical strategy is introduced to divide the drone swarm;

[0016] Specifically: First, set the maximum node threshold for a single sub-cluster. ,like Then all drones are grouped together as a single cluster; if Then all drones are divided into Sub-clusters Each sub-cluster follows the same topology, and sub-clusters maintain a preset safety offset. .

[0017] Step 3: Convert the latitude and longitude coordinates of the lead aircraft and each slave aircraft in the formation to Cartesian coordinates.

[0018] Assume the geographical coordinates of the lead aircraft or a slave aircraft are as follows: Convert to Mercator plane coordinates The formula is as follows:

[0019]

[0020] in Longitude Latitude For height; Given the Earth's average radius, the lead aircraft's real-time planar coordinates are marked as follows: .

[0021] Step 4: Based on the total number of drone nodes Given the target formation type, calculate the total number of rows required using a deterministic closed-form solution. :

[0022] The target formation type is triangle: number of rows in the formation. satisfy The smallest integer solution is calculated using the following formula:

[0023]

[0024] in This represents the function for rounding up.

[0025] The target formation type is rectangular: set the fixed number of columns for the rectangular formation. The number of columns is determined by the task coverage width, then the number of rows... The calculation is as follows: ;

[0026] The target formation type is a wedge-shaped or V-shaped formation: A wedge-shaped formation typically has one lead aircraft in the first row, followed by two aircraft in each subsequent row, one on the left and one on the right. The number of rows... The calculation is as follows: ;

[0027] Step 5: Calculate the maximum number of drone nodes that each row in the target formation type can accommodate;

[0028] Let the first OK( The maximum number of drone nodes that can be accommodated is :

[0029] For triangle formation: the The capacity of the row is ;

[0030] For rectangular arrays: the capacity of each row is fixed, i.e. ;

[0031] For wedge formations: first row Each subsequent line .

[0032] Step Six, for The drones are arranged in ascending order by ID, and each drone independently calculates its own row number. and column number .

[0033] With the first Taking a drone as an example:

[0034] For triangular formations: by accumulating the previous... The total number of rows determines the position, if the condition is met. Then the machine is located at the th Row; its column number .

[0035] For rectangular arrays: row number ;column number .in To round down, This is a modulo operation.

[0036] For wedge formations: when hour, ;when Time: line number ;column number .

[0037] Step 7: Based on the row and column positions of each drone Generate a globally unique secondary number for each. And verify it;

[0038] Secondary numbering for different formations The calculation formula is as follows:

[0039] The calculation for a triangle formation is as follows: .

[0040] The calculation for rectangular formation is as follows: .

[0041] For wedge formations: the calculation is as follows: when Time (lead aircraft): .

[0042] when Time (slave): .

[0043] The verification standard is: automatic comparison within the airborne computing unit. With the physical properties of the drone itself :like If the position calculation is correct, then proceed to the next step of trajectory generation; otherwise, trigger exception handling.

[0044] Step 8: Each slave device determines its row and column position. and reference interval parameters Calculate the expected relative position vector of each aircraft relative to the lead aircraft;

[0045] The expected relative position vector of each slave device Specifically as follows:

[0046] Vertical position All formations extend row by row, with the lead aircraft as the origin. Then: ;

[0047] Altitude position The design employs a layered anti-collision system, with the layer height increasing as the number of rows increases. ;

[0048] Horizontal position The specific calculations are as follows:

[0049] First, the basic horizontal offset: first calculate the center point of the row, then calculate the nodes. Offset: ;

[0050] Then perform a wave-shaped alternation correction: if If it is even, then ;like If it is an odd number, then .

[0051] Step 9: Using the real-time status information of the master machine, combined with the expected relative position of each slave machine relative to the master machine, calculate the global expected position of each slave machine at the next moment.

[0052] Obtain the current Mercator plane coordinates of the lead aircraft. and heading angle (With true north as 0 degrees and clockwise as positive). Calculate the global expected position of each slave device at the next time step. :

[0053]

[0054] in: This is a rotation matrix used to align relative formations to the flight heading; This represents the current velocity vector of the lead aircraft. This is the estimated delay of the system control cycle, and it is used for feedforward compensation.

[0055] Step 10: Based on the improved distributed artificial potential field method, perform tracking control to guide each UAV to safely reach its respective global desired position and form the target formation.

[0056] First, for slave device s, define its total potential field function. :

[0057]

[0058] in This represents the current position of slave device s. For slave device s to point to its desired position gravitational field, For targeting nearby drones or obstacles of the slave device s The repulsive field.

[0059] Then, calculate the control force on slave device s. That is, the negative gradient of the potential field: ;

[0060] Finally, the tracking process is executed: based on control force. Calculate the desired acceleration of the slave device 's', input it into the underlying flight control system to adjust its speed and attitude, and achieve the desired coordinates. Accurate tracking, and the calculated planar coordinates The coordinates are converted back to latitude and longitude using the inverse Mercator projection for issuing actual navigation commands.

[0061] The advantages of this invention are:

[0062] 1. Highly efficient model with low resource requirements: Based on deterministic closed-form solution and distributed mapping, the algorithm has low time complexity, making it very suitable for embedded systems of commercial drones with limited computing power.

[0063] 2. Built-in safety: Through three-dimensional layering, "wave-shaped" arrangement and improved artificial potential field method, multiple mechanisms are in place to ensure collision avoidance between machines, significantly improving the safety of cluster operations.

[0064] 3. High flexibility and scalability: By modifying the topology matrix T and baseline parameters, various complex formations can be quickly generated, scaled and switched, and a hierarchical strategy can be used to support ultra-large-scale clusters.

[0065] 4. Strong anti-interference capability: Combining the trajectory generation model based on the lead aircraft's state prediction and distributed potential field control, the formation is more robust to lead aircraft maneuvers and environmental disturbances. Attached Figure Description

[0066] Figure 1 This is a flowchart of a method for rapidly generating multi-UAV formations according to the present invention;

[0067] Figure 2 This is a diagram showing the coordinate system transformation relationship used in this invention. Detailed Implementation

[0068] To facilitate understanding and implementation of the present invention by those skilled in the art, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are merely some, not all, embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0069] This invention proposes a computationally efficient method for rapid generation and cooperative control of multi-UAV swarm formations applicable to various commercial scenarios, such as... Figure 1 As shown, the specific steps are as follows:

[0070] Step 1: Determine the target formation of the drones (such as triangle, rectangle, or wedge) based on the task requirements (e.g., the pattern of a light show, the width of a plant protection operation), and set the reference parameters for the spacing between the drones:

[0071] The horizontal (between columns) safety interval benchmark;

[0072] The safety interval benchmark for the vertical direction (between rows);

[0073] This serves as the safety interval benchmark in the vertical direction (between floors). The purpose of this is to construct a three-dimensional collision avoidance formation, and its value should be greater than the sum of the drone's maximum vertical error and its fuselage height.

[0074] Step 2: Determine the total number of drone nodes participating in the formation. Furthermore, a hierarchical strategy is introduced to divide the drone swarm;

[0075] To accommodate ultra-large-scale clusters and reduce computational dimensionality, a hierarchical strategy is introduced: setting a maximum node threshold for a single sub-cluster. .like Then all drones are grouped together as a single cluster; if Then all drones are divided into Sub-clusters Each sub-cluster follows the same topology T structure, and the sub-clusters maintain a preset safety offset. .

[0076] Step 3: To eliminate the influence of the Earth's curvature on the accuracy of large-scale formation and achieve precise relative position control, the latitude and longitude coordinates of each UAV (including the lead and slave UAVs) are converted into Cartesian coordinates.

[0077] like Figure 2 As shown, assume the geographical coordinates of the lead aircraft or a slave aircraft are... ,in Longitude latitude, For height; convert to Mercator plane coordinates The formula is as follows:

[0078]

[0079] in, The Earth's average radius is approximately 6371 km. and All units are in radians; the real-time planar coordinates of the lead aircraft are marked as follows: .

[0080] Step 4: Based on the total number of drone nodes Given the target formation type, calculate the total number of rows required using a deterministic closed-form solution. This invention uses a rigorous mathematical closed-form solution for calculation, ensuring deterministic computational efficiency (O(1) time complexity), which is superior to traditional iterative search methods.

[0081] For triangle formation: number of rows satisfy The smallest integer solution is calculated using the following formula: in This represents the function for rounding up.

[0082] For rectangular arrays: Set the fixed number of columns for the rectangular array to... The number of columns is determined by the task coverage width, then the number of rows... The calculation is as follows: ;

[0083] For wedge-shaped or V-shaped formations: Wedge-shaped formations typically have one lead aircraft in the first row, followed by two aircraft in each subsequent row, one on the left and one on the right. The number of rows is... The calculation is as follows: ;

[0084] Step 5: Calculate the maximum number of drone nodes that each row in the target formation type can accommodate;

[0085] Calculate the first OK( The maximum number of drone nodes that can be accommodated :

[0086] Triangle formation: the row capacity ;

[0087] Rectangular array: The capacity of each row is fixed, i.e. ;

[0088] Wedge formation: first row Each subsequent line .

[0089] Step 6: Sort all drones in ascending order by ID ( Each machine independently calculates its own row number. and column number .

[0090] With the first Taking a drone as an example:

[0091] For triangular formations: by accumulating the previous... The total number of rows determines the position. If the following conditions are met... Then the machine is located at the th Row; its column number .

[0092] For rectangular arrays: row number ;column number .in To round down, This is a modulo operation.

[0093] For wedge formations: when hour, ;when Time: line number ;column number

[0094] Step 7: Based on the row and column positions of each drone Generate a globally unique secondary number for each drone. And verify it;

[0095] This step is used to reverse verify the correctness of the position allocation, ensuring that the calculated position is correct. Can uniquely map back to the original For different formations, secondary numbering is used. The calculation formula is as follows:

[0096] For triangle formation: Triangle formation Line accommodate A drone, front The total number of rows is the sum of an arithmetic sequence. The formula is: ;

[0097] It was before The sum of all drones (i.e. ), plus the column number of the current row It should be exactly equal to the machine's original ID.

[0098] For rectangular arrays: Each row of a rectangular array is fixed. The calculation formula is: .in, Indicates the preceding The total number of rows that are full, plus the number of rows currently in the current row. One position.

[0099] For wedge formations: The first row of a wedge formation contains only one lead aircraft, and from the second row onwards, each row contains two aircraft (left and right wingmen). The calculation formula needs to be described in segments: When Time (lead aircraft): .when Time (slave): Where 1 represents the lead aircraft in the first row; Represents from the 2nd row to the 3rd row. OK (Total The cumulative total number of dual machines (in rows); The ordinal number of the current row (1 or 2).

[0100] Verification standard: The system automatically compares data within the onboard computing unit. With the physical properties of the drone itself .like If the position calculation is correct, then proceed to the next step of trajectory generation; otherwise, trigger exception handling.

[0101] Step 8: Each slave device determines its row and column position. and reference interval parameters Calculate the expected relative position vector of each aircraft relative to the lead aircraft;

[0102] To optimize formation stability and wind resistance, this invention improves upon the traditional straight-line arrangement by introducing a "wave-shaped" or staggered layout, where the projections of adjacent drones in the X and Y directions are not perfectly aligned, further reducing the probability of collisions.

[0103] The expected relative position vector of each slave device The specific calculations are as follows:

[0104] Longitudinal position ( ): All formations extend row by row, with the lead machine defined as the origin, then: (The minus sign indicates that it is located behind the lead aircraft)

[0105] Altitude position ( The design employs a layered anti-collision system, with the layer height increasing as the number of rows increases. ;

[0106] Horizontal position ( ):

[0107] Basic lateral offset: First calculate the center point of the row, then calculate the nodes. offset ;

[0108] Wavy staggered correction: To prevent drones in front and behind rows from being in a straight line, add a horizontal offset of half a gap to even-numbered rows. If If it is even, then ;like If it is an odd number, then .

[0109] Step 9: Using the real-time status information of the master machine, combined with the expected relative position of each slave machine relative to the master machine, calculate the global expected position of each slave machine at the next moment.

[0110] Obtain the current Mercator plane coordinates of the lead aircraft. and heading angle (With true north as 0 degrees and clockwise as positive). Calculate the global expected position of each slave device at the next time step. :

[0111]

[0112] in: This is a rotation matrix used to align relative formations to the flight heading; This represents the current velocity vector of the lead aircraft. This is the estimated delay of the system control cycle, and it is used for feedforward compensation.

[0113] Step 10: Based on the improved distributed artificial potential field method, perform tracking control to guide each UAV to safely reach its respective global desired position and form the target formation.

[0114] To achieve safe and smooth trajectory tracking, an improved distributed artificial potential field method is used for closed-loop control.

[0115] First, for slave device s, define its total potential field function. :

[0116]

[0117] in This represents the current position of slave device s. For slave device s to point to its desired position gravitational field, For targeting nearby drones or obstacles of the slave device s The repulsive field.

[0118] Then, calculate the control force on slave device s. That is, the negative gradient of the potential field: ;

[0119] Finally, the tracking process is executed: based on control force. Calculate the desired acceleration of the slave device 's', input it into the underlying flight control system to adjust its speed and attitude, and achieve the desired coordinates. Accurate tracking, and the calculated planar coordinates The coordinates are converted back to latitude and longitude using the inverse Mercator projection for issuing actual navigation commands. Example

[0120] Task Scenario Description: A logistics park needs to utilize a swarm of drones to conduct coordinated inspections of the central warehouse and surrounding roads. A triangular formation provides a wide forward field of view and effectively covers the rectangular structure of the warehouse, while the lead drone at the apex of the swarm is guided by the salient drone, making it suitable for patrolling along a pre-defined route.

[0121] This embodiment performs automatic formation control based on the formation control type, X-direction spacing, Y-direction spacing, Z-direction spacing, and formation quantity. The position information of each UAV can be calculated and generated based on the position of the lead UAV and the offset of each UAV relative to the lead UAV.

[0122] Triangle formation generation and control based on this method:

[0123] Step 1: Formation Planning and Parameter Initialization

[0124] A triangular formation will be generated, and the inter-machine spacing reference parameters will be set. Considering the UAV's wingspan of approximately 1 meter and for safety margins, the parameters are preferably set as follows:

[0125] ;

[0126] in, This is the key design feature for achieving three-dimensional collision avoidance in this embodiment.

[0127] Step 2: Determine the cluster size

[0128] The number of drone nodes participating in the formation is determined to be N = 10 (1 leader drone and 9 slave drones).

[0129] Step 3: Global Coordinate Unification

[0130] Obtain the current GPS coordinates (e.g., longitude) of the lead aircraft (ID 1). ,latitude ,high ).according to Figure 2 The transformation relationship is used to convert it into plane coordinates using the Mercator projection formula.

[0131] Step 4: Calculate the number of rows r in the formation.

[0132] According to the formula for calculating the number of rows in a triangular formation:

[0133]

[0134] Calculations show that 4 rows are needed to accommodate 10 drones;

[0135] Step 5: Calculate the capacity of each row

[0136] Based on the characteristics of a triangle formation, the first row holds 1 drone (the lead drone), the second row holds 2 drones, the third row holds 3 drones, and the fourth row holds 4 drones. Since there are only 10 drones in total, the fourth row actually holds only 10 - (1+2+3) = 4 drones, which is just enough to fill the space.

[0137] Step 6: ID-based distributed row and column allocation

[0138] Sort all drones in ascending order by ID (1 to 10). ID=1 is the lead drone, fixed in row 1, column 1. The remaining drones are then assigned sequentially:

[0139] ID 2, 3 -> Row 2, Columns 1 and 2

[0140] ID 4, 5, 6 -> Row 3, Columns 1, 2, 3

[0141] IDs 7, 8, 9, 10 -> Row 4, Columns 1, 2, 3, 4

[0142] This process requires no inter-machine communication; each machine can independently calculate its location based solely on the known total number N and its own ID, demonstrating the distributed advantage of the method.

[0143] Step 7: Secondary Numbering

[0144] This step, consistent with the original ID order within the triangle formation, essentially verifies the correctness of the position allocation. For example, the drone located in row 4, column 2 (ID=8) can be verified using the triangle numbering formula: j(j-1) / 2 + k = 4 * 3 / 2 + 2 = 6 + 2 = 8

[0145] Step 8: Determine the relative position

[0146] Based on the set benchmark parameters Given the defined logical positions (j=4, k=2), calculate the relative position coordinates of UAV ID=8 relative to the lead aircraft.

[0147] Calculation process:

[0148] 1. Vertical offset Since it is in the 4th row, it is (j-1) body lengths behind the lead aircraft.

[0149]

[0150] 2. Vertical offset To achieve three-dimensional collision avoidance, it is (j-1) layers higher than the main aircraft.

[0151]

[0152] 3. Lateral offset This is the core of the "wave-shaped" arrangement. It is the second in this row (out of 4 aircraft).

[0153] First, calculate the leftmost position in this row, which is offset by (k-1) positions relative to the leftmost position:

[0154]

[0155] Thus, the precise offset vector of ID=8 relative to the lead aircraft is obtained: .

[0156] Step 9: Calculate the global desired position (from relative coordinates to absolute coordinates)

[0157] Objective: To convert relative coordinates into global absolute coordinates by combining the actual position and real-time motion status of the lead aircraft.

[0158] Calculation process:

[0159] 1. Obtain the lead aircraft status: Read the Mercator coordinates of the lead aircraft from step 3 from the flight control system. and current speed and heading angle .

[0160] 2. Coordinate Rotation: Because the formation needs to follow the lead aircraft's heading, the relative offset must be rotated. A rotation matrix is ​​used. Multiply by the relative position vector This gives the offset in the global coordinate system.

[0161] 3. Motion prediction: To compensate for control delay, the motion prediction of the lead aircraft in the next control cycle is performed. The position after:

[0162]

[0163] The same applies to other coordinates.

[0164] 4. Final synthesis position: The expected global position for ID=8 is:

[0165]

[0166]

[0167] Altitude typically does not rotate with heading.

[0168] This step involves "pasting" the static formation blueprint onto the dynamically moving lead aircraft.

[0169] Step 10: Tracking and Control Based on Artificial Potential Field (From "Target Coordinates" to "Flight Maneuvers")

[0170] Objective: To drive the robot safely and smoothly from its current position (which can be obtained via GPS / RTK and converted to Mercator coordinates) to the desired global position.

[0171] Control loop (showing closed loop):

[0172] 1. Gravity Calculation: Calculates the robot's current position. With the target location The vectors between (from step nine) generate a "gravity" pointing towards the target.

[0173] 2. Repulsion Calculation (Collision Avoidance): Real-time detection of the distance to other drones in the vicinity (target position and real-time position can be used as references). If the distance is less than the safe distance, a "repulsion force" is generated. This step dynamically ensures real-time collision avoidance even when executing formation commands in steps one through seven.

[0174] 3. Resultant force and control: The gravitational and repulsive forces are vector-synthesized to obtain a total control force, which is converted into acceleration and velocity commands and sent to the UAV's underlying flight controller (such as PIXHAWK).

[0175] 4. Cyclic execution: In each control cycle (e.g., 10 times per second), repeat steps nine (using the latest helm status) and ten to form a closed loop until the drone stabilizes at the target position.

Claims

1. A method for rapidly generating multi-UAV formations, characterized in that, The specific steps are as follows: Step 1: Determine the target formation of the UAVs according to the mission requirements, and set the three-dimensional spacing reference parameters between the UAVs. ; The target formation includes: triangle, rectangle, or wedge; Step 2: Determine the total number of drone nodes participating in the formation. Furthermore, a hierarchical strategy is introduced to divide the drone swarm; Step 3: Convert the latitude and longitude coordinates of the lead aircraft and each slave aircraft in the formation to Cartesian coordinates; The real-time planar coordinates of the lead aircraft are marked as follows ; Step 4: Based on the total number of drone nodes Given the target formation type, calculate the total number of rows required using a deterministic closed-form solution. : The target formation type is triangle: number of rows in the formation. satisfy The smallest integer solution is calculated using the following formula: ;in This represents the floor function; The target formation type is rectangular: set the fixed number of columns for the rectangular formation. The number of columns is determined by the task coverage width, then the number of rows... The calculation is as follows: ; The target formation type is a wedge-shaped or V-shaped formation: A wedge-shaped formation typically has one lead aircraft in the first row, with each subsequent row fixed at two aircraft. The number of rows... The calculation is as follows: ; Step 5: Calculate the maximum number of drone nodes that each row in the target formation type can accommodate; Step Six, for The drones are arranged in ascending order by ID, and each drone independently calculates its own row number. and column number ; With the first Taking a drone as an example: For triangular formations: by accumulating the previous... The total number of rows determines the position, if the condition is met. Then the machine is located at the th Row; its column number ; For rectangular arrays: row number ;column number ; in To round down, For modulo operation; For wedge formations: when hour, ;when Time: line number ;column number ; Step 7: Based on the row and column positions of each drone Generate a globally unique secondary number for each. And verify it; Secondary numbering for different formations The calculation formula is as follows: The calculation for a triangle formation is as follows: ; The calculation for rectangular formation is as follows: ; For wedge formations: the calculation is as follows: when At the time, it was the lead aircraft: ; when When it is the slave device: ; Step 8: Each slave device determines its row and column position. and reference interval parameters Calculate the expected relative position of each aircraft with respect to the lead aircraft. ; Step 9: Using the real-time status information of the master machine, combined with the expected relative position of each slave machine relative to the master machine, calculate the global expected position of each slave machine at the next moment. Obtain the current Mercator plane coordinates of the lead aircraft. and heading angle Calculate the global expected position of each slave device at the next time step. : ;in: It is a rotation matrix; This represents the current velocity vector of the lead aircraft. This is the estimated delay of the system control cycle; Step 10: Based on the improved distributed artificial potential field method, perform tracking control to guide each UAV to safely reach its respective global desired position and form the target formation.

2. The method as described in claim 1, characterized in that, The interval reference parameters in step one include: the safety interval reference between horizontal columns. Safety spacing benchmark between vertical rows Safety spacing benchmark between floors .

3. The method as described in claim 1, characterized in that, The layered strategy in step two specifically involves: first, setting a maximum node threshold for a single sub-cluster. ,like Then all drones are grouped together as a single cluster; if Then all drones are divided into Sub-clusters Each sub-cluster follows the same topology, and sub-clusters maintain a preset safety offset. .

4. The method as described in claim 1, characterized in that, In step three, it is assumed that the geographical coordinates of the lead aircraft or a slave aircraft are... Convert to Mercator plane coordinates The formula is as follows: ； in Longitude Latitude For height; This is the average radius of the Earth.

5. The method as described in claim 1, characterized in that, In step five, let the first... The maximum number of drone nodes that can be accommodated is ; For triangle formation: the The capacity of the row is ; For rectangular arrays: the capacity of each row is fixed, i.e. ; For wedge formations: first row Each subsequent line .

6. The method as described in claim 1, characterized in that, In step seven, the verification standard is: automatic comparison within the airborne computing unit. With the physical properties of the drone itself :like If the position calculation is correct, then proceed to the next step of trajectory generation; otherwise, trigger exception handling.

7. The method as described in claim 1, characterized in that, In step eight, the expected relative position vector of slave device s The calculation is as follows: Vertical position All formations extend row by row, with the lead aircraft as the origin. Then: ; Altitude position The design employs a layered anti-collision system, with the layer height increasing as the number of rows increases. ; Horizontal position The specific calculations are as follows: First, the basic horizontal offset: first calculate the center point of the row, then calculate the nodes. Offset: ; Then perform a wave-shaped alternation correction: if If it is even, then ;like If it is an odd number, then .

8. The method as described in claim 1, characterized in that, Step ten specifically involves: First, for slave device s, define its total potential field function. : ;in The current position of slave device s. For slave device s to point to the corresponding desired position gravitational field, For targeting nearby drones or obstacles of the slave device s The repulsive field; Then, calculate the control force on slave device s. That is, the negative gradient of the potential field: ; Finally, the tracking process is executed: based on control force. Calculate the desired acceleration of slave device s, input it into the underlying flight control system to adjust the speed and attitude of slave device s, so that it eventually reaches the desired coordinates. and the calculated plane coordinates Convert back to latitude and longitude coordinates using the inverse Mercator projection.

Images