A method for optimizing drone formation control in real urban scenarios

Abstract

This invention relates to the field of UAV cooperative control technology and discloses a UAV formation optimization control method for real urban scenarios, comprising the following steps: S1: Three-dimensional operation scene modeling to construct an urban low-altitude UAV flight scenario; S2: UAV formation modeling to construct a leader-follower-based UAV formation, adopting a leader-follower formation structure, where the leader is responsible for target guidance and the followers maintain the formation according to a preset relative offset, constructing three formations: regular pentagon, rhombus, and V-shape; S3: UAV formation control optimization to construct a UAV formation control method based on optimization strategies. This invention constructs a formation control mechanism that integrates formation holding force and neighbor coordination force, and introduces a flight dynamic weight adjustment mechanism and a three-dimensional obstacle avoidance and altitude constraint strategy, enabling UAV formations to achieve stable formation holding, safe obstacle avoidance, and smooth and efficient flight in complex urban environments.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology for unmanned aerial vehicles (UAVs), and more specifically, to a method for optimizing UAV formation control in real urban scenarios. Background Technology

[0002] Currently, in the field of drone formation control, existing solutions mainly include formation control methods based on traditional artificial potential fields (APF). This method achieves drone formation control by directly superimposing the attraction of the target, the repulsion of the obstacle, and the formation holding force. Due to its simple calculation, it is widely used in drone formation obstacle avoidance and path guidance. However, it is very easy to get trapped in local minima under complex obstacle distribution, and the drone formation is prone to deformation during obstacle avoidance, resulting in loose formation.

[0003] Traditional Dynamic Window (DWA)-based formation control methods achieve dynamic obstacle avoidance through velocity space search. In recent years, an improved Virtual Structure Dynamic Window (VS-DWA) method has been proposed. This method pre-defines the formation as a geometric structure and performs formation control through velocity space sampling and local prediction. However, it typically focuses more on individual drone feasibility and insufficiently considers the cooperative relationships within the formation, making it difficult to maintain a stable formation structure over long periods. Furthermore, control theory-based methods, such as the quadratic programming method combining control Lyapunov functions and obstacle control functions (CLF-CBF-QP), can strictly guarantee safety constraints in simple scenarios. However, as the number of obstacles and drones increases, its computational complexity rises significantly, and the formation trajectory often becomes conservative, leading to a significant increase in flight path and energy consumption.

[0004] Therefore, although existing technical solutions have achieved certain results in specific scenarios, they still have significant shortcomings in complex urban 3D environments and formation control tasks: First, they are prone to disrupting the established formation structure during obstacle avoidance, resulting in loose formations or individual members separating; second, they are prone to path oscillations, rapid ascents, or discontinuous control in densely built areas, affecting flight safety and energy consumption; and third, they do not adequately consider the altitude constraints and safety margins in real urban 3D environments, making it difficult to meet the needs of actual engineering applications. Summary of the Invention

[0005] The technical problem to be solved by this invention is to provide a drone formation optimization control method for real urban scenarios. By constructing a formation control mechanism that integrates formation holding force and neighbor coordination force, and introducing a flight dynamic weight adjustment mechanism and a three-dimensional obstacle avoidance and altitude constraint strategy, the drone formation can achieve stable formation holding, safe obstacle avoidance and smooth and efficient flight in complex urban environments.

[0006] The present invention achieves its objective by employing the following technical solution: A method for optimizing drone formation control in real-world urban scenarios, characterized by the following steps: S1: 3D operational scene modeling to construct urban low-altitude drone flight scenarios; S2: Drone formation modeling, constructing a leader-follower drone formation, adopting a leader-follower formation structure, where the leader is responsible for target guidance, and the followers maintain the formation according to the preset relative offset, constructing three formations: regular pentagon, rhombus, and V-shape; S3: UAV formation control optimization. Construct an UAV formation control method based on optimization strategy. The UAV formation control optimization strategy consists of constructing an optimized potential field, introducing a flight dynamic weight adjustment mechanism, improving three-dimensional obstacle avoidance and altitude constraint control, and iteratively updating the UAV state.

[0007] As a further limitation of this technical solution, the leader-follower model is as follows: (1) (2) in: P i (t) is the th i The ideal target position of the drone at time t; P L (t) represents the leader's position vector; r i For the first i The ideal relative offset vector of the followers relative to the leader; In equation (2), take r i,z = 0 is used to maintain a planar formation, requiring that the initial formation height be consistent, all drones be on the same plane, and the regular pentagonal formation have a ring-shaped vertex distribution. (Index) i =2,…,6 correspond to five vertices, and the formation radius is... R form Let the initial rotation angle be... 0 is used to adjust the orientation, making a vertex face forward or rotating the initial formation clockwise. j One follower, making j = i -1, defines the plane offset: (3) (4).

[0008] As a further limitation of this technical solution, a diamond formation of four UAVs is constructed using basic vectors; First, define four base points in the local coordinate system. Then, the set of basic two-dimensional basis vectors is: (5) Then through a fixed rotation angle i Obtain the direction unit vector: (6) Among them: || || represents the norm of a vector, i.e., the magnitude of the vector; (7) Pick i =-π / 2, direction element vector is dirs, rotation matrix is... R θ ; For the arrangement of the four drones, take the first, third, and fourth rows of the follower set as the base, that is, the three followers are distributed in the upper right, lower left, and lower right points respectively, and then multiply by . R form Then we get: (8) in: k ∈{1,3,4} corresponds to the selected row number.

[0009] As a further limitation of this technical solution, the V-shaped formation consists of five drones, with the leader located at the apex of the V and the followers symmetrically extending backward on the left and right wings. Let the vertical interval between each level be The horizontal spacing is d w Then for the first k For the wing, we get two offsets, left and right: (9) (10) (11) in: i It represents the V-shaped angle.

[0010] As a further limitation of this technical solution, the control process of S3 is as follows: S31: Introducing a formation control mechanism that integrates formation maintenance and neighbor coordination, enabling UAVs to maintain the desired formation position while enhancing local consistency through neighborhood information, as follows: (12) in: This indicates that the new formation maintains its strength. F attIndicates the attractiveness of the target point; F rep Indicates the repulsive force of an obstacle; Indicates the total resultant force; The new formation maintains its strength. Defined as: (13) in: real deal This indicates the desired position of the drone relative to the leader within a pre-defined formation; p Indicates the current actual location of the drone; This indicates the average position of other drones within the neighborhood; k 1. k 2 represents the formation maintenance gain coefficient and the neighbor coordination gain coefficient, respectively; The dynamic update equation for the UAV is: (14) Where: v i To indicate the first i The velocity vector of the drone; S32: Based on the distance between the UAV and the target point, the distance to the obstacle, and the formation error, the weights of the target attraction force, the obstacle repulsion force, and the formation keeping force are adaptively adjusted to achieve smooth switching of control strategies in different flight phases; During three-dimensional flight, the mission of a UAV can be divided into three phases: S321: Long-distance guidance phase: The formation needs to quickly approach the target point. At this time, the attraction weight should be increased and the repulsion force and formation constraints should be reduced. S322: Mid-range coordination phase: As the formation gradually enters complex terrain areas, it should maintain a stable formation while taking obstacle avoidance into account; S323: Precise Phase Near Target: As the leader approaches the target point, it is necessary to enhance formation maintenance and high degree of discipline to ensure that the entire formation arrives in unison. Define the weight coefficients of each item as the target distance. dg Distance to nearest obstacle do Functions: (15) in: Katt , Crepe , Kform These are the dynamic weights of attraction, repulsion, and formation holding force, respectively. α , β , cThis is an adjustment coefficient used to control the convergence speed and sensitivity to weight changes; pgoal Indicates the location of the target point; pobs, j Indicates the location of the obstacle; Kmax att Indicates the weight of maximum attraction; Kmax rep Indicates the weight of the maximum repulsive force; Kmax form This indicates that the largest formation retains its weight; Kmin form This indicates that the smallest team maintains its weight; || p - p goal || represents calculating the magnitude of a vector, i.e., the distance between the UAV and the target point. d g Length; || p - p obs,j || represents calculating the magnitude of a vector, i.e., the distance between the drone and the obstacle, and then... minutes j Take the minimum distance, i.e., the nearest obstacle distance. d o ; To ensure the continuity and smoothness of the formation control system, a normalized flight dynamics weight fusion mechanism is further introduced: (16) in: oh att Indicates the attractiveness weighting factor; oh rep This represents the repulsive force weighting factor; oh form This indicates that the formation maintains the weighting factor; The optimized resultant force expression for the UAV is defined as follows: (17) S33: Combining three-dimensional spatial distance and height information, a three-dimensional obstacle avoidance potential field is constructed, and a vertical repulsive force is applied when the drone's altitude is below the safety threshold to ensure that the drone safely crosses obstacles and avoids drastic altitude fluctuations;

[0011] S331: The UAV position p is defined using three-dimensional Euclidean distance. x , y , z ] T With any 3D point p in the obstacle point cloud obs=[ xobs , yobs , zobs ] T Spatial distance: (18) S332: To reflect the repulsive effect of obstacles on the drone, a three-dimensional repulsive potential energy is defined based on the artificial potential field theory: (19) in: or The repulsive potential field gain coefficient controls the strength of the repulsive force. dr The effective radius of influence of the obstacle; S333: Then, the three-dimensional repulsive force is obtained from the potential energy gradient: (20) in: Represents the gradient operator; Indicates the location p At that point, the direction and rate of the fastest increase in repulsive potential energy are repelled; S334: Further, based on digital surface model elevation data h ( x , y Construct height constraints: (twenty one) in: z ( t ) indicates the drone at a certain time t The vertical position; h ( x , y This corresponds to the actual ground height at that location; Δ h For safety margin, it is used to resist systematic errors and air disturbances.

[0012] S335: To integrate this constraint into the potential field optimization framework in a continuous manner, a highly repulsive potential is defined: (twenty two) in: k Weights are for the vertical direction; The corresponding vertical repulsive force is: (twenty three) S336: Combining horizontal and vertical repulsion forces, the total three-dimensional obstacle avoidance force of a UAV is defined as follows: (twenty four) in: Nobs Indicates the number of obstacle point clouds; F( j ) rep Indicates the first j The three-dimensional repulsive force of an obstacle; The upward vertical thrust is solely due to insufficient height. S34: The above force terms are weighted and synthesized into the UAV control input, and the speed and position of the UAV are updated until the formation as a whole safely reaches the target point and restores the preset formation.

[0013] Compared with related technologies, the UAV formation optimization control method provided by this invention for real urban scenarios has the following beneficial effects: (1) This invention uses real high-precision GIS data to model the scene, constructs a leader-follower-based UAV formation, introduces a formation control mechanism that combines formation maintenance force and neighbor coordination force, and combines a flight dynamic weight adjustment mechanism and a three-dimensional obstacle avoidance and altitude constraint strategy to enable UAV formation to maintain formation, avoid obstacles safely and fly smoothly and efficiently in complex urban environments, thereby improving the safety, stability, environmental adaptability and engineering universality of UAV formation control method.

[0014] (2) Based on the traditional potential field control model, this invention introduces a formation control mechanism that integrates formation holding force and neighbor coordination force. By simultaneously constraining the geometric positional relationship of UAVs in the desired formation structure and their local neighborhood coordination relationship, it effectively improves the overall consistency of the formation and suppresses individual decoupling and formation deformation. At the same time, a dynamic weight adjustment mechanism based on the flight phase is constructed. According to the spatial relationship between the UAV and the target point, obstacles and the formation center, the intensity of the interaction between target guidance, obstacle avoidance control and formation holding is adaptively adjusted, thereby forming a continuous control system that takes into account both global guidance and local coordination. In addition, this invention combines real urban three-dimensional environmental information to establish a three-dimensional obstacle avoidance and height control model that includes building height and safety margin constraints, enabling UAV formations to achieve safe and smooth cooperative flight in complex building environments. Therefore, the protection scope of this invention not only covers the above-mentioned control modules themselves, but also their cooperative operation mode and overall control framework in real urban scenarios. It is a key technical solution for achieving stable formation holding and safe obstacle avoidance of UAVs in complex urban environments. Attached Figure Description

[0015] Figure 1 This is a flowchart of the present invention.

[0016] Figure 2 This is a schematic diagram of a typical urban area according to the present invention.

[0017] Figure 3This is a typical urban area visualization code for the present invention.

[0018] Figure 4 This is a typical three-dimensional model of an urban area according to the present invention.

[0019] Figure 5 This is a schematic diagram of the regular pentagonal formation of the present invention.

[0020] Figure 6 This is a schematic diagram of the drone diamond formation according to the present invention.

[0021] Figure 7 This is a schematic diagram of the V-shaped formation of drones according to the present invention.

[0022] Figure 8 This is a flowchart of the drone formation control process of the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] Existing drone formation control technologies still have significant shortcomings in real-world urban scenarios. On the one hand, most methods prioritize individual safety during obstacle avoidance, neglecting overall formation consistency, which can easily lead to loose formations or isolated drones, resulting in insufficient formation stability. On the other hand, in complex urban environments, some methods exhibit uneven flight paths, such as path oscillations, sharp ascents, or frequent turns, significantly increasing energy consumption and impacting flight efficiency. Regarding formation adaptability, most existing technologies are based on simplified, idealized environmental models, failing to adequately consider building height, terrain undulations, and safety altitudes. They lack validation in real, high-resolution urban environments, limiting their engineering applicability and making it difficult to guarantee reliability and effectiveness. Furthermore, existing solutions often use fixed control parameters, making it difficult to dynamically adjust the trade-offs between target guidance, obstacle avoidance, and formation maintenance according to flight phases.

[0025] A method for optimizing drone formation control in real-world urban scenarios includes the following steps: S1: 3D operational scene modeling to construct urban low-altitude drone flight scenarios.

[0026] First, a typical urban area was selected as the scenario for the drone formation mission. The coverage area of ​​this area is approximately 5km × 5km. Figure 2 As shown.

[0027] Subsequently, a 3D urban model was constructed based on high-precision GIS data and building distribution data. The model's matrix size was 500×500, and building heights were visualized using pseudo-color encoding, ranging from 0 to 40 meters. Figure 4 As shown in the diagram, in the height matrix, areas with a ground elevation greater than 0m are defined as obstacle areas.

[0028] S2: Drone formation modeling, constructing a leader-follower drone formation based on a leader-follower structure. The leader is responsible for target guidance, and the followers maintain the formation according to the preset relative offset. Three formations are constructed: regular pentagon, rhombus, and V-shape.

[0029] The leader-follower model is as follows: (1) (2) in: P i (t) is the th i The ideal target position of the drone at time t; P L (t) represents the leader's position vector; r i For the first i The ideal relative offset vector of the followers relative to the leader is determined by... r i With proper design, formations of different shapes and sizes can be achieved; In equation 2, take r i,z = 0 is used to maintain a planar formation, requiring that the initial formation height be consistent, all drones be on the same plane, and the regular pentagonal formation have a ring-shaped vertex distribution. (Index) i =2,…,6 correspond to five vertices, and the formation radius is... R form Let the initial rotation angle be... 0 is used to adjust the orientation, making a vertex face forward or rotating the initial formation clockwise. j One follower, making j = i -1, defines the plane offset: (3) (4).

[0030] Construct a diamond formation of four drones using basic vectors; First, define four base points in the local coordinate system. Then, the set of basic two-dimensional basis vectors is: (5) Then through a fixed rotation angle i Obtain the direction unit vector: (6) Among them: || || represents the norm of a vector, i.e., the magnitude of the vector; (7) Pick i =-π / 2, direction element vector is dirs, rotation matrix is... R θ ; For the arrangement of the four drones, take the first, third, and fourth rows of the follower set as the base, that is, the three followers are distributed in the upper right, lower left, and lower right points respectively, and then multiply by . R form Then we get: (8) in: k ∈{1,3,4} corresponds to the selected row number. Therefore... r i The design is complete, resulting in a symmetrical rhomboid structure, with each follower positioned horizontally at a distance of [missing information]. R form It can be adjusted according to the overall formation compactness requirements. R form The size of the four drones in a diamond formation is shown in Figure 6.

[0031] The V-shaped formation consists of five drones, with the leader at the apex of the V and the followers spreading out symmetrically to the left and right flanks. Let the vertical interval between each level be The horizontal spacing is d w Then for the first k For the wing, we get two offsets, left and right: (9) (10) (11) in: i Indicates the V-angle, which can be adjusted... d ℓ and d w The V-angle can be adjusted; to achieve a smaller angle, the lateral spacing can be reduced. d w Or increase the longitudinal spacing Finally, a unified rotation can be performed to adjust the direction of the "V". A diagram of the V-shaped formation is shown below. Figure 7 As shown.

[0032] S3: UAV formation control optimization. Construct an UAV formation control method based on optimization strategy. The UAV formation control optimization strategy consists of constructing an optimized potential field, introducing a flight dynamic weight adjustment mechanism, improving three-dimensional obstacle avoidance and altitude constraint control, and iteratively updating the UAV state.

[0033] The control process of S3 is as follows: S31: First, based on the traditional artificial potential field model, a formation control mechanism that integrates formation maintenance force and neighbor coordination force is introduced, enabling UAVs to maintain the desired formation position and enhance local consistency through neighborhood information, as follows: (12) in: This indicates the new formation holding force term, used to maintain the stability of the overall formation shape; F att It indicates the attraction of the target point and guides the drone to move towards the target point; F rep This represents the repulsive force of obstacles, preventing drones from colliding with them; Indicates the total resultant force; The new formation maintains its strength. Defined as: (13) in: real deal This indicates the desired position of the drone relative to the leader within a pre-defined formation; p Indicates the current actual location of the drone; This indicates the average position of other drones within the neighborhood; k 1. k 2 represents the formation maintenance gain coefficient and the neighbor coordination gain coefficient, respectively; First component k 1( p ideal - p This reflects macroscopic formation constraints, ensuring that drones can maintain operation within a pre-defined geometric structure, thereby achieving spatial position maintenance within a leader-follower system. The second component... k 2( p neighbor - p This reflects local coordination constraints. By providing feedback on the average position of neighboring individuals, it enhances the spatial consistency and coordination capabilities among UAVs, and reduces formation deformation and local isolation.

[0034] The dynamic update equation for the UAV is: (14) Where: v i To indicate the first i The velocity vector of the drone; S32: Based on the distance between the UAV and the target point, the distance to the obstacle, and the formation error, the weights of the target attraction force, the obstacle repulsion force, and the formation keeping force are adaptively adjusted to achieve smooth switching of control strategies in different flight phases; During three-dimensional flight, the mission of a UAV can be divided into three phases: S321: Long-distance guidance phase: The formation needs to quickly approach the target point. At this time, the attraction weight should be increased and the repulsion force and formation constraints should be reduced. S322: Mid-range coordination phase: As the formation gradually enters complex terrain areas, it should maintain a stable formation while taking obstacle avoidance into account; S323: Precise Phase Near Target: As the leader approaches the target point, it is necessary to enhance formation maintenance and high degree of discipline to ensure that the entire formation arrives in unison. Define the weight coefficients of each item as the target distance. dg Distance to nearest obstacle do Functions: (15) in: Katt , Crepe , Kform These are the dynamic weights of attraction, repulsion, and formation holding force, respectively. α , β , c This is an adjustment coefficient used to control the convergence speed and sensitivity to weight changes; pgoal Indicates the location of the target point; pobs, j Indicates the location of the obstacle; Kmax att Indicates the weight of maximum attraction; Kmax rep Indicates the weight of the maximum repulsive force; Kmax form This indicates that the largest formation retains its weight; Kmin form This indicates that the smallest team maintains its weight; || p - p goal || represents calculating the magnitude of a vector, i.e., the distance between the UAV and the target point. d g Length; || p -p obs,j || represents calculating the magnitude of a vector, i.e., the distance between the drone and the obstacle, and then... minutes j Take the minimum distance, i.e., the nearest obstacle distance. d o ; When the drone moves away from the obstacle ( d o → ∞), the repulsive force weight approaches zero, while the attractive force weight approaches its maximum value. Conversely, when approaching an obstacle, the repulsive force term will be significantly amplified; at the same time, as the drone gradually approaches the target point ( d g →0), Formation Maintenance Item K form This will enhance and ensure the accurate restoration of the formation structure.

[0035] To ensure the continuity and smoothness of the formation control system, a normalized flight dynamics weight fusion mechanism is further introduced: (16) in: oh att This represents the attraction weighting factor, which is obtained by normalizing the dynamic attraction weights. oh rep This represents the repulsive force weighting factor, which is obtained by normalizing the dynamic weighting of the repulsive force. oh form This indicates the formation holding weight factor, which is obtained by normalizing the dynamic weight of the formation holding force. The optimized resultant force expression for the UAV is defined as follows: (17) When the formation is in an open area oh att The dominant motion ensures rapid global convergence; when the drone approaches tall buildings or terrain protrusions... oh rep Automatic enhancement causes a natural deviation in the individual trajectory, enabling flexible obstacle avoidance; as the leader approaches the target point... oh form The dynamic magnification causes the followers to recover to the ideal formation in three-dimensional space, achieving a stable landing or hovering.

[0036] S33: Combining three-dimensional spatial distance and altitude information, a three-dimensional obstacle avoidance potential field is constructed, and a vertical repulsive force is applied when the drone's altitude is below the safety threshold to ensure that the drone safely crosses obstacles and avoids drastic altitude fluctuations.

[0037] S331: Traditional artificial potential fields mainly rely on two-dimensional planar distances to construct repulsive forces, which are difficult to handle the numerous vertical structures present in real buildings. Therefore, this invention uses three-dimensional Euclidean distance to define the UAV position p=[ x , y , z ] T With any 3D point p in the obstacle point cloud obs =[ xobs , yobs , zobs ] T Spatial distance: (18) S332: To reflect the repulsive effect of obstacles on the drone, a three-dimensional repulsive potential energy is defined based on the artificial potential field theory: (19) in: or The repulsive potential field gain coefficient controls the strength of the repulsive force. dr The effective radius of influence of the obstacle; When the drone is far from the obstacle d At 0, the repulsive potential energy disappears naturally, but when the drone approaches the obstacle, the potential energy increases exponentially, ensuring that the repulsive force is continuous, differentiable, and increases rapidly.

[0038] S333: Then, the three-dimensional repulsive force is obtained from the potential energy gradient: (20) in: Represents the gradient operator; Indicates the location p The direction and rate of fastest increase in repulsive potential energy; repulsive force. F rep It equals the negative gradient of potential energy, that is, the direction of the force is the direction in which the potential energy decreases the fastest (here, the repulsive force will push the drone away from the obstacle, corresponding to a decrease in potential energy).

[0039] Three-dimensional repulsive force F rep The direction is always in the opposite direction to the obstacle the drone is pointing, and the smaller the distance, the greater the repulsive force. This model enables the drone not only to move away from the horizontal projection of the obstacle, but also to automatically correct its flight trajectory upward or downward according to the height of the obstacle to bypass it.

[0040] S334: Further, based on Digital Surface Model (DSM) elevation data h ( x ,y Construct height constraints: (twenty one) in: z ( t ) indicates the drone at a certain time t The vertical position, i.e., the flight altitude; h ( x , y This corresponds to the actual ground height at that location; Δ h For safety margin, it is used to resist systematic errors and air disturbances.

[0041] S335: To integrate this constraint into the potential field optimization framework in a continuous manner, a highly repulsive potential is defined: (twenty two) in: k The vertical weight determines the conservatism of the system. When the drone's altitude is insufficient, i.e. below the safe altitude, a positive vertical repulsive force occurs, pushing the drone upward. The corresponding vertical repulsive force is: (twenty three) When the drone's altitude exceeds the safety boundary area, the repulsive force automatically returns to zero, preventing interference with flight. Compared to directly truncating the altitude, this flexible potential field avoids sudden jumps or altitude oscillations in the drone, better aligning with actual aircraft dynamics.

[0042] S336: Combining horizontal and vertical repulsion forces, the total three-dimensional obstacle avoidance force of a UAV is defined as follows: (twenty four) in: Nobs Indicates the number of obstacle point clouds; Indicates the first j The three-dimensional repulsive force of an obstacle; The vertical upward thrust is generated solely by insufficient altitude; the combined effect of these two factors allows the drone to automatically select either a "fly around" or "jump" strategy, enabling flexible obstacle avoidance. S34: The above force terms are weighted and synthesized into the UAV control input, and the speed and position of the UAV are updated until the formation as a whole safely reaches the target point and restores the preset formation.

[0043] The proposed method is compared with the traditional APF method, VS-DWA method, and CLF-CBF-QP method. The experimental results are shown in the table below.

[0044] Table 1: Method Comparison Table

[0045] Table 1 shows that the present invention performs best in terms of path length, obstacle avoidance performance, and formation keeping accuracy.

[0046] The experimental three-dimensional results of this scheme and method, including the Artificial Potential Field (APF) method, the Virtual Structure-Dynamic Window Approach (VS-DWA) method, and the Control Lyapunov Function-Control Barrier Function-Quadratic Programming (CLF-CBF-QP method), are presented. As can be seen from the three-dimensional trajectory diagram, the path generated by this invention is smoother, and the formation remains more stable.

[0047] This invention offers significant advantages in terms of environmental adaptability, formation stability, and flight safety in real-world urban scenarios. Traditional UAV formation control methods are often based on simplified environmental models or two-dimensional assumptions, making it difficult to simultaneously account for building height constraints, complex obstacle distributions, and the collaborative relationships among multiple UAVs. This can easily lead to formation disruption or discontinuous flight trajectories during obstacle avoidance. This invention introduces a formation control mechanism that integrates formation-keeping force and neighbor coordination force, effectively enhancing the collaboration among UAVs in complex three-dimensional urban scenarios, enabling the formation to maintain a stable structure during obstacle avoidance and path adjustment. Regarding control strategies, compared to existing methods that often employ fixed control parameters or a single control objective, this invention constructs a dynamic weight adjustment mechanism based on flight phases. This mechanism adaptively adjusts the control action according to changes in the relationship between the UAV and the target point, obstacles, and formation errors, thereby avoiding path oscillations caused by excessive obstacle avoidance or frequent maneuvers, resulting in smoother flight trajectories and lower energy consumption. Simultaneously, this invention fully integrates real-world three-dimensional urban environmental information, introducing height constraints and safety margin control, effectively improving the safety and feasibility of UAV formations in densely built-up areas. Furthermore, this invention maintains good robustness and scalability even when the size of the drone formation expands and the formation structure changes, achieving stable collaborative control while keeping computational complexity low. Therefore, in real-world urban drone formation missions, this invention demonstrates superior overall performance compared to existing best-in-class technologies in terms of safety, stability, and engineering applicability.

[0048] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for optimizing the control of drone formations in real urban scenarios, characterized in that, Includes the following steps: S1: 3D operational scene modeling to construct urban low-altitude drone flight scenarios; S2: Drone formation modeling, constructing a leader-follower drone formation, adopting a leader-follower formation structure, where the leader is responsible for target guidance, and the followers maintain the formation according to the preset relative offset, constructing three formations: regular pentagon, rhombus, and V-shape; S3: UAV formation control optimization. Construct an UAV formation control method based on optimization strategy. The UAV formation control optimization strategy consists of constructing an optimized potential field, introducing a flight dynamic weight adjustment mechanism, improving three-dimensional obstacle avoidance and altitude constraint control, and iteratively updating the UAV state.

2. The method for optimizing drone formation control in real urban scenarios according to claim 1, characterized in that: The leader-follower model is as follows: （1） （2） in: P i (t) is the th i The ideal target position of the drone at time t; P L (t) represents the leader's position vector; r i For the first i The ideal relative offset vector of the followers relative to the leader; In equation (2), take r i,z = 0 is used to maintain a planar formation, requiring that the initial formation height be consistent, all drones be on the same plane, and the regular pentagonal formation have a ring-shaped vertex distribution. (Index) i =2,…,6 correspond to five vertices, and the formation radius is... R form Let the initial rotation angle be... 0 is used to adjust the orientation, making a vertex face forward or rotating the initial formation clockwise. j One follower, making j = i -1, defines the plane offset: （3） （4）。 3. The method for optimizing drone formation control in real urban scenarios according to claim 2, characterized in that: Construct a diamond formation of four drones using basic vectors; First, define four base points in the local coordinate system. Then, the set of basic two-dimensional basis vectors is: （5） Then through a fixed rotation angle θ Obtain the direction unit vector: （6） Among them: || || represents the norm of a vector, i.e., the magnitude of the vector; （7） Pick θ =-π / 2, direction element vector is dirs, rotation matrix is... R θ ; For the arrangement of the four drones, take the first, third, and fourth rows of the follower set as the base, that is, the three followers are distributed in the upper right, lower left, and lower right points respectively, and then multiply by . R form Then we get: （8） in: k ∈{1,3,4} corresponds to the selected row number.

4. The method for optimizing drone formation control in real urban scenarios according to claim 3, characterized in that: V The V-shaped formation consists of five drones, with the leader positioned at the apex of the V, and the followers spreading out symmetrically to the left and right flanks. Let the vertical interval between each level be The horizontal spacing is d w Then for the first k For the wing, we get two offsets, left and right: （9） （10） （11） in: θ It represents the V-shaped angle.

5. The method for optimizing drone formation control in real urban scenarios according to claim 1, characterized in that: The control process of S3 is as follows: S31: Introducing a formation control mechanism that integrates formation maintenance and neighbor coordination, enabling UAVs to maintain the desired formation position while enhancing local consistency through neighborhood information, as follows: （12） in: This indicates that the new formation maintains its strength. F att Indicates the attractiveness of the target point; F rep Indicates the repulsive force of an obstacle; Indicates the total resultant force; The new formation maintains its strength. Defined as: （13） in: pideal This indicates the desired position of the drone relative to the leader within a pre-defined formation; p Indicates the current actual location of the drone; This indicates the average position of other drones within the neighborhood; k 1. k 2 represents the formation maintenance gain coefficient and the neighbor coordination gain coefficient, respectively; The dynamic update equation for the UAV is: （14） Where: v i To indicate the first i The velocity vector of the drone; S32: Based on the distance between the UAV and the target point, the distance to the obstacle, and the formation error, the weights of the target attraction force, the obstacle repulsion force, and the formation keeping force are adaptively adjusted to achieve smooth switching of control strategies in different flight phases; During three-dimensional flight, the mission of a UAV can be divided into three phases: S321: Long-distance guidance phase: The formation needs to quickly approach the target point. At this time, the attraction weight should be increased and the repulsion force and formation constraints should be reduced. S322: Mid-range coordination phase: As the formation gradually enters complex terrain areas, it should maintain a stable formation while taking obstacle avoidance into account; S323: Precise Phase Near Target: As the leader approaches the target point, it is necessary to enhance formation maintenance and high degree of discipline to ensure that the entire formation arrives in unison. Define the weight coefficients of each item as the target distance. dg Distance to nearest obstacle do Functions: （15） in: Katt , Krep , Kform These are the dynamic weights of attraction, repulsion, and formation holding force, respectively. α , β , γ This is an adjustment coefficient used to control the convergence speed and sensitivity to weight changes; pgoal Indicates the location of the target point; pobs, j Indicates the location of the obstacle; Kmax att Indicates the weight of maximum attraction; Kmax rep Indicates the weight of the maximum repulsive force; Kmax form This indicates that the largest formation retains its weight; Kmin form This indicates that the smallest team maintains its weight; || p - p goal || represents calculating the magnitude of a vector, i.e., the distance between the UAV and the target point. d g Length; || p - p obs,j || represents calculating the magnitude of a vector, i.e., the distance between the drone and the obstacle, and then... min j Take the minimum distance, i.e., the nearest obstacle distance. d o ; To ensure the continuity and smoothness of the formation control system, a normalized flight dynamics weight fusion mechanism is further introduced: （16） in: ω att Indicates the attractiveness weighting factor; ω rep This represents the repulsive force weighting factor; ω form This indicates that the formation maintains the weighting factor; The optimized resultant force expression for the UAV is defined as follows: （17） S33: Combining three-dimensional spatial distance and height information, a three-dimensional obstacle avoidance potential field is constructed, and a vertical repulsive force is applied when the drone's altitude is below the safety threshold to ensure that the drone safely crosses obstacles and avoids drastic altitude fluctuations; S331: The UAV position p is defined using three-dimensional Euclidean distance. x , y , z ] T With any 3D point p in the obstacle point cloud obs =[ xobs , yobs , zobs ] T Spatial distance: （18） S332: To reflect the repulsive effect of obstacles on the drone, a three-dimensional repulsive potential energy is defined based on the artificial potential field theory: （19） in: η The repulsive potential gain coefficient controls the strength of the repulsive force. dr The effective radius of influence of the obstacle; S333: Then, the three-dimensional repulsive force is obtained from the potential energy gradient: （20） in: Represents the gradient operator; Indicates the location p At that point, the direction and rate of the fastest increase in repulsive potential energy are repelled; S334: Further, based on digital surface model elevation data h ( x , y Construct height constraints: （21） in: z ( t ) indicates the drone at time t The vertical position; h ( x , y This corresponds to the actual ground height at that location; Δ h To provide a safety margin to resist system errors and air disturbances; S335: To integrate this constraint into the potential field optimization framework in a continuous manner, a highly repulsive potential is defined: （22） in: κ Weights are for the vertical direction; The corresponding vertical repulsive force is: （23） S336: Combining horizontal and vertical repulsion forces, the total three-dimensional obstacle avoidance force of a UAV is defined as follows: （24） in: Nobs Indicates the number of obstacle point clouds; F( j ) rep Indicates the first j The three-dimensional repulsive force of an obstacle; The upward vertical thrust is solely due to insufficient height. S34: The above force terms are weighted and synthesized into the UAV control input, and the speed and position of the UAV are updated until the formation as a whole safely reaches the target point and restores the preset formation.

Images