A heterogeneous UAV swarm interactive clustering formation and trajectory planning method

Abstract

This invention discloses a heterogeneous UAV swarm interactive clustering formation and trajectory planning method, comprising the following steps: The UAV swarm is clustered according to the task to be processed and the dynamic characteristics of each UAV within the swarm, with each cluster corresponding to one of the tasks to be processed; for each cluster, the target position of its corresponding task, the positions of identified obstacles, and the current positions of each UAV within the cluster are obtained; based on the target position, obstacle positions, and current UAV positions, the gravitational and repulsive forces experienced by each UAV at its current position are calculated using gravitational and repulsive functions, wherein the gravitational and repulsive functions are obtained using an improved artificial potential field function; and each UAV is driven by the resultant force of the gravitational and repulsive forces experienced by it at its current position.

Description

Technical fields:

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) swarm control and trajectory planning, and discloses a method for interactive clustering and trajectory planning of heterogeneous UAV swarms. Background technology:

[0002] Unmanned aerial vehicles (UAVs) possess advantages such as high flexibility, good maneuverability, fast response speed, and low operational requirements, enabling them to replace manned aircraft in performing complex and dangerous tasks to a certain extent. However, limited by factors such as the UAV's own energy, capabilities, and payload, a single UAV is insufficient to cope with complex and ever-changing environments and autonomously complete tasks. Therefore, UAV swarm collaborative operations have become a trend. With the rapid development of new-generation information technologies such as computers, networks, and communications, multi-UAV swarm technology has increasingly broad and important military and civilian value, especially in multi-target, multi-mission application scenarios, such as collaborative reconnaissance, saturation strikes, precision agriculture, and environmental monitoring. However, homogeneous swarms composed of single-function UAVs still have significant limitations in terms of payload and functionality. When performing multi-target control tasks, multiple heterogeneous UAVs often need to collaborate. Moreover, existing UAV swarms face challenges such as fixed cluster formation, preventing them from adjusting to changes in tasks or completion levels, which severely limits their effectiveness in scenarios with variable targets or tasks. In addition, the flight path planning process of drone swarms has problems such as unreachable targets, local minima traps, and excessively long paths. It is impossible to determine whether a drone should continue to participate and complete the mission based on its real-time status, thus preventing the drone swarm from achieving its maximum effectiveness.

[0003] In response to this situation, there is an urgent need to develop a safe, reliable, and operable method for heterogeneous UAV swarm interactive clustering and flight path planning to meet the requirements of real-world mission scenarios. Therefore, this invention aims to solve existing technical challenges, improve the efficiency of UAV swarms in intelligently completing tasks, and enhance the quality of UAV swarms in accomplishing complex multi-objective tasks. Summary of the Invention:

[0004] This invention aims to at least partially solve one of the aforementioned technical problems or at least provide a useful construction method. To this end, this invention proposes a method for interactive clustering and flight path planning of heterogeneous UAV swarms with security, reliability, and operability, enabling heterogeneous UAV swarms with different functions to autonomously complete multi-target control tasks.

[0005] According to one aspect of the present invention, a method for interactive clustering and flight path planning of heterogeneous unmanned aerial vehicle (UAV) swarms is provided, comprising the following steps: clustering the UAV swarm according to the task to be processed and the dynamic characteristics of each UAV within the swarm, wherein each cluster corresponds to one of the tasks to be processed; for each cluster, obtaining the target position of its corresponding task, the position of identified obstacles, and the current position of each UAV within the cluster; calculating the gravitational and repulsive forces acting on each UAV at its current position using a gravitational function and a repulsive function based on the target position, obstacle position, and current UAV position, wherein the gravitational and repulsive functions are obtained using an improved artificial potential field function; and driving each UAV based on the resultant force of the gravitational and repulsive forces acting on each UAV at its current position.

[0006] According to another aspect of the present invention, a method for interactive clustering and flight path planning of heterogeneous unmanned aerial vehicle (UAV) swarms is provided, comprising: Step 1: setting initial parameters of the heterogeneous UAV swarm, such as the number of swarms and swarm performance; determining the mission type, mission nature, and permissible formation form of the UAV swarm; Step 2: based on the preliminary mission planning, autonomous clustering of the swarm is achieved according to the mission requirements and different functions of the UAV swarm, following the principle of balanced partitioning; Step 3: based on the clustering forms corresponding to different mission natures, flight path planning is performed using an improved artificial potential field method to effectively solve the problems of unreachable targets, local minima traps, and excessively long paths; Step 4: based on changes or adjustments in missions due to internal or external factors, and in response to real-time requirements, autonomous clustering and flight path planning of the UAV swarm is achieved.

[0007] The invention is further characterized by:

[0008] In step 1, the mission nature includes mission modes such as target reconnaissance, patrol, and fire attack. The permitted formation of the aircraft group depends on the external environment. For example, when performing missions in high mountains or canyons, a loose lateral formation cannot withstand strong wind interference and cannot be selected for mission execution.

[0009] In step 2, each cluster of the UAV swarm is either a sub-cluster with a single mission function or a mixed cluster composed of multiple functions. The specific functions of each cluster include target detection, fire support, mobile backup, jamming, and counter-jamming. Furthermore, the interactive clustering formation method proposed in this invention can cope with diverse mission scenarios, meet real-time requirements, and solve the problem of fixed clustering formations. At the same time, compared with existing clustering formation methods, this method can effectively resist anti-swarm attacks and has strong robustness.

[0010] In step 3, the proposed improved artificial potential field method not only avoids the occurrence of target unreachability and local minima traps, but also optimizes the obstacle avoidance path of the UAV swarm, thereby reducing path length and computational load.

[0011] In step 4, real-time updates and dynamic adjustments are required based on changes in the task and the drone's state. The drone's flight path is dynamically planned according to updates to the gravitational and repulsive fields, thereby enabling autonomous clustering and dynamic trajectory planning for the drone swarm.

[0012] According to the method of the present invention, heterogeneous UAV swarms can achieve autonomous clustering and dynamic trajectory planning based on the real-time requirements of the mission, thereby improving the efficiency of UAV swarms in intelligently completing multi-target tasks and having the advantages of high efficiency, stability and reliability. Attached image description:

[0013] Figure 1 This is a topology example diagram of the unmanned aerial vehicle (UAV) swarm in this invention;

[0014] Figure 2 This is a topological example diagram of the unmanned aerial vehicle (UAV) swarm clustering in this invention;

[0015] Figure 3 This is a flowchart of the overall process for unmanned aerial vehicle (UAV) swarm clustering and flight path planning in this invention.

[0016] Figure 4 This is a flowchart of the drone swarm trajectory planning process in this invention;

[0017] Figure 5 This is a flowchart of drone swarm trajectory planning according to another embodiment of the present invention. Detailed implementation method:

[0018] To more clearly illustrate the objectives, technical solutions, and advantages of the present invention, the present 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 of the invention and are not intended to limit the invention.

[0019] Consider a heterogeneous drone swarm system consisting of N drones and m clusters, i.e. And there are That is, the first cluster It contains c1 drones, and the i-th cluster Includes c i -c i-1 There are m drones in the m-th cluster. Includes Nc m-1 This section focuses on the research and development of UAV swarm clustering and trajectory planning, including the functional settings of each cluster based on the mission, feasible formations for different missions in the mission chain, the topology of the swarm formation, and the trajectory planning of the swarm.

[0020] The following provides a specific example to describe the method of the present invention. See also: Figure 1 In the topology graph, each node represents a drone, and its dynamic characteristics can be expressed as f. i (·), i = 1, ..., 5, and the edge weights in the topological graph are A. ij Let i, j = 1, ..., 5, where i and j represent the i-th and j-th drones, respectively. ij This represents the information transmitted by the j-th drone to the i-th drone.

[0021] The permissible cluster formation of the UAV swarm is determined based on the nature of the mission and the flight environment of the UAV swarm. The nature of the mission can be preset, such as target reconnaissance, patrol, and fire attack, which specifies the mission and feasible formation for each cluster. For example, if the flight environment of the swarm is in a mountainous or canyonous area, a loose lateral formation cannot withstand strong wind interference and should not be selected for the mission.

[0022] After clarifying the nature of the mission and the flight environment, the principle of balanced allocation is followed:

[0023]

[0024] That is, all UAVs within the same cluster have the same dynamic characteristics and the same external (external) information input. and (Indicates different clusters). See also Figure 2 middle (Representing a clustering method), the drone swarm is first divided into 2 clusters. The drones in each cluster have the same function and can complete a specific task. This division method is called minimal division because it has the fewest clusters.

[0025] When the task changes, such as when a task is added temporarily, in Based on this, there exists Six clustering methods, among which This means that each drone forms a separate cluster, each completing a specific task. This interactive clustering formation method can handle diverse mission scenarios, meet real-time requirements, and solve the problem of fixed clustering. Furthermore, when subgroups... When subjected to anti-cluster attacks, subgroups The coordination will not be affected (e.g.) (as shown), but when the subgroup When subjected to anti-cluster attacks, subgroups Unable to maintain coordination (e.g.) As shown), that is One-way dependence Compared to existing clustering methods, this method has stronger robustness.

[0026] Next, for the target task of each swarm, this invention proposes an improved artificial potential field method to complete the trajectory planning of the UAV swarm, thereby solving the problems of target unreachability and local minima traps in existing methods.

[0027] In this example, taking the i-th cluster as an example, the potential field function of the improved artificial potential field method is:

[0028]

[0029]

[0030] Where X is the current position of a single UAV in the i-th cluster, X goal Let X be the target location of the i-th cluster of machines. ob Let k be the coordinates of the obstacle in the i-th cluster of aircraft; att and k rep U represents the gravitational and repulsive potential coefficients, respectively; att (X) and U rep (X) represent the gravitational and repulsive potential field functions, respectively; ρ(X, X) goal )=||XX goal || represents the relative distance between the UAV and the target point, ρ(X, X) ob )=||XX ob || represents the relative distance between the drone and the obstacle; ρ0 represents the maximum effective range of the obstacle, beyond which there is no effect; t is the adjustment factor, which is a variable normal number.

[0031] By taking the negative gradient of the above potential field function, the gravitational, repulsive, and net forces acting on the UAV can be obtained as follows:

[0032]

[0033]

[0034] F(X)=F att (X)+F rep (X) (6)

[0035] in,

[0036]

[0037]

[0038] Repulsive component F rep1 The direction of (X) is from the obstacle towards the drone, and the repulsive force component F rep2The direction of (X) is from the drone towards the target position, in the same direction as gravity. As the drone performs its mission, it gradually approaches the target position. If t∈(0,1), ρ(X,X) goal If )→0, then the repulsive component F rep1 (X)→0, repulsive component F rep2 (X)→∞, under the gravitational force F att (X) and repulsive component F rep2 The combined effect of (X) solves the problem of target unreachability (even if there are obstacles around the target location, the drone can still fly away from the obstacles toward the target location and complete the target mission).

[0039] To avoid the local minimum trap (where a local minimum occurs when the gravitational and repulsive forces acting on the drone are equal in magnitude and opposite in direction), this invention introduces a control force to break the force equilibrium state caused by the drone falling into a local minimum:

[0040] F ct (X)=F′ rep (X)+μF att (X) (9)

[0041] Among them, F ct (X) represents the regulating force, F′ rep (X) represents the new repulsive force, the magnitude of which is the sum of the repulsive force components of each obstacle, and the direction of which is perpendicular to the resultant force of the repulsive force components of each obstacle; μ represents the control coefficient, which is defined as follows:

[0042]

[0043] Where, ρ j (X, X) ob Let ρj be the relative distance between the drone and the j-th obstacle, and ρ0 be the maximum effective range of the obstacle. It is easy to see that the value of μ ranges from 0 to μ to 1. When μ is large, the distance between the drone and the obstacle is still far, and the drone will try to escape the local minimum by moving closer to the obstacle. When μ approaches 0, the distance between the drone and the obstacle is close, and the control force is approximately F′. rep (X) The drone will escape the local minimum in the direction of the resultant force perpendicular to the repulsive force components of each obstacle, thus ensuring the safety of the trajectory planning.

[0044] Under the influence of the control force, the drone swarm can break out of the local minimum state, escape the local minimum trap, cancel the control force, and fly towards the target position under the influence of gravity and repulsion to complete the target mission assigned to the swarm.

[0045] Furthermore, in order to optimize the path length of the drone swarm during the trajectory planning process (when avoiding obstacles, drones inevitably plan obstacle avoidance routes that deviate from the original straight path, resulting in excessively long paths), this invention proposes the following threshold function:

[0046] S j =|ρ j (X, X) ob sinθ j | (11)

[0047] Where, ρ j (X, X) ob θ represents the relative distance between the drone and the j-th obstacle. j Let S represent the angle between the line connecting the drone to the j-th obstacle and the line connecting the drone to the target position. j If the distance S is greater than the safe distance R for the drone's flight, the drone will continue along its current route without colliding with the obstacle. The obstacle is then considered invalid, and the repulsive force generated by the obstacle is disregarded. j When the distance is less than the safe distance R, the repulsive force generated by the obstacle must be considered. The formula for the repulsive force function in this case is:

[0048]

[0049] In other words, when an obstacle does not pose a significant threat to the current path, its impact is ignored, and the path continues along the original route, introducing S. j This will make the path planning of the fleet smoother and the path shorter.

[0050] The above-mentioned trajectory planning for UAV swarms based on the improved artificial potential field method can not only avoid the occurrence of unreachable targets and local minima traps, but also optimize the obstacle avoidance path of the UAV swarm, thereby reducing the path length and computational load.

[0051] Based on the real-time requirements of the mission and the operational needs of the drones themselves, the drone swarm can be updated and dynamically adjusted in real time, thereby completing the autonomous clustering and flight path planning of the drone swarm. Figure 3 This is the overall flowchart of UAV swarm clustering and trajectory planning in this invention.

[0052] Step 1: Determine the mission type, mission nature, and permissible formation of the UAV swarm; set initial parameters for the heterogeneous UAV swarm, such as the number of swarm members and their performance. The types and meanings of mission types and mission natures can be pre-defined, for example, target reconnaissance or patrol. The mission is also decomposed to suit multi-UAV swarm collaboration. The mission also corresponds to the required number of UAVs and feasible formation methods.

[0053] Step 2: Based on the initial task planning, and according to the task requirements and the different functions of the UAV swarm, autonomous clustering of the swarm is achieved according to the principle of balanced partitioning. Optionally, each UAV cluster is used to execute one task, and multiple UAV clusters cooperate to complete the overall task through their respective tasks.

[0054] Step 3: Based on the clustering forms corresponding to different mission characteristics, a modified artificial potential field method is used for trajectory planning, effectively solving the problems of target unreachability, local minima traps, and excessively long paths. Through the modified artificial potential field function, the gravitational and repulsive functions of each UAV in each cluster are obtained, and the resultant force acting on the UAV is calculated as the driving force. The UAV is driven according to the calculated driving force, enabling it to achieve the trajectory used to perform the mission.

[0055] Step 4: Based on changes or adjustments in the mission due to internal or external factors, and addressing real-time requirements, achieve autonomous clustering and trajectory planning for the UAV swarm. As the mission executes, various factors change, such as changes in UAV positions, the identification of new obstacles, changes in the position of the mission target, and / or the disappearance of previously identified obstacles. The resultant force is updated based on the gravitational and repulsive functions of each UAV to update the driving force on the UAVs, achieving dynamic trajectory planning. Optionally, during mission execution, the mission may change, such as the emergence of new or additional targets. In response, the UAV swarm clustering method is changed, autonomously resetting the mission for each cluster and updating the driving force.

[0056] According to embodiments of the present invention, the calculation and implementation can be performed by the information processing equipment onboard the UAV, or by a server located outside the UAV and transmitted to the UAV swarm.

[0057] Figure 4 This is a flowchart of the drone swarm trajectory planning process in this invention.

[0058] During the flight of the drone swarm, dynamic implementation is based on Figure 4 The example illustrates flight path planning in an embodiment of the present invention. For instance, during flight, the flight path is planned based on specified time intervals or changes in the drone's state. Figure 4 The process of updating the driving force of drones.

[0059] Calculating the driving force requires the use of gravitational and repulsive functions. The gravitational and repulsive functions are obtained from the improved artificial potential field function (see formulas (2)-(6) above). Understandably, the acquisition of gravitational and repulsive functions does not need to occur in each trajectory planning, but rather one or more gravitational and repulsive functions suitable for different scenarios can be obtained offline.

[0060] During mission execution (e.g., in flight), the system acquires the target position, identified obstacle positions, and the position of each drone within the cluster for each drone cluster. Based on this positional information, it applies gravitational and repulsive functions to calculate the gravitational and repulsive forces acting on each drone at its current position. The resultant force of these calculated gravitational and repulsive forces is then used as the driving force to propel the drones into flight.

[0061] By repeatedly acquiring multiple location information and updating the driving force, the flight path of the aircraft group is continuously corrected until the set mission is completed.

[0062] Optionally, new obstacles may be discovered during flight, or previously identified obstacles may be overcome, thus updating the obstacle positions used in calculating gravity / repulsion as obstacles are identified. Overcome obstacles may be disregarded in gravity / repulsion calculations.

[0063] Figure 5 This is a flowchart of drone swarm trajectory planning according to another embodiment of the present invention.

[0064] same Figure 4 Compared to the previous embodiment, Figure 5 The demonstrated processing flow also responds to task updates. For example, if a new target is discovered while an already clustered drone swarm is performing a task, the existing drone swarm needs to be re-clustered to create new drone clusters to perform new tasks targeting the new target, while ensuring that a corresponding drone cluster continues to complete the original target task.

[0065] In response to task updates, such as detecting changes in the number of tasks (increase, decrease, or alteration), the UAV swarm is re-clustered according to the changed tasks. Following the principle of balanced partitioning, the re-clustering still satisfies formula (1). Besides creating new UAV clusters to perform new tasks, existing UAV clusters performing tasks may also be adjusted (e.g., the number of UAVs decreases, and some UAVs join new clusters). Each re-clustered UAV cluster has its own corresponding task to perform.

[0066] The trajectory planning process implemented by each drone cluster to perform its corresponding task is the same as... Figure 4 The process shown is consistent.

[0067] The method according to embodiments of the present invention can greatly improve the efficiency of unmanned aerial vehicle swarms in intelligently completing multi-target tasks, and has the advantages of high efficiency, stability and reliability.

[0068] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for interactive clustering and trajectory planning of heterogeneous unmanned aerial vehicle (UAV) swarms, comprising the following steps: Based on the task to be processed and the different functions and dynamic characteristics of each UAV in the heterogeneous UAV swarm, the heterogeneous UAV swarm is divided into multiple clusters. Each cluster corresponds to one of the tasks to be processed and is used to execute one task. The multiple UAV clusters cooperate to complete the overall task through their respective tasks. The UAVs in each cluster have the same dynamic characteristics and the same functions. For each of the divided clusters, obtain the target location of its corresponding task, the location of the identified obstacles, and the current location of each UAV within that cluster; Based on the target position of each cluster's task, the position of obstacles, and the current position of each UAV within the cluster, the gravitational and repulsive forces experienced by each UAV within the cluster at its current position are calculated using the gravitational and repulsive functions obtained based on the improved artificial potential field function. The improved artificial potential field function generates gravitational forces for the UAVs within the cluster based on the target position of the task corresponding to each cluster and the current position of the UAVs within the cluster. The combined force of the gravitational and repulsive forces acting on each UAV at its current position drives each UAV, causing each cluster to fly toward the target position and complete the target task assigned to that cluster. The improved artificial gravitational potential field function is: Where X is the current position of a single UAV in the i-th cluster of UAVs, X goal Let X be the target location of the i-th cluster of machines. ob Let k be the location of the obstacle encountered by the i-th cluster of aircraft; att and k rep U represents the gravitational potential coefficient and the repulsive potential coefficient, respectively; att (X) and U rep (X) represent the gravitational potential field function and the repulsive potential field function, respectively; ρ(X,X) goal )=||XX goal || represents the relative distance between the single UAV and the target position, ρ(X,X) ob )=||XX ob || represents the relative distance between the single UAV and the obstacle; ρ0 represents the maximum effective range of the obstacle; t is an adjustment factor, which is a variable normal number.

2. The method according to claim 1, wherein The sum of inputs received by each drone in each of the divided clusters from drones in other clusters is the same.

3. The method according to claim 2, wherein The available formations are determined based on the nature of the task to be performed and the flight environment of the drone swarm. Configure the cluster settings and grouping for the tasks to be processed.

4. The method according to claim 1, wherein The gravitational function is: The repulsion function is: The function of the resultant force is F(X) = F att (X)+F rep (X) in, Repulsive component F rep1 The direction of (X) points from the location of the obstacle to the single drone, and the repulsive force component F rep2 The direction of (X) is from the single UAV to the target position, in the same direction as gravity.

5. The method according to claim 4, wherein The function of the resultant force is F(X) = F att (X)+F rep (X)+F ct (X) F ct (X)=F′ rep (X)+μF att (X) Where F ct (X) represents the control force, which is used to drive the UAV out of the force equilibrium state caused by being trapped in a local minimum, F′ rep (X) represents the new repulsive force, the magnitude of which is the sum of the repulsive force components of all obstacles, and the direction of which is perpendicular to the resultant force of the repulsive forces of all obstacles; μ represents the control coefficient, which is defined as follows: Where ρ j (X,X ob ) represents the relative distance between the single UAV and the j-th obstacle, and ρ0 represents the maximum effective range of the obstacle.

6. The method of claim 4, wherein The repulsion function in, R represents the safe distance for the drone during flight. S j =|ρ j (X,X ob )sinθ j | θ j Let S represent the angle between the line connecting the single UAV to the j-th obstacle and the line connecting the single UAV to the target position, where S is introduced. j This will make the path planning of the drone swarm smoother and the path shorter, when S j When the obstacle posed no threat to the current path, the repulsive force function is zero, ignoring the repulsive force effect of the obstacle, allowing the single drone to continue along the original path.

7. The method according to any one of claims 1-6, further comprising: In response to an update of the task to be processed, or if the number of tasks to be processed changes, the drone swarm is re-clustered, with each cluster corresponding to one of the tasks to be processed.

8. The method according to claim 7, further comprising: During the flight of one or more clusters of drones, update the target position of their corresponding mission, the position of identified obstacles, and / or the current position of each drone within the cluster; Based on the updated target position, obstacle position, and UAV current position, the gravitational and repulsive forces acting on each UAV at its current position are calculated using the gravitational and repulsive functions. Each drone is driven by the combined force of the gravitational and repulsive forces acting on it at its current position.

9. The method according to claim 8, further comprising: Based on the updated target position, obstacle position, and UAV current position, select the gravitational and repulsive functions to calculate the gravitational and repulsive forces acting on each UAV at its current position.

10. An information processing device, comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method according to any one of claims 1-9.

Images