A method and system for high-dynamic unmanned aerial vehicle swarm autonomous path planning

Abstract

The application discloses a kind of high dynamic unmanned aerial vehicle group autonomous path planning method, comprising the following steps;Step one: initialize the position of unmanned aerial vehicle, speed and termination condition and introduce constraint condition;Get unmanned aerial vehicle group group N;Step two: evaluate unmanned aerial vehicle cluster situation, reconstruct communication topology structure;Step three: introduce evaluation parameter system to reconstructed communication topology structure, according to the current unmanned aerial vehicle velocity vector, relative distance, the information of unmanned aerial vehicle quantity, calculate evaluation parameter;Step four: with the parameter in the aforementioned evaluation parameter system is optimized, obtains temporary fitness function value, calculates fitness function.Step five: optimization unmanned aerial vehicle flight parameter;If the fitness function value is updated, the various parameters to be optimized are also updated to corresponding parameter value;Step six: judge termination condition;The application has the characteristics of reliability, stability, real-time.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, and specifically relates to a method, system, device and storage medium for autonomous trajectory planning of a high-dynamic unmanned aerial vehicle (UAV) swarm. Background Technology

[0002] Autonomous cooperative control of unmanned aerial vehicles (UAVs) refers to a control method in which a group of UAVs, based on inter-UAV communication and swarm intelligence, cooperate and divide tasks to complete a common task. It is a multi-agent system that executes tasks based on local rules, offering advantages such as high efficiency, high flexibility, and high reliability. Due to these advantages, UAV swarms are widely used in scenarios such as remote sensing, communication, agriculture, and search and rescue. In these scenarios, UAV swarms need to form specific topologies (or configurations) based on task requirements and promptly diagnose and handle faults to improve system performance under specific functional demands.

[0003] As mission requirements become increasingly sophisticated and updated, and mission demands continue to rise, the environments in which UAVs operate become increasingly complex, placing higher demands on the autonomous and collaborative control of UAVs: (1) The platform space configuration for the missions performed by UAVs is complex, and environmental constraints are becoming increasingly stringent; (2) The flight attitude of UAVs and the topology of the cluster are heavily coupled, and an unreasonable cluster configuration may lead to a decrease in detection accuracy; (3) As the scale of UAVs increases, inter-UAV communication becomes more frequent, placing a greater burden on onboard computing resources; (4) Higher requirements are placed on anti-interference and strike capabilities. When a UAV in the cluster fails due to interference, it is necessary to quickly diagnose the fault and switch configurations to reorganize. In order to improve the adaptability of UAV clusters to the environment and ensure the stability of mission execution, further research is needed on UAV cluster configuration optimization based on control and other theories, as well as strategies for reorganizing the cluster after UAV failure.

[0004] With the disruptive transformation brought about by artificial intelligence technology, autonomous trajectory planning technology for UAV swarms has become a key technology for improving the autonomy of UAVs. Autonomous trajectory planning for UAVs aims to generate an optimal mission route connecting the starting point and the destination, while meeting performance constraints and ensuring flight safety. Essentially, it is a type of optimization problem under multiple constraints. UAV trajectory planning can be divided into pre-planning and real-time trajectory planning according to mission requirements. Pre-planning is a global optimization based on prior environmental information, thus obtaining a globally optimal solution within a reasonable timeframe is more practically significant. Real-time trajectory planning addresses local route optimization in response to sudden threats, requiring minimizing planning time to meet real-time requirements. Multi-UAV collaborative trajectory planning is more complex than single-UAV trajectory planning, requiring consideration of not only the constraints of the UAVs themselves but also the spatiotemporal constraints of the mission to avoid trajectory overlaps and mission conflicts. Research on multi-UAV collaborative trajectory planning in large-scale complex scenarios and dynamic uncertain environments is of greater practical significance.

[0005] Traditional optimization algorithms, such as graph search, spatial sampling, and potential field methods, are relatively simple to implement. However, they suffer from low search efficiency in large-scale, high-dimensional spaces, struggle to solve multi-constraint trajectory planning problems, or are prone to local oscillations and local minima in large-scale, high-dimensional spaces, rendering flight paths unusable. While traditional optimization algorithms can achieve good trajectory planning results in simple environments, they are inadequate for efficiently solving autonomous trajectory planning problems for UAVs in complex, dynamic, and highly uncertain battlefield environments.

[0006] Traditional optimization algorithms have significant advantages in small-scale environments, but their computational complexity increases exponentially with the size of the space, easily leading to a "combinatorial explosion." Intelligent algorithms, due to their high parallelism, strong robustness, and self-organizing and self-learning capabilities, can be applied to autonomous trajectory planning problems for unmanned aerial vehicles (UAVs) in complex environments. By constructing a heuristic objective (reward) function using intelligent algorithms and employing parallel computing and self-organizing / self-learning methods, they demonstrate good applicability in complex environments. The complexity of the environment is also reflected in the complexity of the constraints, requiring a comprehensive consideration of the UAV's own performance constraints and dynamically changing threats and obstacles. Therefore, the autonomous trajectory planning problem is a type of multi-extremum optimization problem. Summary of the Invention

[0007] To overcome the shortcomings of the existing technology, the present invention aims to provide a method and system for autonomous trajectory planning of highly dynamic unmanned aerial vehicle (UAV) swarms. This method features reliability, stability, and real-time performance. It solves the problems related to autonomous trajectory planning of highly dynamic UAV swarms in obstacle-prone environments.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] A method for autonomous trajectory planning of a highly dynamic unmanned aerial vehicle (UAV) swarm includes the following steps;

[0010] Step 1: Initialize the drone's position, speed, and termination conditions, and introduce constraints; obtain the drone swarm;

[0011] Step 2: Every Δt time interval, assess the drone cluster status, remove missing drones, and reconstruct the communication topology;

[0012] Step 3: Introduce an evaluation parameter system into the reconstructed communication topology. Calculate the evaluation parameters based on the current UAV velocity vector, relative distance, and number of UAVs. These evaluation parameters are used to calculate the objective function in the optimization algorithm.

[0013] Step 4: Optimize the evaluation parameters in the evaluation parameter system based on the algorithm proposed in the specific steps to obtain temporary fitness function values ​​and calculate the fitness function; to improve the stability of the UAV swarm operation.

[0014] Step 5: Based on the updated fitness function value, optimize the UAV flight parameters in the fitness function accordingly;

[0015] Step 6: Determine the termination condition; if the current time t > t max If the loop ends, then update the drone's speed and position at time t+Δt and return to step two.

[0016] The specific steps of step one are as follows:

[0017] N drones are randomly generated, and the initial coordinates of the i-th drone are x. i The velocity vector of the i-th drone is v i The relative distance between the i-th drone and the j-th drone is r. ij The distance of the i-th drone from the origin is r. i Let i = 1, 2, ..., N, j = 1, 2, ..., N, and set the system runtime t. max The drone swarm is constructed as a connected graph, with individual drones as nodes. If the relative distance between drones is less than a given communication distance, an edge is established between them to achieve a distributed architecture. An update control variable time interval Δt = 0.5s is given.

[0018] Furthermore, set the drone's velocity vector. ( It is the calculated velocity vector, v i (This is the actual velocity vector)

[0019]

[0020] Set constraints: the number of missing drones N disc This refers to the number of disconnected drones, which is less than one-fifth of the total number of drones.

[0021] N disc <N / 5.

[0022] Step two specifically involves:

[0023] Step one involves assessing the drone swarm situation, which includes missing drones.

[0024] The number of missing drones is divided into those that were damaged and those that flew too far.

[0025] If no drones are damaged or fly too far, the communication topology between drones remains unchanged.

[0026] If a drone encounters an emergency during flight, such as being attacked or colliding with an obstacle, resulting in damage, it needs to exchange information with each other, discard the damaged drones, optimize the drone swarm structure, detect the relative distance between the drones, and then reconstruct an executable communication topology based on the method in the first step.

[0027] Step three specifically involves:

[0028] To maintain the stability of the UAV swarm, velocity correlation is introduced into the evaluation parameter system to evaluate the consistency of the velocity vectors of each UAV, as follows:

[0029]

[0030] The i-th drone v i N represents the velocity vector of the drone. i J represents the number of neighbors of drone i. i Let v represent the set of neighbor indices of drone i, T represent the upper limit of system runtime, and v j This represents the velocity vector of drone j. The closer the above parameters are to 1, the better the consistency.

[0031] Secondly, based on the flight environment, collision risks between UAVs and collision risks between UAVs and obstacles are introduced into the evaluation parameter system:

[0032]

[0033]

[0034] Where Θ(·) is the Heaviside function, N is the number of drones, and r coll It is the safe distance for collisions. This is the distance sign (positive or negative) from drone i to virtual drone s (when an obstacle is detected, a virtual drone is created at the obstacle location, thus obstacle avoidance can be achieved by simulating maintaining distance between drones). When drone i is inside an obstacle (due to autonomous trajectory planning simulation, sometimes drones may autonomously determine they are inside an obstacle; to avoid this, although it is rare, optimization is still needed, hence the introduction of a negative sign; when this occurs, this indicator deteriorates, causing the system to not adopt this approach, thus achieving the goal), r is negative. is (t) is the distance between UAV i and virtual UAV s at time t; the closer this parameter is to 0, the lower the risk of collision, the more flexible the decision-making when encountering obstacles, and the higher the fault tolerance between UAVs.

[0035] Introduce the average speed index of UAVs into the evaluation parameter system to evaluate whether the actual flight speed of UAVs meets the set swarm speed:

[0036]

[0037] The parameters are defined as above; the closer the parameters are to the common velocity v, the better. flock (This speed is set manually, i.e., the ideal cluster speed), indicating that the faster it can converge to the set cluster speed, the more this step will calculate the above evaluation parameters and substitute the results into the next step.

[0038] Step four specifically involves:

[0039] Calculate the fitness function:

[0040] F = F speed ·F coll ·F disc ·F cluster ·F obstacle ·F corr

[0041]

[0042] Where F1(φ,φ0,d)=1-S(φ,φ0,d);

[0043]

[0044]

[0045] F speed It's an indicator of the speed of drone swarms, F coll It is an indicator of the risk of collisions between drones, F disc F cluster It is a metric about connectivity, N disc It is the number of missing drones, F obstacle It's about collision risk indicators between drones and obstacles, F corr It is an indicator related to the motion correlation of drones;

[0046] Let the cluster speed tolerance be v. tol Tolerance for collision risk between drones (a) tol The tolerance for the risk of collision between drones and obstacles (r) tol .

[0047] The above model transforms the distributed control problem into an optimization problem, where the fitness function, which is also the objective function, needs to be maximized. This step calculates a temporary fitness function value. In the next step, through the optimization algorithm, if a better result is found, the temporary fitness function value will be updated.

[0048] Step five specifically involves:

[0049] Based on the current flight environment, and considering the current flight spacing safety constraints, repulsive field constraints, swarm flight consistency constraints, and spherical / cylindrical obstacle constraints (these constraints are also artificially set safety constraints to ensure that collisions between drones and between drones and obstacles are minimized), the optimization parameters are updated to maximize the fitness function, i.e., max F = max(F speed ·F coll ·F disc ·F cluster ·F obstacle ·F corr Update the expected velocity vector for the next moment:

[0050]

[0051]

[0052] in The expected speed of drone i, v flock It is the self-driven speed of each drone as an independent individual, which enables the drone to have a certain cruising ability even when it loses contact with the group. The direction and magnitude of this speed are determined by the user.

[0053] It is the speed generated by the drone i under the combined repulsive force of the surrounding drones, used to avoid collisions between drones;

[0054] It is the repulsive velocity between drones i and j, pointing towards i, and is defined as follows:

[0055]

[0056] p rep These are the linear coefficients to be optimized. This is the safe distance to be optimized; the closer the distance between drone i and drone j, the greater the value of this speed will be, meaning the stronger the repulsive force generated.

[0057] It is the velocity generated by drone i under the speed synchronization drag, which makes the magnitude and direction of the velocities of each drone as consistent as possible;

[0058]

[0059]

[0060] It is v i and v j The speed synchronization term between them, where C frict These are the linear coefficients to be optimized.

[0061] v frict This is the maximum permissible speed difference. D(·) is a decay function, defined to allow UAVs to decelerate rapidly while maintaining a large distance and high speed, thus controlling the speed synchronization time within a small range; a frict and p frict These are all parameters that need optimization. The safety distance to be optimized is D(·), defined as follows:

[0062]

[0063] This is the speed generated by the repulsive force exerted on the drone i by the obstacle, so that the drone can avoid walls and obstacles during flight. The idea behind establishing this speed term is to generate "virtual drones" at the locations of walls and obstacles, so that this term can be defined with reference to the speed synchronization term mentioned above.

[0064] The model is as follows:

[0065] in, r is It is the distance from drone i to virtual drone s. The safe distance to be optimized is a. shill and p shill These are the parameters to be optimized. This step provides the rules for updating the drone's speed in the next moment.

[0066] All the parameters to be optimized mentioned above will be optimized in step five and replaced with parameters that maximize the fitness function.

[0067] Another object of the present invention is to provide a method for autonomous trajectory planning based on highly dynamic unmanned aerial vehicle swarms, comprising:

[0068] Step 1: Initialize the drone. Randomly initialize the parameters to be optimized according to certain rules. First, set the value of the objective function F (the same as defined in the previous module) to 0, and the maximum number of iterations itermax.

[0069] Step 2: Perform the first iteration. At each moment during the simulation test time, work in conjunction with the distributed algorithm module to update the speed and position of each UAV. The following (1) to (4) are all calculated and delivered by the distributed algorithm module:

[0070] (1) Calculation

[0071] (2) Calculation

[0072] (3) Calculation

[0073] (4) Calculation

[0074] Then according to

[0075]

[0076] Update the speed of each drone, and then update the drone's location information;

[0077] Step 3: Calculate φ according to the evaluation parameter system introduced in the distribution optimization module. corr ,φ wall ,φ obstacle and φ vel Then, the value of the objective function F is calculated. If the current calculated F value is greater than the F before the iteration, the previous F is changed to the F value now sought, and the corresponding parameters to be optimized are updated.

[0078] Step 4: The current iteration ends and the next iteration begins. The parameters to be optimized are reset according to certain rules and substituted into the solution until the maximum number of iterations is reached.

[0079] A system for implementing the method of autonomous trajectory planning based on highly dynamic unmanned aerial vehicle (UAV) swarms, comprising:

[0080] The flight parameter control module is used to control the size of the UAV swarm during simulation, taking integer values ​​between 1 and 50.

[0081] The simulation parameter setting module is used to control the upper limit of the test time during simulation;

[0082] The resilience verification module is used to set the percentage of drone swarms that are attacked, which can be set to a number between 0-40% by sliding a slider.

[0083] The algorithm selection module is used to select a distributed optimization algorithm.

[0084] The flight direction control module is used to control the flight direction of the drone swarm during simulation. It can rotate freely by 270° and ensures real-time adjustment during operation.

[0085] The two methods described above have different focuses. The first method focuses on calculating the objective function to optimize the parameters, while the second method focuses on calculating the speed and position of the UAV at the next moment.

[0086] The beneficial effects of this invention are:

[0087] This invention provides autonomous setting of the UAV environment and algorithm environment. To simulate the spatial trajectory control of the UAV, the UAV is simulated as a point mass. That is, the software performs simulation and optimization based on the UAV being approximately viewed as a point object.

[0088] This invention integrates two algorithms for the autonomous function of a super-maneuverable UAV: ​​a distributed control algorithm and a trajectory planning algorithm. The user interface provides corresponding algorithm selection buttons. Both algorithms are feasible, and there is no need to consider which is superior or inferior or how to choose. The only difference is the final visualization effect.

[0089] For the high-maneuverability control of UAV swarms, the software platform provides two algorithm options. Algorithm 1 is a distributed control optimization algorithm that solves the spatial flight control of UAV swarms, including flight direction, flight speed, and flight spacing.

[0090] Algorithm 2 is a path-solving algorithm based on distributed control optimization. Algorithm 2 provides calculation of the UAV maneuver flight trajectory within the spatial obstacle range, and real-time flight control includes flight direction, flight speed, flight distance, etc. Attached Figure Description

[0091] Figure 1 This is a schematic diagram of the interface of the autonomous function integrated design platform system for super-maneuverable unmanned aerial vehicles.

[0092] Figure 2 This is a schematic diagram of the simulation test space of the present invention.

[0093] Figure 3 This is a flowchart of the real-time optimization computing framework of the present invention.

[0094] Figure 4(a) is a sample diagram of the dynamic flight trajectory provided in an embodiment of the present invention.

[0095] Figure 4(b) is a summary example of flight obstacle distances provided in an embodiment of the present invention.

[0096] Figure 5 This is a schematic diagram of the distributed optimization algorithm process of this invention.

[0097] Figure 6This is a security performance graph-time curve under attack provided in an embodiment of the present invention.

[0098] Figure 7 This is a schematic diagram of the acceleration of an attacked drone and a time curve provided in an embodiment of the present invention. Detailed Implementation

[0099] The present invention will now be described in further detail with reference to the accompanying drawings.

[0100] Faced with harsh flight environments, and the need for multi-tasking, multi-objective UAV swarm systems lacking a leader and centralized computing and control system, relying solely on localized information exchange between drones, a distributed intelligent optimization algorithm is required to control the flight speed and angle of the UAVs in real-time. This ensures smoother operation of the UAV swarm and meets requirements for reliability, real-time performance, resilience, stability, and robustness. Furthermore, considering the increased computational scale and optimization complexity as the UAV swarm grows, this project, building upon previous research, improves the distributed optimization algorithm for application in practical autonomous UAV flight planning problems.

[0101] like Figures 1-7 As shown below, the process of distributed optimization will be introduced:

[0102] Step 1: Initialize the UAV's position, velocity, termination conditions, and introduce constraints. Randomly generate N initial coordinate positions x for the UAV. i i = 1, 2, ..., N, velocity vector v i , i = 1, 2, ... N, calculate the relative distance r of UAV i. ij i≠j and the distance r from the origin i Set runtime t max The drone swarm is constructed as a connected graph, with individual drones acting as nodes. If the distance between drones is less than a given communication distance, an edge can be established between them to achieve a distributed architecture. An update control variable time interval is given, such as Δt = 0.5s.

[0103] Because there is no leader and the drones rely solely on exchanging localized information to achieve their objectives, they are autonomous decision-makers and controllers. Therefore, the drone's velocity vector cannot be too fast or too slow.

[0104]

[0105] The speeds mentioned above are scalars because only magnitude is considered. Secondly, the cluster may lose some drones due to attacks; here, we define the number of lost drones as N. discThis refers to the number of disconnected drones, which we assume to be less than one-fifth of the total number of drones.

[0106] N disc <N / 5

[0107] Step 2: Assess the drone swarm situation and reconstruct the communication topology. If no drones are damaged or have flown too far, the communication topology between drones remains unchanged. If a drone encounters an emergency during flight, such as being attacked or colliding with an obstacle, resulting in damage, the drones need to exchange information, discard the damaged drones, optimize the drone swarm structure, and reconstruct an executable communication topology.

[0108] Step 3: Introduce evaluation parameters. Based on current speed control parameters, flight spacing, number of drones, and other information, calculate some evaluation parameters. To maintain the stability of the drone swarm, introduce speed correlation to evaluate the consistency of the speed vectors of each drone, as follows:

[0109]

[0110] The closer the above parameters are to 1, the better the consistency.

[0111] Secondly, based on the flight environment, the risks of collisions between drones and between drones and obstacles are introduced:

[0112]

[0113]

[0114] Where Θ(·) is the Heaviside function, N is the number of drones, and r coll It is the safe distance for collisions. This is the distance (positive or negative) from drone i to virtual drone s, with the sign indicating whether drone i is inside an obstacle. The closer this parameter is to 0, the lower the risk of collision, and the more flexible the drones are in their decision-making regarding obstacles, as well as the higher the fault tolerance between drones.

[0115] Finally, the average speed metric for drones was introduced to evaluate whether the actual flight speed of the drones met the set swarm speed:

[0116]

[0117] The closer the above parameters are to the common velocity v flock This indicates that the faster the cluster converges to the set cluster speed, the better.

[0118] Step 4: Calculate the fitness function. Calculate the following function:

[0119] F = F speed ·F coll ·F disc ·F cluster ·F obstacle ·F corr

[0120]

[0121] Where F1(φ,φ0,d)=1-S(φ,φ0,d),

[0122] F speed It's an indicator of the speed of drone swarms, F coll It is an indicator of the risk of collisions between drones, F disc F cluster It is a metric about connectivity, N disc It is the number of missing drones, F obstacle It's about collision risk indicators between drones and obstacles, F corr Regarding the indicators related to the motion of drones, the previously mentioned evaluation indicators also need to have a tolerance level set for each, so that some unsatisfactory results can be discarded. Let v tol ,a tol ,r tol This corresponds to the tolerance for cluster speed, collision risk between drones, and collision risk between drones and obstacles. The above model transforms the distributed control problem into an optimization problem, where the fitness function, or objective function, needs to be maximized.

[0123] Step 5: Optimize UAV flight parameters. Based on the current flight environment, and considering the current flight spacing safety constraints, repulsive field constraints, swarm flight consistency constraints, and spherical / cylindrical obstacle constraints, update and optimize the parameters to maximize the fitness function, i.e., max F = max(F speed ·F coll ·F disc ·F cluster ·F obstacle ·F corr Update the expected velocity vector for the next moment:

[0124]

[0125]

[0126] in The expected speed of drone i, v flockIt is the self-driven speed of each drone as an independent individual, which enables the drone to have a certain cruising ability even when it loses contact with the group. The direction and magnitude of this speed are determined by the user.

[0127] It is the speed generated by the combined repulsive force of the surrounding drones on drone i, used to avoid collisions between drones.

[0128] It is the repulsive velocity between drones i and j, pointing towards i, and is defined as follows:

[0129]

[0130] p rep These are the linear coefficients to be optimized. This is the safe distance to be optimized. It can be seen that the closer the distance between drone i and drone j, the greater the speed value, meaning the stronger the repulsive force generated.

[0131] It is the velocity generated by drone i under the speed synchronization drag, which makes the magnitude and direction of the velocities of each drone as consistent as possible.

[0132]

[0133]

[0134] It is v i and v j The speed synchronization term between them, where C frict These are the linear coefficients to be optimized.

[0135] v frict This is the maximum permissible speed difference. D(·) is a decay function, defined to allow UAVs to rapidly decelerate while maintaining a large distance and high speed, thus controlling the speed synchronization time within a small range. Here, a frict and p frict These are all parameters that need optimization. This is the safety distance to be optimized. D(·) is defined as follows:

[0136]

[0137] This is the velocity of drone i caused by the repulsive force exerted on it by obstacles, enabling the drone to avoid walls and obstacles during flight. The idea behind establishing this velocity term is to generate "virtual drones" at the locations of walls and obstacles, thus defining this term by referring to the velocity synchronization term mentioned above. The model is as follows:

[0138]

[0139] in, r is It is the distance from drone i to virtual drone s. The safe distance to be optimized is a. shill and p shill These are parameters that need to be optimized.

[0140] Step 6: Determine the termination condition. If the current time t > t0 max If the loop ends, then the loop ends; otherwise, update the drone's velocity and position at time t+Δt and return to step two.

[0141] To address the integrated mission design requirements of highly dynamic UAV swarms, this invention develops algorithms for irregular networks formed by swarms under fault-tolerant conditions, decentralized cooperative navigation, and planning. This invention designs an autonomous trajectory planning technology based on intelligent algorithms. While meeting performance constraints and ensuring flight safety, it considers the constraints of the UAV itself and the spatiotemporal constraints of the mission, avoiding trajectory intersections and mission conflicts, and utilizes intelligent algorithms to quickly generate an optimal mission route connecting the start and end points.

[0142] The technical effects of the present invention will be described in detail below with reference to simulation experiments.

[0143] 1. Experimental setup: The platform of this invention is developed based on Matlab 2018b and has a GUI design. The computer CPU is a 12th generation Intel Core i7-12650H, the memory is 16GB, and the operating system is 64-bit Windows 11.

[0144] 2. Simulation Parameters: Flight control parameters for the UAV swarm, including: formation flight direction, formation size settings, formation safety distance settings, formation flight speed, etc., used for UAV swarm calculation and control simulation demonstration. Basic operating parameters are as follows:

[0145]

[0146] 3. Simulation content and results:

[0147] Performance metrics: Key performance metrics include reliability and survivability. Reliability requires that, under non-human interference conditions, the drone swarm autonomously traverses obstacle spaces within fixed flight parameter constraints. Simultaneously, the software provides real-time control of highly maneuverable flight, including flight direction, speed, and spacing, enabling real-time intervention. Under autonomous drone flight conditions, the drones are controllable and commandable. Survivability assesses the drone system's performance against different types of threats and attacks. This testing can cover multiple aspects, including physical attacks, electromagnetic interference, and communication interference.

[0148] Simulation 1: Reliability of the optimization algorithm; First, the effectiveness of the optimization algorithm is verified. To demonstrate the optimization process, a drone swarm is simulated in both obstacle-free and obstacle-filled environments. The drone swarm size is set to 20, 30, and 40 drones, respectively. In the obstacle-filled environment, the flight direction is set to due north and northeast. The distance between all experimental drones is set to 10 meters, the speed is set to 6, and the test time is 100 seconds. It is assumed that the drones will not be attacked by external forces and that no drones will lose contact.

[0149] Simulation 2: Resilience of Distributed Optimization Algorithm; For comparison, the flight environment was divided into an unobstructed environment and an obstructed environment. The simulation parameters were set as follows: swarm size of 20 drones, spacing between drones of 10 meters, speed of 6, and test time of 100 seconds for both environments. In both environments, it was assumed that after the drone swarm had been running for a period of time, 30% of the drones would be randomly attacked (causing these drones to lose contact and become undetectable by other drones). By observing the behavior of the drone swarm after the attack, its resilience and adaptability could be evaluated. The attack time was set at 30 seconds.

[0150] When the drone swarm is attacked after 30 seconds, several drones disappear from the image, simulating that these drones have been damaged and no longer appear in the swarm. The remaining drones quickly rebuild the communication network, immediately regroup to strengthen their connection, and continue to complete their flight missions. This simulates the actual situation after a drone is attacked, demonstrating the resilience of the distributed optimization algorithm.

[0151] Before the attack, the drone swarm operated normally towards the target, maintaining good coordination and trajectory planning. At the 30-second mark, some drones were attacked. The attacked drones disappeared from the swarm's view, lost connection with other drones, and could no longer participate in trajectory planning and mission execution. The remaining normal drones needed to adapt and regroup to maintain swarm operation and complete the mission. Comparing this to the unattacked experiment, we can see that after the drone swarm was attacked at 30 seconds, the drones no longer evenly divided into two columns. Instead, more drones flew to the left, and fewer flew to the right. After bypassing the first obstacle, the drones on the right quickly regrouped with the ones on the left to establish a communication network with the majority of drones, adjust their control parameters, and complete the mission objective. This demonstrates the intelligence and effectiveness of the distributed optimization algorithm, as well as the autonomous decision-making and control capabilities of the distributed drones. Although the time it took for the drone swarm to complete the mission increased compared to the unattacked experiment, this also reflects the algorithm's stability. To ensure mission completion, internal adjustments were made first to stabilize the system before proceeding with the mission, demonstrating the system's intelligence.

[0152] like Figure 1 The diagram shown is a screenshot of the system's functional interface. Figure 2 The image shown is a screenshot of the system in operation. Figure 3 The diagram shows a real-time optimization calculation and solution framework; Figure 4 shows the flight path simulation diagram and the real-time parameter numerical diagram, respectively; Figure 5 The diagram shown is a flowchart of the algorithm. Figure 6 The graph shows the changes of various evaluation indicators over time during runtime; in the graph, there are instances where curves are equal in value and instances where curves are overlapped; for example... Figure 7 The attached diagram shows the acceleration of an attacked drone over time. The diagram shows the drone's acceleration fluctuating up and down due to the frequent changes in acceleration.

[0153] As can be seen from the attached diagram, the system has complete functional modules and many debuggable functions; the system's runtime visualization is concise and clear, and the effect is good. It can still maintain stable operation when attacked or encountering obstacles; the flowchart shows that the entire system's operating logic is clear; and the changes of various operating parameters and indicators over time are shown in the graphs, which can facilitate parameter adjustments according to different needs during subsequent testing.

[0154] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented, in whole or in part, as a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0155] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for autonomous trajectory planning of a highly dynamic unmanned aerial vehicle (UAV) swarm, characterized in that, Includes the following steps; Step 1: Initialize the drone's position, speed, and termination conditions, and introduce constraints; obtain the drone swarm; Step 2: Assess the drone swarm situation, remove missing drones, and reconstruct the communication topology; Step 3: Introduce an evaluation parameter system into the reconstructed communication topology. Calculate the evaluation parameters based on the current UAV velocity vector, relative distance, and number of UAVs. These evaluation parameters are used to calculate the objective function in the optimization algorithm. Step 4: Optimize the parameters in the evaluation parameter system based on the algorithm proposed in the previous step to obtain a temporary fitness function value, and calculate the fitness function; Step 5: Based on the updated fitness function value, optimize the UAV flight parameters in the fitness function accordingly; Step Six: Determine the termination condition; if the current time... The loop ends if the loop ends, otherwise update. Record the drone's speed and position at any given moment, and return to step two; Step five specifically involves: Based on the current flight environment, and considering constraints such as flight spacing safety, repulsive field, swarm flight consistency, and spherical / cylindrical obstacle constraints, the parameters are updated and optimized to maximize the fitness function. , It's an indicator about the speed of drone swarms. It is an indicator of the risk of collisions between drones. , It's an indicator about connectivity. It's about collision risk indicators between drones and obstacles. It is an indicator related to the motion correlation of drones; Update the expected velocity vector for the next time step: in It is a drone The expected speed It is the self-driving speed of each drone as an independent individual; It is a drone The speed generated by the combined and repulsive forces of surrounding drones is used to avoid collisions between drones; It is a drone and drones The repulsive velocity between them, and the direction of the repulsive force. Its definition is as follows: These are the linear coefficients to be optimized. This is the safety distance to be optimized; It is a drone A velocity generated under the influence of velocity synchronization resistance; yes and The speed synchronization terms between them, among which These are the linear coefficients to be optimized; This is the maximum permissible speed difference. It is a decay function. and These are all parameters that need optimization. This is the safe distance to be optimized. The definition is as follows: It is a drone The velocity generated by the repulsive force exerted on the obstacle; The model is as follows: in, , It is a drone To virtual drones distance, This is the safe distance to be optimized. and These are parameters that need to be optimized.

2. The method for autonomous trajectory planning of a high-dynamic unmanned aerial vehicle (UAV) swarm according to claim 1, characterized in that, The specific steps of step one are as follows: Randomly generate N drones, the first... The initial coordinates of the drones are , No. The velocity vector of the drone is , No. The drone and the first The relative distance between the drones is , No. The distance of each drone from the origin is , Set the system uptime The drone swarm is constructed as a connected graph, with individual drones as nodes. If the relative distance between drones is less than a given communication distance, an edge is established between them to achieve a distributed architecture. An update control variable time interval is provided. .

3. The method for autonomous trajectory planning of a high-dynamic unmanned aerial vehicle (UAV) swarm according to claim 2, characterized in that, Set the drone velocity vector : Set constraints, number of missing drones This refers to the number of disconnected drones, which is less than one-fifth of the total number of drones. 。 4. The method for autonomous trajectory planning of a high-dynamic unmanned aerial vehicle (UAV) swarm according to claim 3, characterized in that, Step two specifically involves: Step one involves assessing the drone swarm situation, which includes missing drones. The number of missing drones is divided into two categories: drones that were damaged and drones that flew too far. If no drones are damaged or fly too far, the communication topology between drones remains unchanged. If a drone encounters an emergency during flight, such as being attacked or colliding with an obstacle, resulting in damage, it needs to exchange information with each other, discard the damaged drones, optimize the drone swarm structure, detect the relative distance between the drones, and then reconstruct an executable communication topology based on the method in the first step.

5. The method for autonomous trajectory planning of a high-dynamic unmanned aerial vehicle swarm according to claim 3, characterized in that, Step three specifically involves introducing velocity correlation into the evaluation parameter system to evaluate the consistency of the velocity vectors of each UAV, as detailed below: No. One drone represent The drone's velocity vector represent The number of neighbors of the drone; Indicates drone The neighbor subscript set, Represents the upper limit of system runtime. Representing drones The velocity vector; Secondly, based on the flight environment, collision risks between UAVs and collision risks between UAVs and obstacles are introduced into the evaluation parameter system: in It's the Heaviside function. It's the number of drones. It is the safe distance for collisions. It is a drone To virtual drones Distance symbol, when drone Take the negative value within the obstacle; It is the drone at time t With virtual drones The distance between them; Introduce the average speed index of UAVs into the evaluation parameter system to evaluate whether the actual flight speed of UAVs meets the set swarm speed: The above evaluation parameters will be calculated, and the results will be substituted into the next step.

6. The method for autonomous trajectory planning of a high-dynamic unmanned aerial vehicle swarm according to claim 5, characterized in that, Step four specifically involves calculating the fitness function. in ; ； ， ； It's an indicator about the speed of drone swarms. It is an indicator of the risk of collisions between drones. , It's an indicator about connectivity. It is the number of missing drones. It's about collision risk indicators between drones and obstacles. It is an indicator related to the motion correlation of drones; Set the tolerance for cluster speed , Tolerance for collision risk between drones and obstacles .

7. A method for autonomous trajectory planning based on a highly dynamic unmanned aerial vehicle (UAV) swarm, applying any one of claims 1-6. Its characteristics are: Step 1: Initialize the drone, randomly initialize the parameters to be optimized, first set the objective function value F to 0, and the maximum number of iterations itermax; Step 2: Perform the first round of iteration. At each moment during the simulation test time, work in conjunction with the distributed algorithm module to update the speed and position of each UAV. The following (1) to (4) are all calculated and delivered by the distributed algorithm module: (1) Calculation ; (2) Calculation ; (3) Calculation ; (4) Calculation ; Then according to Update the speed of each drone, and then update the drone's location information; Step 3: Calculate the evaluation parameter system as described in the distribution optimization module. and Then, the value of the objective function F is calculated. If the current calculated F value is greater than the F before the iteration, the previous F is changed to the F value now sought, and the corresponding parameters to be optimized are updated. Step 4: The current iteration ends and the next iteration begins. The parameters to be optimized are reset according to certain rules and substituted into the solution until the maximum number of iterations is reached.

8. The system based on the autonomous trajectory planning method for highly dynamic unmanned aerial vehicle swarms according to claim 7, comprising: The flight parameter control module is used to control the size of the UAV swarm during simulation, taking integer values ​​between 1 and 50. The simulation parameter setting module is used to control the upper limit of the test time during simulation; The resilience verification module is used to set the percentage of drone swarms that are attacked, which can be set to a number between 0% and 40% by sliding a slider. The algorithm selection module is used to select a distributed optimization algorithm. The flight direction control module is used to control the flight direction of the drone swarm during simulation, allowing it to rotate freely at 270°.

Images