Unmanned aerial vehicle formation obstacle avoidance control method based on predefined time and event triggering

Abstract

This invention belongs to the field of UAV obstacle avoidance, specifically disclosing a UAV formation obstacle avoidance control method based on predefined time and event triggering. This method designs a predefined time controller based on the dynamic model of the UAV formation system. The specific process is as follows: First, an improved artificial potential field method is designed, incorporating the distance between the UAV and the target point into the obstacle repulsive potential field function, and constructing an artificial potential field force model; then, a convergence time parameter is set, and a predefined time Lyapunov criterion is determined; simultaneously, an event triggering mechanism is designed, updating the control input only when the measurement error is greater than or equal to a trigger threshold constant; finally, adaptive laws for the position subsystem and attitude subsystem are designed based on the backstepping control method. This invention enables rapid formation, stable tracking, and safe obstacle avoidance control of multiple UAVs in complex environments, effectively improving control efficiency and engineering practicality while ensuring system stability.

Description

Technical Field

[0001] This invention belongs to the field of drone obstacle avoidance, specifically relating to a drone formation obstacle avoidance control method based on predefined time and event triggering. Background Technology

[0002] With the rapid development of UAV technology, multi-UAV systems have been widely used in formation flying, cooperative reconnaissance, target tracking, and disaster relief due to their significant advantages in mission execution efficiency and adaptability to complex environments. Therefore, high-performance formation control methods for multi-UAV systems have become a current research hotspot. In multi-UAV formation control, the control objective is usually to enable each following UAV to maintain a predetermined formation structure while accurately tracking the leader's trajectory. To achieve this objective, researchers have proposed various control methods, such as consistency control methods, adaptive control methods, and neural network-based nonlinear control methods.

[0003] These methods can address system uncertainties and external disturbances to some extent, but they still have limitations. First, most existing formation control methods are based on asymptotic stability or finite-time stability theories, whose convergence time depends on the initial state of the system, making it difficult to accurately set the convergence time in advance during the control design phase. In practical tasks, such as time-sensitive tasks (e.g., coordinated strikes or emergency response), the system state needs to converge within a pre-set time, making traditional methods unsuitable for such applications. Second, multi-UAV systems typically rely on communication networks for information exchange, while traditional control strategies often employ periodic sampling or continuous communication mechanisms, leading to high communication resource consumption, network congestion, and increased energy consumption, limiting the system's application capabilities in resource-constrained environments. Furthermore, multi-UAV system dynamics often exhibit strong nonlinearity, uncertainty, and external disturbances, especially in complex environments where accurate models are difficult to obtain. Although some studies use neural networks or adaptive methods to approximate unknowns, a trade-off still exists between ensuring convergence speed, control accuracy, and system stability.

[0004] In summary, existing technologies still have the following shortcomings in the control of multi-UAV formations: First, it is difficult to achieve control performance with a pre-set convergence time; second, the communication burden is heavy, making it difficult to balance control performance and communication efficiency; and third, the robustness to system uncertainties and external disturbances still needs to be improved. Summary of the Invention

[0005] The purpose of this invention is to propose a UAV formation obstacle avoidance control method based on predefined time and event triggering. This method designs an improved artificial potential field method, a predefined time Lyapunov criterion, and an event triggering mechanism, and combines an adaptive law with a backstepping control method to enable the UAV formation system to complete the formation tracking task within a predefined time, while reducing the communication burden and improving the system robustness.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: The drone formation obstacle avoidance control method based on predefined time and event triggers includes the following steps: Step 1. Establish a dynamic model of the UAV formation system, including a position subsystem model and an attitude subsystem model; Step 2. Based on the UAV formation system dynamics model established in Step 1, design a predefined time controller, including a predefined time controller for the position subsystem and a predefined time controller for the attitude subsystem. The specific process is as follows: First, an improved artificial potential field method is designed, which introduces the distance between the UAV and the target point into the obstacle repulsive potential field function, and constructs an artificial potential field force model. Then, the convergence time parameter is set, and the predefined time Lyapunov criterion is determined; At the same time, an event triggering mechanism is designed to update the control input only when the measurement error is greater than or equal to the trigger threshold constant; Finally, adaptive laws for the position subsystem and attitude subsystem are designed based on the backstep control method. Step 3. Use the predefined time controller designed in Step 2 to realize the real-time obstacle avoidance movement of the drone formation.

[0007] As described above, this invention proposes a UAV formation obstacle avoidance control method based on predefined time and event triggering. This method sets a convergence time parameter and introduces a predefined time Lyapunov criterion, allowing the system convergence time to be pre-set during the control design phase and independent of the system's initial state, thus meeting the requirements of time-sensitive tasks. Furthermore, this invention combines an improved artificial potential field method, introducing a distance adjustment term on top of the traditional potential field function. This gradually weakens the repulsive force as the UAV approaches the target point, avoiding the problem of the UAV being unable to reach the target when an obstacle approaches. This achieves a unified design of formation control and collision avoidance constraints, improving system safety. Simultaneously, this invention designs a time-triggered mechanism, updating the control input only when the measurement error is greater than or equal to a trigger threshold constant. This significantly reduces the update frequency of the control input and the communication burden, improving the system's applicability under resource-constrained conditions. This invention employs a radial basis function neural network to approximate unknown nonlinear terms and external disturbances online, improving the system's adaptability and robustness to uncertainties. Attached Figure Description

[0008] Figure 1 This is a flowchart of the UAV formation obstacle avoidance control method based on predefined time and event triggering in an embodiment of the present invention; Figure 2 This is a control block diagram of the unmanned aerial vehicle (UAV) formation system in an embodiment of the present invention; Figure 3 This is a simulated communication topology diagram in an embodiment of the present invention; Figure 4 This is a three-dimensional obstacle avoidance trajectory diagram of the UAV formation after adopting the control method of the present invention; Figure 5 The diagram shows the obstacle avoidance trajectory of the UAV formation in the inertial coordinate system after adopting the control method of the present invention. Figure 6 The position tracking response curve is shown after applying the control method of the present invention; where (a) is... The position tracking response curve in the axial direction, (b) is The position tracking response curve in the axial direction, (c) is Position tracking response curve in the axial direction; Figure 7 The speed tracking response curve is shown after applying the control method of the present invention; where (a) is... The velocity tracking response curve in the axial direction, (b) is The velocity tracking response curve in the axial direction, (c) is axial velocity tracking response curve; Figure 8 The following are attitude angle tracking response curves after adopting the control method of the present invention; where (a) is the pitch angle tracking response curve, (b) is the yaw angle tracking response curve, and (c) is the roll angle tracking response curve. Figure 9 The figure shows the tracking error response curve of the position subsystem after adopting the control method of the present invention; where (a) is... Position tracking error in the axial direction, (b) is The position tracking error in the axial direction, (c) is Position tracking error in the axial direction; Figure 10 The following are the tracking error response curves of the attitude subsystem after adopting the control method of the present invention; where (a) is the pitch tracking error, (b) is the yaw tracking error, and (c) is the roll tracking error. Figure 11 The diagram shows the control input simulation after applying the control method of this invention; where (a) is... The simulation diagram, (b) is The simulation diagram, (c) is The simulation diagram, (d) is Simulation diagram; Figure 12 This is a stem-and-trough diagram of event triggering for UAV1 after adopting the control method of the present invention; where (a) is The event-triggered stem diagram, (b) is The event-triggered stem diagram, (c) is The event-triggered stem diagram, (d) is The event-triggered stem graph; Figure 13 This is a stem-and-trough diagram of event triggering in UAV2 after adopting the control method of the present invention; where (a) is The event-triggered stem diagram, (b) is The event-triggered stem diagram, (c) is The event-triggered stem diagram, (d) is The event-triggered stem graph; Figure 14 This is a stem-and-trough diagram of event triggering in UAV3 after adopting the control method of the present invention; where (a) is... The event-triggered stem diagram, (b) is The event-triggered stem diagram, (c) is The event-triggered stem diagram, (d) is The event triggers the stem graph. Detailed Implementation

[0009] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: Example 1 For multi-UAV formation systems that consider disturbances and nonlinear terms, this embodiment 1 describes a UAV formation obstacle avoidance control method based on predefined time and event triggering. This method combines an event triggering mechanism and an artificial potential field method to achieve safe operation of UAV formations in complex environments.

[0010] like Figure 1 As shown, the drone formation obstacle avoidance control method based on predefined time and event triggering includes the following steps: Step 1. Establish a dynamic model of the UAV formation system, including a position subsystem model and an attitude subsystem model.

[0011] In the drone formation system, the first The dynamic model of the drone is represented as follows: (1) in, Indicates the first The acceleration vector of the drone; Indicates the first The position vector of the drone, , , , They represent the first A drone in axis, axis, Position component in the axial direction.

[0012] Indicates the first The attitude angular acceleration vector of the drone; Indicates the first The attitude angle vector of the drone. , Indicates the first The pitch angle of the drone, Indicates the first The yaw angle of the drone, Indicates the first The roll angle of the drone.

[0013] Indicates the first The position control input vector of the UAV is formed by the total thrust.

[0014] Indicates the first The attitude control input vector of the drone. , , , They represent the first The control inputs for the pitch angle, yaw angle, and roll angle of the drone.

[0015] This represents the bounded external perturbation vector acting on the position subsystem. , , , They represent exist axis, axis, Components in the axial direction.

[0016] This represents the time-varying bounded external perturbation vector acting on the attitude subsystem. , , , They represent The components of pitch angle, yaw angle, and roll angle.

[0017] This represents the position channel input allocation vector determined by the attitude angle. ; in .

[0018] Represents a three-dimensional identity matrix.

[0019] This represents the unknown lumped nonlinear term acting on the position subsystem. ; in ; , , These represent the air drag coefficients in the three directions of the position subsystem, respectively; Indicates the total mass of the drone; Represents gravitational acceleration; , , This represents the uncertainty term in the location subsystem model.

[0020] This represents the unknown lumped nonlinear term acting on the attitude subsystem. ; in, ; , , These represent the damping coefficients on the three corner channels of the attitude subsystem; , , They represent the first The rotational inertia of the drone about its three principal axes; This represents the distance from the center of each rotor blade to the center of mass of the fuselage; , , This represents the uncertainty term in the attitude subsystem model.

[0021] Indicates the drone's serial number. , This indicates the number of drones following in the formation.

[0022] Since the UAV is an underactuated system with six outputs and four inputs, it is necessary to solve the inverse problem by analyzing the inputs of the position subsystem.

[0023] The virtual controller for the location subsystem is defined as follows: (2) in, , , They represent the first The position control input of the drone is in axis, axis, The component along the axial direction.

[0024] The position subsystem model and attitude subsystem model are shown in Equation (3) and Equation (4) respectively: (3) (4) in, Indicates the first The velocity vector of the drone , , , They represent exist axis, axis, Components along the axial direction; ; Indicates the first The attitude angular velocity vector of the drone. .

[0025] Step 2. Based on the UAV formation system dynamics model established in Step 1, design a predefined time controller, including a predefined time controller for the position subsystem and a predefined time controller for the attitude subsystem. The specific process is as follows: First, an improved artificial potential field method is designed, which introduces the distance between the UAV and the target point into the obstacle repulsive potential field function, and constructs an artificial potential field force model. Then, the convergence time parameter is set, and the predefined time Lyapunov criterion is determined; At the same time, an event triggering mechanism is designed to update the control input only when the measurement error is greater than or equal to the trigger threshold constant; Finally, adaptive laws for the position subsystem and attitude subsystem are designed based on the backstep control method.

[0026] Currently, the artificial potential field method has been widely applied to obstacle avoidance in multi-UAV swarm systems, achieving good control results. However, traditional artificial potential fields suffer from local minima and the inability to reach the target point. Therefore, to address the safe flight problem of multi-UAV swarms in obstacle-prone environments, this invention introduces an improved artificial potential field method to construct a target attraction potential field function. The artificial potential field force is generated through the synthesis of the attraction and repulsion potential fields, serving as the guiding force for obstacle avoidance control.

[0027] The specific process of constructing an artificial potential field force model is as follows: Define the distance function between the drone and the obstacle. for: ; in, Indicates the first Location of the drone; Indicates the first The location of the obstacle. , This indicates the number of obstacles present in the operating space of the drone formation. , , , The first An obstacle in axis, axis, Position components along the axial direction; This indicates a search for the norm.

[0028] Define the distance function between the UAV and the target point. for: ; in, Indicates the location of the target point. , , , The target points are respectively at axis, axis, Position component in the axial direction.

[0029] Construct the repulsive potential field function of the obstacle for: (5) in, This represents the set repulsion gain coefficient. This indicates the effective distance range of the repulsive force exerted by the set obstacle.

[0030] The obstacle repulsive potential field function introduces a distance adjustment term based on the traditional repulsive potential field function. This causes the repulsive force to gradually weaken as the drone approaches the target point, thus avoiding the problem of "obstacles approaching the target point causing the drone to be unable to reach the target".

[0031] Construct the target attractive potential field function for: (6) in, This represents the set attraction gain coefficient. This represents the boundary threshold of the set attractive potential field.

[0032] The target attraction potential field function is designed as a piecewise function to reduce the attraction intensity when the UAV moves away from the target point, thereby reducing the risk of collision.

[0033] Taking the negative gradient of the obstacle repulsive potential field function yields the obstacle repulsive vector. Its expression is as follows: (7) Taking the negative gradient of the target attraction potential function yields the target attraction vector. Its expression is as follows: (8) When the environment exists When there is an obstacle, the first... Total repulsive force on the drone for: (9) Then the first The drone is at its current location Artificial potential force at the location for: (10) This combined force As an obstacle avoidance compensation term, it is introduced into the controller design to enable real-time obstacle avoidance movement of the drone formation.

[0034] Under the condition that the drone maintains a non-zero safe distance from obstacles and the operating space is bounded, the artificial potential field force is bounded, that is, there exists a normal number. ,satisfy: .

[0035] Assumption It is a continuous function with a weight matrix. and basis function vectors , so that: ;in This is the neural network approximation error term.

[0036] The basis functions are Gaussian radial basis functions, and their expressions are as follows: ; in, Denotes the center vector of the basis functions. Z represents the basis function width parameter, and Z represents the input vector of the radial basis function neural network.

[0037] To describe the information interaction relationships among multiple UAVs, a graph theory approach is used to model the communication topology of the UAV formation.

[0038] The communication topology of a drone formation system is represented as a weighted graph. ; in Represents a set of drone nodes. , Indicates the first One drone node; Represents the set of communication edges. ; Represents the adjacency matrix. , Indicates the first drones and the first Communication weights between drones.

[0039] when At that time, drones Able to receive signals from drones Information; At that time, drones With drones There is no information exchange connection between them.

[0040] Furthermore, define the degree matrix. for: ;in, Indicates the first The weighted degree of each node. .

[0041] The Laplace matrix of the communication topology is defined based on the adjacency matrix and the degree matrix. for: ; Used to describe the relative coupling relationship between individual drone nodes in a drone formation.

[0042] For a leader-follower structured drone swarm system, a virtual leader adjacency matrix is ​​defined. for: ; in, Indicates the first The weight of the connection between the drone and the virtual leader. When the... When a drone connects directly to a virtual leader When the first When the drone is not directly connected to the virtual leader .

[0043] In this embodiment, the communication topology diagram It is a connected graph, and there exists at least one drone that can acquire information about the virtual leader.

[0044] Define systematic error , , , They are respectively: ; in, Indicates the first The position vector of the drone, It is the position vector of the virtual leader. , , , Virtual leaders in axis, axis, Position component in the axial direction.

[0045] It is the first The expected relative distance between drones and virtual leaders , , , They are respectively exist axis, axis, Components along the axial direction; It is the first The expected relative distance between a drone and a virtual leader.

[0046] and These are the virtual control laws for the position subsystem and the attitude subsystem, respectively.

[0047] Indicates the first The attitude angle vector of the drone; It is the desired attitude angle of the drone formation. ; Indicates the first The desired pitch angle of the drone. Indicates the first The expected yaw angle of the drone. Indicates the first The expected roll angle of the drone.

[0048] Regarding the first Select a drone , The expected roll-off angle provided to leaders.

[0049] Due to the underactuated nature, the desired yaw and pitch angles cannot be directly input. Therefore, by inverse solving equation (2), the first... Desired pitch angle of the drone and the Desired yaw angle of the drone Since there is no direct input, this invention only requires that these two angles achieve stability based on practical considerations. The drone formation of this invention requires that the attitude angles of each drone be the same; therefore, the desired attitude angle of the first drone is selected. This serves as a reference to demonstrate that tracking can be completed at each attitude angle.

[0050] in This represents the expected pitch angle of the first drone. This represents the expected yaw angle of the first drone. This represents the expected roll angle of the first drone.

[0051] The inverse solution process is as follows: ; ; .

[0052] This invention enables the state error of the UAV formation system to converge to the bounded residual set within a preset time by pre-setting the convergence time parameter and designing the position subsystem control law and attitude subsystem control law based on the predefined time Lyapunov criterion. Moreover, the upper bound of the convergence time does not depend on the initial state of the system.

[0053] The predefined time Lyapunov criterion is stated as follows: ; in, , For design parameters, , ; The bounded term is caused by disturbances, approximation errors, and event-triggered errors. ; To preset the convergence time parameter, Indicates information about the system state A positive definite Lyapunov function.

[0054] If the above predefined time Lyapunov criterion is satisfied, then the system state is... It converges within the actual predefined time, which is: And system status Ultimately, they all converge to the residual set. In China; among them It is a positive number. .

[0055] During the operation of drone formations, excessive communication resources are consumed. Therefore, this invention introduces an event triggering mechanism based on a fixed threshold to save communication resources. At the same time, the controller is designed based on the predefined time Lyapunov criterion to ensure that the various states of the drones can converge within a predefined time.

[0056] Based on a fixed threshold, the following event triggering mechanism is defined: (11) in, To control the input, Indicates the first The control input sampled and sent at the next trigger moment; This indicates the measurement error caused by the event. , This indicates the control input to be triggered.

[0057] Indicates the current moment. Indicates the first Next trigger time Indicates the first Next trigger time .

[0058] This represents a pre-set trigger threshold constant. ; The sum of the absolute values ​​of all elements in the vector is represented by ; inf{} represents the infimum of the set of times that satisfy the conditions within the parentheses, i.e., the earliest triggering time.

[0059] At the trigger time Control input Updated and sent to the drone flight controller actuator, within the range Within this period, the control input remains unchanged until the next trigger time. arrival.

[0060] From formula (11), we can see that in the interval Internal satisfaction: .

[0061] Therefore, there exists a time-varying parameter vector. This makes for have: , , ,and: ; in, , , They are respectively exist axis, axis, Components in the axial direction, , They are respectively , The time-varying parameter vector at time t.

[0062] The event-triggered mechanism does not continuously update the control input at every moment. Instead, it only updates the control quantity when the measurement error is greater than or equal to the trigger threshold constant, thereby reducing the communication frequency and the update burden of the UAV flight controller actuator while ensuring system stability.

[0063] The backstepping method requires constructing a virtual control law based on the Lyapunov function, and then designing the actual control input through the virtual control law.

[0064] The specific process for designing a predefined time controller for the position subsystem is as follows: Selecting the first-level Lyapunov function for: .

[0065] From the position subsystem model and the definition of systematic error, we get: ; in, , , Indicates the first The velocity vector of the drone This represents the velocity vector of the virtual leader.

[0066] but derivative for: (12) To ensure that the position error meets the predefined time convergence characteristics, a virtual control law for the position subsystem is designed. for: (13) Substituting the above virtual control law, i.e., formula (13), into formula (12), we get: (14) Constructing extended Lyapunov functions for: (15) Substituting formula (15) into formula (14), we get The derivative satisfies: (16) The unknown terms and disturbance terms of the position subsystem are combined to obtain Its expression is: .

[0067] Radial basis function neural network is used to By approximation, we obtain: ; in, Weight matrix, Represents a basis function vector. This represents the approximation error.

[0068] After further calculation, we get: ; in, , , , They are respectively exist axis, axis, Components along the axial direction; , The number of nodes in the radial basis function neural network; This indicates that the absolute value of each element in the vector is taken.

[0069] The event triggering mechanism, control input, and adaptive law of the position subsystem are as follows: (17) in, For the first The actual control input of the UAV position subsystem. Indicates the position subsystem at the 1st The control input is updated and sent at the next trigger moment.

[0070] , This is the control input to be triggered for the position subsystem; , , , They are respectively exist axis, axis, Components in the axial direction.

[0071] This represents the pre-defined trigger threshold constant for the location subsystem. For the constant to be designed, ; for The estimated value, , ; .

[0072] It is a saturation function. , , They are respectively exist axis, axis, Components along the axial direction; This represents a diagonal matrix operator used to arrange the elements within the brackets sequentially along the main diagonal of a matrix, with the remaining elements being 0.

[0073] The expression for the saturation function is as follows: ; in , , Indicates precision.

[0074] The specific process of designing a predefined time controller for the attitude subsystem is as follows: Selecting the first-level Lyapunov function for: .

[0075] From the attitude subsystem model and the definition of system error, we get: ; in, , Indicates the first The attitude angular velocity vector of the drone. This represents the attitude angular velocity vector corresponding to the desired attitude angle.

[0076] but The derivative is: (18) Design the virtual control law for the attitude subsystem. for: (19) Substituting the above virtual control law, i.e., formula (19), into formula (18), we get: (20) To further address velocity errors, an extended Lyapunov function is constructed. for: (twenty one) Substituting formula (21) into formula (20), we get The derivative satisfies: (twenty two) To facilitate subsequent approximation, the unknown terms and perturbation terms of the attitude subsystem are combined to obtain... Its expression is: .

[0077] Radial basis function neural network is used to By approximation, we obtain: ; in, Represents the weight matrix. Represents a basis function vector. This represents the approximation error.

[0078] After further calculation, we get: ; in, , , , They are respectively The components of pitch angle, yaw angle, and roll angle; .

[0079] The event triggering mechanism, control input, and adaptive law of the attitude subsystem are as follows: (twenty three) in, For the first The actual control inputs of the UAV attitude subsystem The attitude subsystem is represented in the first... The control input is updated and sent at the next trigger moment.

[0080] , This is the control input to be triggered for the position subsystem; This represents the pre-set attitude subsystem trigger threshold constant. For the constant to be designed, .

[0081] for The estimated value, ; , , , They are respectively exist axis, axis, Components in the axial direction.

[0082] After completing the design of the predefined time controller for the position subsystem and the predefined time controller for the attitude subsystem, stability analysis was performed on the predefined time controller for the position subsystem and the predefined time controller for the attitude subsystem, proving that they can be stable within the predefined time and that no Zeno phenomenon occurs.

[0083] The specific process for performing stability analysis on the predefined time controller of the position subsystem is as follows: Substituting formula (17) into formula (16) yields: ; in , This is a time-varying parameter vector.

[0084] exist This makes for have: , , ; in, , They are respectively , The time-varying parameter vector at time t.

[0085] According to the inequality: ,have to: (twenty four) in, .

[0086] According to the inequality: ,have to: (25) Therefore, formula (25) can be written as: (26) Choosing Lyapunov functions for: and to Differentiating, we get: (27) Therefore, all positional errors of the drone formation are within a predefined time. Internal convergence means that the positional subsystem is stable within a predefined time.

[0087] definition ;in, , , They represent exist axis, axis, Components in the axial direction.

[0088] but: ; Where sign represents the sign function. , , , They are respectively exist axis, axis, Components in the axial direction.

[0089] because If it is a differentiable bounded function, then It is also bounded, therefore, it exists. Make: .

[0090] because , There exist positive numbers satisfy Therefore, the position subsystem will not exhibit the Zeno phenomenon.

[0091] The specific process of performing stability analysis on the predefined time controller of the attitude subsystem is as follows: Substituting formula (23) into formula (22) yields: ; in , This is a time-varying parameter vector.

[0092] exist This makes for have: , , ; in , They are respectively , The time-varying parameter vector at time t.

[0093] According to the inequality: ,have to: (28) in, .

[0094] According to the inequality: ,have to: (29) Therefore, formula (29) can be written as: (30) Choosing Lyapunov functions for: and to Differentiating, we get: (31) Therefore, all attitude errors of the drone formation are within a predefined time. Internal convergence means that the attitude subsystem is stable within a predefined time.

[0095] definition ,in , , They represent The components of pitch angle, yaw angle, and roll angle.

[0096] but: ; , , , They are respectively The components of pitch angle, yaw angle, and roll angle.

[0097] because If it is a differentiable bounded function, then It is also bounded, therefore, it exists. Make: .

[0098] because , There exist positive numbers satisfy Therefore, the attitude subsystem will not exhibit the Zeno phenomenon.

[0099] In summary, all states of the drone formation satisfy the predefined time stability and do not exhibit the Zeno phenomenon.

[0100] Step 3. Use the predefined time controller designed in Step 2 to realize the real-time obstacle avoidance movement of the drone formation.

[0101] This invention addresses the challenge of drone formations struggling to form ranks, track virtual leaders, and safely avoid obstacles within a predefined timeframe in complex environments with obstacles and limited communication resources. It proposes a drone formation obstacle avoidance control method based on predefined timeframes and event triggers. Figure 2As shown, the system first acquires the state information of each following UAV and obstacle. The trajectory information of the virtual leader is determined through pre-set formation reference commands. Each following UAV transmits its position and attitude signals to the communication network (i.e., the communication topology). A position subsystem and an attitude subsystem are constructed based on the UAV formation system dynamics model. Then, an improved artificial potential field method is used to calculate the target attraction and obstacle repulsion, which are then combined into an artificial potential field force to guide the UAVs to avoid obstacles while tracking the target trajectory. Furthermore, a predefined time controller is designed using the predefined time Lyapunov criterion, enabling the UAV formation error to converge within a preset time. Simultaneously, a radial basis function neural network is used to approximate the unknown nonlinear terms and external disturbances of the system online, and the estimated parameters are updated using an adaptive law to improve the system's robustness. Finally, an event triggering mechanism is introduced, updating and sending control input signals to the UAV flight controller actuator only when the measurement error is greater than or equal to a trigger threshold constant. This reduces the communication burden and control update frequency while achieving rapid formation, stable tracking, and safe obstacle avoidance of the UAV formation.

[0102] To verify the effectiveness of the method proposed in this invention, the following specific experiments are presented: Figure 3 This is the topology diagram of the leader and followers in this experiment. The numbers "1", "2", and "3" in the diagram represent three drones, UAV1, UAV2, and UAV3, respectively, i.e., n=3. The number "0" represents the virtual leader.

[0103] The control method proposed in this invention will be simulated in a virtual environment to verify its feasibility.

[0104] The parameters are selected as follows: , , ; , ; , ; ; , , ; , , ; , .

[0105] The controller-related parameters are selected as follows: , , , , , , , , , , .

[0106] The virtual leader's trajectory, i.e., the formation reference command, is as follows: ; ; ; ;in , , Virtual leaders in axis, axis, Position component in the axial direction.

[0107] The desired formation offset is: , , .

[0108] The initial state selection for the three follower drones is as follows: ; ; in Indicates the first The initial position of the drone. Indicates the first The initial attitude angle of the drone.

[0109] The simulation results of the UAV formation obstacle avoidance control method based on predefined time and event triggering used in this experiment are as follows: Figure 4-14 As shown in the diagram, "VL" represents the virtual leader. In this experiment, a predefined time... Set to 3 seconds.

[0110] from Figure 4-5 As can be seen, the control method employed in this invention enables drone formations to complete their formation and effectively avoid obstacles in complex environments. Figure 6-10 It can be seen that the various states (position, velocity, attitude angle) of the UAV formation can be effectively tracked, and the tracking error of each state can converge within a predefined time. Figure 6-10 The fluctuations are caused by the drone deviating from the desired point during obstacle avoidance. After obstacle avoidance is completed, the drone will return to the desired position and angle. Figure 11 As shown, the control input in this experiment is stepped, which conforms to the characteristic of the event triggering mechanism that triggers again after reaching the error condition. Figure 12-14The experiment demonstrates the trigger times of the predefined time controllers for three UAVs (UAV1, UAV2, and UAV3). The total number of trigger events for the three UAVs (UAV1, UAV2, and UAV3) are 37410, 9222, and 10850, respectively, with a sampling interval of 0.001 seconds. The event triggering mechanism proposed in this invention saves 86.3% of communication resources. The experiment clearly shows that this invention enables UAV formations to converge and fly safely and effectively within a predefined timeframe, even with obstacles and limited communication resources.

[0111] This invention addresses the challenges of uncontrollable convergence time, difficulty in obstacle avoidance, and limited communication resources in multi-UAV formations in complex environments. By introducing a predefined time, it achieves rapid convergence of the formation system within a user-specified timeframe. Combined with an improved artificial potential field method, it effectively avoids the local minima problem inherent in traditional potential field methods, thus improving obstacle avoidance performance. Simultaneously, an event-triggered control mechanism significantly reduces communication frequency, minimizing system resource consumption. Furthermore, this invention utilizes a radial basis function neural network to approximate and compensate for system uncertainties, enhancing system robustness. This invention enables rapid formation, stable tracking, and safe obstacle avoidance control of multiple UAVs in complex environments, effectively improving control efficiency and engineering practicality while ensuring system stability.

[0112] Of course, the above description is only a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. It should be noted that any equivalent substitutions or obvious modifications made by those skilled in the art under the guidance of this specification fall within the scope of this specification and should be protected by the present invention.

Claims

1. A method for drone formation obstacle avoidance control based on predefined time and event triggering, characterized in that, Includes the following steps: Step 1. Establish a dynamic model of the UAV formation system, including a position subsystem model and an attitude subsystem model; Step 2. Based on the UAV formation system dynamics model established in Step 1, design a predefined time controller, including a predefined time controller for the position subsystem and a predefined time controller for the attitude subsystem. The specific process is as follows: First, an improved artificial potential field method is designed, which introduces the distance between the UAV and the target point into the obstacle repulsive potential field function, and constructs an artificial potential field force model. Then, the convergence time parameter is set, and the predefined time Lyapunov criterion is determined; At the same time, an event triggering mechanism is designed to update the control input only when the measurement error is greater than or equal to the trigger threshold constant; Finally, adaptive laws for the position subsystem and attitude subsystem are designed based on the backstep control method. Step 3. Use the predefined time controller designed in Step 2 to realize the real-time obstacle avoidance movement of the drone formation.

2. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 1, characterized in that, Step 1 specifically involves: In the drone formation system, the first The dynamic model of the drone is represented as follows: (1) in, Indicates the first The acceleration vector of the drone; Indicates the first The position vector of the drone, , , , They represent the first A drone in axis, axis, Position components along the axial direction; Indicates the first The attitude angular acceleration vector of the drone; Indicates the first The attitude angle vector of the drone. , Indicates the first The pitch angle of the drone, Indicates the first The yaw angle of the drone, Indicates the first The roll angle of the drone; Indicates the first The position control input vector of the UAV is formed by the total thrust. Indicates the first The attitude control input vector of the drone; , , , They represent the first The control inputs for the pitch angle, yaw angle, and roll angle of the UAV; This represents the bounded external perturbation vector acting on the position subsystem; This represents the time-varying bounded external perturbation vector acting on the attitude subsystem; This represents the position channel input allocation vector determined by the attitude angle; Represents a three-dimensional identity matrix; This represents the unknown lumped nonlinear term acting on the position subsystem; This represents the unknown lumped nonlinear term acting on the attitude subsystem; Indicates the drone's serial number. , Indicates the number of drones following in the formation; The virtual controller for the location subsystem is defined as follows: (2) in, , , They represent the first The position control input of the drone is in axis, axis, Components along the axial direction; The position subsystem model and attitude subsystem model are shown in Equation (3) and Equation (4) respectively: (3) (4) in, Indicates the first The velocity vector of the drone , , , They represent exist axis, axis, Components along the axial direction; ; Indicates the first The attitude angular velocity vector of the drone. .

3. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 2, characterized in that, In step 2, the specific process of constructing the artificial potential field force model is as follows: Define the distance function between the drone and the obstacle. for: ； in, Indicates the first Location of the drone; Indicates the first The location of the obstacle. , This indicates the number of obstacles present in the operating space of the drone formation; , , , The first An obstacle in axis, axis, Position components along the axial direction; This indicates a search for the norm. Define the distance function between the UAV and the target point. for: ； in, Indicates the location of the target point. , , , The target points are respectively at axis, axis, Position components along the axial direction; Construct the repulsive potential field function of the obstacle for: (5) in, This represents the set repulsion gain coefficient. Indicates the effective distance range of the repulsive force of the set obstacle; Construct the target attractive potential field function for: (6) in, This represents the set attraction gain coefficient. This represents the boundary threshold of the set attractive potential field; Taking the negative gradient of the obstacle repulsive potential field function yields the obstacle repulsive vector. Its expression is as follows: (7) Taking the negative gradient of the target attraction potential function yields the target attraction vector. Its expression is as follows: (8) When the environment exists When there is an obstacle, the first... Total repulsive force on the drone for: (9) Then the first The drone is at its current location Artificial potential force at the location for: (10) Under the condition that the drone maintains a non-zero safe distance from obstacles and the operating space is bounded, the artificial potential field force is bounded, that is, there exists a normal number. ,satisfy: .

4. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 3, characterized in that, In step 2, Define systematic error , , , They are respectively: ； in, Indicates the first The position vector of the drone, It is the position vector of the virtual leader. , , , Virtual leaders in axis, axis, Position components along the axial direction; It is the first The expected relative distance between drones and virtual leaders , , , They are respectively exist axis, axis, Components along the axial direction; It is the first The expected relative distance between drones and virtual leaders; Indicates the first The attitude angle vector of the drone. It is the desired attitude angle of the drone formation. ; Indicates the first The expected roll angle of the drone. , The expected roll-off angle provided for leaders; Indicates the first The expected yaw angle of the drone. ; Indicates the first The desired pitch angle of the drone; ； ； and These are the virtual control laws for the position subsystem and the attitude subsystem, respectively. Indicates the first drones and the first Communication weights between drones; Indicates the first The weight of the connection between drones and virtual leaders; The predefined time Lyapunov criterion is stated as follows: ; in, , For design parameters, , ; The bounded term is caused by disturbances, approximation errors, and event-triggered errors. ; To preset the convergence time parameter, Indicates information about the system state Positive definite Lyapunov functions; If the above predefined time Lyapunov criterion is satisfied, then the system state is... It converges within the actual predefined time, which is: And system status Ultimately, they all converge to the residual set. In China; among them It is a positive number. .

5. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 4, characterized in that, In step 2, the event triggering mechanism is specifically as follows: (11) in, To control the input, Indicates the first The control input sampled and sent at the next trigger moment; This indicates the measurement error caused by the event. , Indicates the control input to be triggered; Indicates the current moment. Indicates the first Next trigger moment Indicates the first Next trigger moment ; This represents a pre-set trigger threshold constant. ; This represents the sum of the absolute values ​​of all elements in the vector; inf{} denotes the infimum of the set of moments for the conditions within the parentheses; At the trigger time Control input Updated and sent to the drone flight controller actuator, within the range Within this period, the control input remains unchanged until the next trigger time. arrival; From formula (11), we can see that in the interval... Internal satisfaction: ; Therefore, there exists a time-varying parameter vector. This makes for have: , , ,and: ; in, , , They are respectively exist axis, axis, Components in the axial direction, , They are respectively , The time-varying parameter vector at time t.

6. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 5, characterized in that, In step 2, the specific process of designing the predefined time controller for the position subsystem is as follows: Selecting the first-level Lyapunov function for: ; From the position subsystem model and the definition of systematic error, we get: ; in, ; , Indicates the first The velocity vector of the drone; Represents the velocity vector of the virtual leader; but derivative for: (12) Design the virtual control law for the position subsystem. for: (13) Substituting the above virtual control law, i.e., formula (13), into formula (12), we get: (14) Constructing extended Lyapunov functions for: (15) Substituting formula (15) into formula (14), we get The derivative satisfies: (16) The unknown terms and disturbance terms of the position subsystem are combined to obtain Its expression is: ; Radial basis function neural network is used to By approximation, we obtain: ; in, Weight matrix, Represents a basis function vector. Indicates the approximation error; After further calculation, we get: ; in, , , , They are respectively exist axis, axis, Components along the axial direction; , The number of nodes in the radial basis function neural network; This indicates that the absolute value of each element in the vector is taken; The event triggering mechanism, control input, and adaptive law of the position subsystem are as follows: (17) in, Indicates the first The actual control input of the UAV position subsystem. The position subsystem is in the first... The control input is updated and sent at the next trigger moment; , This indicates the control input to be triggered for the position subsystem; , , , They are respectively exist axis, axis, Components along the axial direction; This represents the pre-defined trigger threshold constant for the location subsystem. Indicates the constant to be designed. ; for The estimated value, , ; ; It is a saturation function. , , They are respectively exist axis, axis, Components along the axial direction; This represents a diagonal matrix operator.

7. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 6, characterized in that, In step 2, the specific process of designing the predefined time controller for the attitude subsystem is as follows: Selecting the first-level Lyapunov function for: ; From the attitude subsystem model and the definition of system error, we get: ; in, , Indicates the first The attitude angular velocity vector of the drone. This represents the angular velocity vector corresponding to the desired attitude angle; but The derivative is: (18) Design the virtual control law for the attitude subsystem. for: (19) Substituting the above virtual control law, i.e., formula (19), into formula (18), we get: (20) Constructing extended Lyapunov functions for: (21) Substituting formula (21) into formula (20), we get The derivative satisfies: (22) The unknowns and disturbances of the attitude subsystem are combined to obtain Its expression is: ; Radial basis function neural network is used to By approximation, we obtain: ; in, Represents the weight matrix. Represents a basis function vector. Indicates the approximation error; After further calculation, we get: ; in, , , , They are respectively The components of pitch angle, yaw angle, and roll angle; ； The event triggering mechanism, control input, and adaptive law of the attitude subsystem are as follows: (23) in, For the first The actual control inputs of the UAV attitude subsystem The attitude subsystem is represented in the first... The control input is updated and sent at the next trigger moment; , This is the control input to be triggered for the attitude subsystem; This represents the pre-set attitude subsystem trigger threshold constant. Indicates the constant to be designed. ; for The estimated value, ; , , , They are respectively exist axis, axis, Components in the axial direction.

8. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 7, characterized in that, In step 2, after completing the design of the predefined time controller for the position subsystem and the predefined time controller for the attitude subsystem, a stability analysis is performed on the predefined time controller for the position subsystem and the predefined time controller for the attitude subsystem.

9. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 8, characterized in that, In step 2, the specific process of performing stability analysis on the predefined time controller of the position subsystem is as follows: Substituting formula (17) into formula (16) yields: ； in , A time-varying parameter vector; exist This makes for have: , , ; in, , They are respectively , The time-varying parameter vector at time t; According to the inequality: ,have to: (24) in, ; According to the inequality: ,have to: (25) Therefore, formula (25) can be written as: (26) Choosing Lyapunov functions for: and to Differentiating, we get: (27) Therefore, all positional errors of the drone formation are within a predefined time. Internal convergence means that the position subsystem is stable within a predefined time. definition ,in, , , They represent exist axis, axis, Components along the axial direction; but: ; Where sign represents the sign function; , , , They are respectively exist axis, axis, Components along the axial direction; because If it is a differentiable bounded function, then It is also bounded, therefore, it exists. Make: ; because , There exist positive numbers satisfy Therefore, the position subsystem will not exhibit the Zeno phenomenon.

10. The UAV formation obstacle avoidance control method based on predefined time and event triggering according to claim 9, characterized in that, In step 2, the specific process of performing stability analysis on the predefined time controller of the attitude subsystem is as follows: Substituting formula (23) into formula (22) yields: ； in , A time-varying parameter vector; exist This makes for have: , , ; in , They are respectively , The time-varying parameter vector at time t; According to the inequality: ,have to: (28) in, ; According to the inequality: ,have to: (29) Therefore, formula (29) can be written as: (30) Choosing Lyapunov functions for: and to Differentiating, we get: (31) Therefore, all attitude errors of the drone formation are within a predefined time. Internal convergence means that the attitude subsystem is stable within a predefined time. definition ,in , , They represent The components of pitch angle, yaw angle, and roll angle; but: ; in , , , They are respectively The components of pitch angle, yaw angle, and roll angle; because If it is a differentiable bounded function, then It is also bounded, therefore, it exists. Make: ; because , There exist positive numbers satisfy Therefore, the attitude subsystem will not exhibit the Zeno phenomenon.

Images