Nonlinear robust tracking control method for multi-UAV system based on position measurement information

Abstract

The present invention relates to the control constraint of a four-rotor UAV formation. Through the anti-disturbance tracking control design of the UAV formation based on azimuth information, a new nonlinear controller is proposed to complete the formation formation and tracking tasks of a multi-UAV system in view of the situation that each UAV can only obtain the azimuth information of the neighboring aircraft. For this reason, the present invention adopts a nonlinear robust tracking control method for a multi-UAV system based on azimuth measurement information, and considers the characteristics of the UAV system being susceptible to external disturbances. The Leader-Follower formation method is adopted, and the position drift of the UAV formation system based on azimuth information is corrected by introducing the Leader position information. The backstepping method design is combined with the adaptive design and the robust control design to compensate for the influence of unknown parameters and unknown external disturbances on the multi-UAV formation system. The present invention is mainly applied to design and manufacturing occasions.

Description

Technical Field

[0001] The present invention belongs to the control and constraint of quad-rotor UAV formation. Aiming at the problems that the sensor requirements for obtaining relative distance information between UAVs in GPS-denied environment are too high and the communication network in traditional formation control system is overloaded, a leader-follower type multi-UAV system formation tracking control design based on position information is proposed. Background Art

[0002] In recent years, drones have been widely used in military and civil activities due to their low cost, wide application scenarios, and strong mobility. However, as the difficulty of the tasks performed by drones continues to increase, single drone systems are no longer competent for increasingly complex flight missions, so researchers at home and abroad have begun to study the coordinated control of multi-drone systems. Compared with single drone systems, multi-drone systems have significant advantages in areas such as regional search and disaster relief. However, many factors such as communication delay, susceptibility to airflow disturbances, and model uncertainty in multi-drone systems have brought challenges to the control of multi-drone systems.

[0003] In order to solve the problems of large central node calculation, heavy communication burden, and easy communication delay in centralized formation control methods, distributed formation methods have become an important research direction of current multi-UAV formation systems.

[0004] Dong, Xiwang et al. (Journal: IEEE Transactions on Control Systems Technology; Author: Dong, Xiwang and Yu, Bocheng and Shi, Zongying and Zhong, Yisheng; Publication Date: 2014; Article Title: Time-varying formation control for unmanned aerial vehicles: Theories and applications; Pages: 340-348) applied the consistency method to the control problem of time-varying formation of unmanned aerial vehicles, carried out detailed mathematical derivation, and gave the necessary and sufficient conditions for multi-UAV system to achieve time-varying formation. Based on the consistency method, the paper designed a time-varying formation control strategy, carried out outdoor actual flight tests, and verified the effectiveness of the method.

[0005] However, most traditional formation control strategies are based on a default condition, that is, each aircraft (or some aircraft in the system) can obtain the distance or position information between each other. In an environment without GPS, this condition requires the UAV system to have strong perception capabilities, which greatly increases the design difficulty and cost of the multi-UAV formation system.

[0006] In order to cope with this limitation, formation control based on position information has attracted the attention of researchers in recent years. Formation control based on position information only requires the aircraft to obtain the position information of itself and its neighbors, without the need for inter-aircraft communication or distance information between aircraft. This control strategy has its own specific advantages for UAV formation control. First, in actual situations, each aircraft can easily obtain the required position information through passive sensors (such as airborne cameras), which greatly reduces the requirements for the perception ability of the aircraft. Second, there is no need for communication between the aircraft, which can effectively avoid various problems caused by communication in the multi-aircraft formation problem.

[0007] Zhao Shiyu et al. (Journal: IEEE Transactions on Automatic Control; Author: Zhao, Shiyu and Zelazo, Daniel; Publication Date: 2015; Article Title: Bearing rigidity and almost global bearing-only formation stabilization; Pages: 1255--1268) extended the theory of orientation rigidity to arbitrary dimensions and proposed the conditions under which the formation of multi-agent formations can be described by orientation constraints, which greatly promoted the development of orientation information formation control. Li, Xiaolei et al. (Journal: IEEE transactions on cybernetics; Author: Li, Xiaolei and Wen, Changyun and Chen, Ci; Publication Date: 2020; Article Title: Adaptive formation control of networked robotic systems with The quasi-Euler-Lagrangian form is used to generalize the theory of bearing rigidity to the solution of nonlinear robot system problems, and the single integral model subject to nonholonomic and dynamic constraints is transformed into a quasi-Euler-Lagrangian model. The formation and maintenance of bearing information formation of a three-dimensional nonholonomic constraint model is realized. Numerical simulation experiments verify the effectiveness of this method.

[0008] Schiano, Fabrizio et al. (Conference: 2017 IEEE International Conference on Robotics and Automation (ICRA); Author: Schiano, Fabrizio and Giordano, Paolo Robuffo; Publication Year: 2017; Article Title: Bearing Rigidity Maintenance for Formations of Quadrotor UAVs; Pages: 1467--1474) considered the sensor constraint problem in the azimuth information formation, including the field of view limitation of the camera, the possible field of view occlusion during flight, and other situations that actually exist in actual flight. The paper mathematically describes the above sensor constraint problem and conducts actual flight tests. The experimental results show that this design can complete the formation maintenance task based on azimuth information.

[0009] However, the above design does not consider the impact of external disturbances on formation control, and cannot complete the formation movement task. For the formation movement problem that relies only on azimuth information, Li, Sulong et al. (Journal: Frontiers in Control Engineering; Author: Li, Sulong and Wang, Qin and Wang, Enci and Chen, Yangyang; Publication Year: 2021; Article Title: Bearing-Only Adaptive Formation Control Using Back-Stepping Method; Page Number: 700053) studied a type of azimuth information formation control problem with a second-order model with unknown disturbances. In this paper, adaptive control design is introduced through the backstepping method to offset the impact of interference on the system, and the movement of the azimuth information formation is realized by embedding the speed-level expected variables into the deviation system. However, for actual systems, the complete inability to obtain position information will cause position drift problems.

[0010] To address this problem, Zhao, Shiyu et al. (Journal: Automatica; Author: Zhao, Shiyu and Zelazo, Daniel; Publication Date: 2016; Article Title: Localizability and distributed protocols for bearing-based network localization in arbitrary dimensions; Pages: 334-341) proposed the concept of bearing Laplacian and gave the necessary and sufficient conditions for constructing unique formations using anchor point and bearing information constraints. On this basis, Zhao, Shiyu et al. (Journal: IEEE Transactions on Automatic Control; Author: Zhao, Shiyu and Li, Zhenhong and Ding, Zhengtao; Publication Date: 2019; Article Title: Bearing-only formation tracking control of multiagent systems; Pages: 4541--4554) adopts the Leader-Follower formation mode, introduces the position information of the Leader to set the anchor point for the entire formation system, and designs a target formation tracking controller based on the azimuth information for the first-order model and the second-order model. The above design proves the convergence of the tracking error and the stability of the formation system through the Lyapunov method, and verifies the effectiveness of the method through numerical simulation. Summary of the invention

[0011] In order to overcome the shortcomings of the prior art, the present invention aims to study the anti-disturbance tracking control design of UAV formation based on azimuth information, and proposes a new nonlinear controller to complete the formation and tracking tasks of a multi-UAV system in view of the situation that each UAV can only obtain the azimuth information of its neighboring aircraft. To this end, the technical solution adopted by the present invention is a nonlinear robust tracking control method for a multi-UAV system based on azimuth measurement information. Taking into account the characteristics of the UAV system that is susceptible to external disturbances, a Leader-Follower formation method is adopted. The position drift of the UAV formation system based on azimuth information is corrected by introducing the Leader position information. The backstepping design is combined with the adaptive design and the robust control design to compensate for the influence of unknown parameters and unknown external disturbances on the multi-UAV formation system.

[0012] The stability of the system is proved based on the Lyapunov analysis method.

[0013] The specific steps are: firstly, analyze the UAV dynamics model, a formation system consisting of n UAVs, represents n×m-dimensional Euclidean space, where the dynamic model of the i-th UAV can be written as follows:

[0014]

[0015] in and Represent the position and linear velocity of the UAV, m i is the mass of the drone, and g is the gravitational acceleration constant. In formula (1): is the attitude rotation matrix from the drone body coordinate system to the inertial coordinate system at time t, f i (t) is the total lift generated by the UAV rotor, e 3 =[0 0 1] T , is the drag coefficient of the drone. The external disturbance to the drone. Define the auxiliary control input signal for

[0016]

[0017] The system dynamics model can be written as

[0018]

[0019] Secondly, we use graph theory to describe the UAV formation system: the connection relationship between the UAVs in the UAV formation can be represented by an undirected graph G = (V, E), where the node set V = {1, 2,, n} represents the set of individual UAVs, and the boundary set Represents the connection relationship between drones, and (i, j)∈E indicates that drone i can measure the position information g of drone j relative to itself ij , and UAV j is its neighbor, that is, N i ={j∈V:(i,j)∈E}, according to the connection characteristics of the undirected graph, The position information g ij Defined as:

[0020]

[0021] In the above formula, || e ij || represents the Euclidean distance between two bodies; is a unit vector, actually measured by a camera or wireless array;

[0022] Defining the orthographic projection matrix as follows:

[0023]

[0024] in is the unit matrix;

[0025] Sequentially put the first n l The nodes are set as the nodes corresponding to the Leader drone, and the remaining n f =nn l Set the nodes as Follower drone nodes, and record the sequence number of the first Follower node as n ff , that is, n ff =n l +1, and the first n l Nodes belong to the Leader set After n f Nodes belong to the Follower set

[0026] In order to ensure the uniqueness of the azimuth formation, the following azimuth Laplace array is introduced:

[0027]

[0028] Arrange the machines in the order of Leader to Follower. Divided into the following forms:

[0029]

[0030] when When it is a non-singular matrix, the azimuth formation p * (t) can be determined by the expected direction With Leader position The only decision is to design the controller and define the potential field function between UAV i and its neighbors as:

[0031]

[0032] Along it ij Taking partial derivative in the direction, we get:

[0033]

[0034] The total potential field function of UAV i is defined as:

[0035]

[0036] Introducing variables Then the system model (3) is transformed into:

[0037]

[0038] Design a virtual speed input v for each Follower drone si (t) are as follows:

[0039]

[0040] Define auxiliary error variables Then the system model of (11) can be rewritten as:

[0041]

[0042] where v i =diag([v ix (t),v iy (t),v iz (t)]) where v ix (t),v iy (t),v iz (t) is the speed of the drone in the x, y and z directions, D i =[D ix ,D iy ,D iz ] T , where D ix ,D iy ,D iz is the drag coefficient of the UAV in the x, y, and z directions;

[0043] For the above system (13), the control input of the i-th UAV is for:

[0044]

[0045] in sgn(*) is a symbolic function, and its specific form is as follows:

[0046]

[0047] In the formula D is the wind resistance parameter i The estimated value of It is given in formula (16). In summary, the formation control rate is proposed:

[0048]

[0049] in is a positive parameter, for Update law, v * is the expected speed of the formation as a whole;

[0050] Under the control law (16) mentioned above, the azimuth angles between the UAVs in the multi-UAV system can converge to the expected value, and the speed of each UAV can converge to the expected speed, thus completing the formation, maintenance and tracking tasks of the target formation.

[0051] The specific steps of verification are as follows: combining the Lyapunov analysis method, it can be proved that the azimuth deviation and velocity deviation converge globally asymptotically, that is,

[0052] The above-mentioned UAV formation control method based on azimuth information is experimentally verified, which is characterized in that the UAVs are controlled only by azimuth information to complete the formation flight function, so as to verify the rationality and effectiveness of the designed control strategy.

[0053] The characteristics and beneficial effects of the present invention are:

[0054] 1. At present, in most formation strategies based on azimuth information, relatively simple first-order and second-order models or the formation maintenance problem of static target formations are mostly considered. Based on the position dynamics model of UAVs, the present invention considers the unknown disturbances and wind resistance in actual flight, designs a formation controller based on azimuth information, and realizes the tracking of mobile target formations.

[0055] 2. The formation is constructed based on the Leader-Follower method to solve the position drift phenomenon in the formation problem based on azimuth information. The azimuth information formation controller is designed by backstepping method. By introducing virtual speed input, the adaptive control method and robust control method are applied to the formation control problem based on azimuth information, and the stability of the formation system is analyzed by Lyapunov method.

[0056] 3. The results of literature search show that most of the current research on azimuth information formation is still at the simulation stage; there are very few research results on UAV formation control design based on azimuth information and experimental verification. The present invention independently built a multi-UAV flight experimental platform and conducted physical experiments to further verify the practicality and effectiveness of the proposed method. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Schematic diagram of the four-rotor formation experimental platform used in the present invention;

[0058] Figure 2 It is a schematic diagram of the expected formation and topological structure designed for the verification experiment of the present invention;

[0059] Figure 3 It is the three-dimensional position map of the four drones in Experiment 1;

[0060] Figure 4is the deviation between the azimuth information in Experiment 1 and the expected azimuth Graphs;

[0061] Figure 5 is the total deviation of the position of each Follower UAV in Experiment 1 △ i (t)

[0062] Figure 6 It is a graph of the position deviation δp3(t) of UAV No. 3 in Experiment 1;

[0063] Figure 7 is the position deviation δp of UAV No. 4 in Experiment 1 4 (t)

[0064] Figure 8 is the total deviation of the position of each Follower UAV in Experiment 2 △ i (t)

[0065] Fig. 9 It is a graph of the position deviation δp3(t) of UAV No. 3 in Experiment 2;

[0066] Fig.10 is the position deviation δp of UAV No. 4 in Experiment 2 4 (t) curve. DETAILED DESCRIPTION

[0067] In order to solve the problems of heavy burden on communication network and high requirements on sensors for obtaining relative positions in traditional formation control methods, the present invention studies the formation problem of multi-UAV system based on azimuth information. The present invention mainly designs a Leader-Follower type azimuth information formation control strategy through backstepping method, solves the position drift phenomenon existing in the formation problem based on azimuth information, and realizes the formation, maintenance and tracking tasks of target formation based on azimuth information of multi-UAV system under the condition of considering disturbance and wind resistance.

[0068] The further specific steps are: firstly, analyze the UAV dynamics model, a formation system consisting of n UAVs, Represents n×m-dimensional Euclidean space, then the dynamic model of the i-th UAV can be written as follows:

[0069]

[0070] in and Represent the position and linear velocity of the UAV, m i is the mass of the drone, and g is the gravitational acceleration constant. In formula (1): is the attitude rotation matrix from the drone body coordinate system to the inertial coordinate system at time t, f i (t) is the total lift generated by the UAV rotor, e 3 =[0 0 1] T , is the drag coefficient of the drone. The external disturbance to the drone. Define the auxiliary control input signal for

[0071]

[0072] The system dynamics model can be written as

[0073]

[0074] Secondly, we use graph theory to describe the UAV formation system. The connection relationship between each UAV in the UAV formation can be represented by an undirected graph G = (V, E). The node set V = {1, 2,, n} represents the set of individual UAVs, and the boundary set Represents the connection relationship between each drone. In the present invention, (i, j)∈E represents that drone i can measure the position information g of drone j relative to itself. ij , and UAV j is its neighbor, that is, N i ={j∈V:(i,j)∈E}. According to the connection characteristics of undirected graphs, The position information g ij Defined as:

[0075]

[0076] In the above formula, ||e ij || represents the Euclidean distance between two bodies. is a unit vector that can be actually measured by a camera or wireless array;

[0077] Defining the orthographic projection matrix as follows:

[0078]

[0079] in is a unit matrix.

[0080] Sequentially put the first n l The nodes are set as the nodes corresponding to the Leader drone, and the remaining n f =nn l Set the nodes as Follower drone nodes, and record the sequence number of the first Follower node as n ff , that is, n ff=n l +1, and the first n l Nodes belong to the Leader set After n f Nodes belong to the Follower set

[0081] In order to ensure the uniqueness of the azimuth formation, the selection of the leader and azimuth constraints needs to meet certain conditions. The following azimuth Laplace matrix is ​​introduced:

[0082]

[0083] Arrange the machines in the order of Leader to Follower. Divided into the following forms:

[0084]

[0085] when When it is a non-singular matrix, the azimuth formation p * (t) can be determined by the expected direction With Leader position The only decision;

[0086] Then the controller is designed. The potential field function between UAV i and its neighbors is defined as

[0087]

[0088] Along it ij Taking the partial derivative in the direction, we can get

[0089]

[0090] The total potential field function of UAV i is defined as

[0091]

[0092] Introducing variables Then the system model (3) is transformed into

[0093]

[0094] Design a virtual speed input v for each Follower drone si (t) are as follows:

[0095]

[0096] Define auxiliary error variables Then the system model of (11) can be rewritten as

[0097]

[0098] where v i =diag([v ix (t),v iy (t),v iz (t)]) where v ix (t),v iy (t),v iz (t) is the speed of the drone in the x, y and z directions, D i =[D ix ,D iy ,D iz ] T , where D ix ,D iy ,D iz is the drag coefficient of the UAV in the x, y, and z directions.

[0099] For the above system (13), the control input of the i-th UAV is for

[0100]

[0101] in sgn(*) is a symbolic function, and its specific form is as follows

[0102]

[0103] In the formula D is the wind resistance parameter i The estimated value of It is given in formula (16). In summary, the formation control rate is proposed:

[0104]

[0105] in is a positive parameter, for Update law, v * is the expected speed of the formation as a whole. Under the control law (16), the azimuth angles between the UAVs in the multi-UAV system can converge to the expected value, and the speed of each UAV can converge to the expected speed, completing the formation, maintenance and tracking tasks of the target formation. The specific steps of verification are: combining the Lyapunov analysis method, it can be proved that the azimuth angle deviation and speed deviation converge globally asymptotically, that is,

[0106] The above-mentioned UAV formation control method based on azimuth information is experimentally verified, which is characterized in that the UAVs are controlled only by azimuth information to complete the formation flight function, so as to verify the rationality and effectiveness of the designed control strategy.

[0107] The present invention is described in detail below with reference to specific examples and accompanying drawings.

[0108] 1. Introduction to the experimental platform

[0109] In order to verify the effectiveness and practicality of the azimuth formation control method designed by the present invention, the present invention uses a multi-UAV formation flight platform to conduct actual flight experiments. In the physical experiment, each UAV obtains the required azimuth information through the position information of itself and its neighbors by formula (4), and the formation movement speed can be given in advance.

[0110] like Figure 1 As shown in the figure, the formation flight platform consists of four quad-rotor drones and a ground station. The mass of each drone is m 1 =0.923kg,m 2 =0.926kg,m 3 =0.905kg,m 4 =0.935kg, and the wheelbase is l=0.25m. Each drone is equipped with an ARM embedded computing board and a flight controller. The drones are networked via WIFI, and can obtain their own and other drones' position information through the OptiTrack motion capture system. The ARM embedded computing board and the flight controller are connected via a serial port. The ARM embedded computing board runs the azimuth formation control method at a frequency of 50hz, and sends the calculated control instructions to the flight controller in the form of MavLink information. The flight controller is responsible for the underlying attitude control of the drone. After receiving the control input instruction, it calculates and converts it into the speed of each motor, and finally realizes the control input instruction from the ARM embedded computing board. The ground station is used to remotely log in to each airborne ARM embedded computing board to start the formation control method and detect various data in real time. The flight field is 5.5m long, 3m wide and 2.5m high.

[0111] 2. Experimental Verification and Analysis

[0112] The present invention mainly conducted two groups of flight experiments. Experiment 1 was a flight experiment of forming, maintaining and tracking a target formation by a multi-UAV system under the action of the controller designed by the present invention. Experiment 2 was a comparative experiment, in which a PD controller based on azimuth information was used to control the multi-UAV system, and the same flight experiment as Experiment 1 was conducted.

[0113] First, select the desired formation and design appropriate orientation constraints and leader selection through Lemma 1. Due to the limitation of the size of the experimental site, each UAV always maintains the same desired altitude during the flight. The plane diagram of the desired formation is as follows Figure 2 As shown. Among them, machines 1 and 2 are leaders, and the constraint between the two leaders || e 12 ||=1.5,g 12 =[1,0]. Other expected directions are

[0114] Empirically, the Laplace matrix of the desired formation designed above is Non-singular, that is, the above expected formation design satisfies the uniqueness condition of the azimuth formation, and the expected formation can be uniquely determined by the above design.

[0115] The expected heights of the four drones are all set to 1.5m, and the initial expected positions are set to p 1 =[-0.75,2.5,1.5] T m,p 2 =[0.75,2.5,1.5] T m,p 3 =[-1,1.3,1.5] T m,p 4 =[1,1.3,1.5] T m. The initial position does not meet the expected formation. The experimental parameters are selected as follows: k 1 =[1.25,1.25] T ,k 2 =[1.25,1] T ,

[0116] When conducting the flight experiment, each UAV is first made to fly to the initial position set in the article. After reaching the initial expected position point, the UAV as the Follower runs the control method designed by the present invention to complete the formation and tracking tasks of the formation under the condition of using only the azimuth information; the two UAVs as the Leader are tracked and controlled by the position information, and wait for 5 seconds after reaching the initial expected position, and then the two Leader UAVs are made to perform v along the y direction under the condition of satisfying the above constraints. d = 0.1m / s reciprocating motion.

[0117] The positions of the four drones in the flight experiment are as follows: Figure 3 As shown in the figure, the dotted lines represent the trajectories of the four drones, and the solid lines represent the trajectories of the formation method. 0= The formation formed by the real-time positions of the four drones after 16s. At this time, the formation is already in a forward moving state. It can be seen from the figure that the formation system effectively completes the task of maintaining and tracking the expected moving formation.

[0118] During the flight, each orientation information is represented by a unit vector, and the difference between the actual orientation information and the expected orientation value is like Figure 4 As shown, the horizontal axis is the running time of the designed control method calculated from the time when each body reaches the expected starting point. The vertical axis is the δ of each participating control operation. ij value.

[0119] In order to more intuitively express the control effect of each orientation information, the total orientation deviation of each Follower is defined as Its trend over time is Figure 5 As shown. Figure 5 It can be seen that as the program runs, each direction gradually converges to its expected value, that is, the multi-UAV system successfully completes the formation and tracking tasks under the designed control law.

[0120] For the formation determined by the leader-orientation constraint, the same formation with different sizes (such as different spacing between leaders) has the same orientation deviation δ ij (t) The formation effect is different. In order to more intuitively show the formation flight effect, the present invention gives the deviation between each Follower position and the expected position. The curve diagram is as follows Figure 6 , 7 As shown. The expected position of each Follower body It can be calculated from the real-time position and expected orientation constraints of any two leaders. Figure 6 , 7 It can be seen that each Follower UAV gradually converges to the desired position under the action of the designed controller, and always fluctuates within a small range near the desired position. It can be intuitively seen that the controller designed by the present invention completes the formation task well, and solves the position drift problem by introducing the Leader UAV.

[0121] Comparative Experiment 2 adopts the same process and initial conditions as Experiment 1, and designs PD comparison controllers for each Follower as follows:

[0122]

[0123] Select parameter k p =[2.5,2] T ,k d =[2.5,2.5] TSame as experiment 1, the experimental data results are as follows Figure 8 , 9 , as shown in 10.

[0124] right Figure 5 , 6 , 7 and Figure 8 , 9 , 10, the steady-state process of the data is quantitatively compared and analyzed, the data from the 10th second to the 70th second are selected, and the maximum deviation and root mean square error of each quantity in the figure are calculated respectively. The calculation results are shown in Table 1 and Table 2. By comparing Table 1 and Table 2, it can be found that in Experiment 1, i With δp i The maximum deviation and root mean square error of are both 50% to 60% of those in Experiment 1, which indicates that under the action of the controller designed by the present invention, the formation formed by the multi-UAV system is closer to the expected formation and has smaller fluctuations.

[0125] Combined with the above charts, it can be seen that the controller designed by the present invention has better performance than the PD controller when the multi-UAV formation system completes the formation, maintenance and tracking tasks of the target formation based on the position information.

[0126] Table 1: Controller designed by the present invention

[0127]

[0128] Table 2: Comparison of PD controllers

[0129]

[0130] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

Claims

1. A nonlinear robust tracking control method for a multi-UAV system based on azimuth measurement information, characterized in that: The UAV formation system is described by graph theory, and the Leader-Follower formation method is adopted. The position drift of the UAV formation system based on the orientation information is corrected by introducing the Leader position information. The backstepping design is combined with the adaptive design and the robust control design to compensate for the influence of unknown parameters and unknown external disturbances on the multi-UAV formation system. The specific steps are as follows: first, the UAV dynamics model is analyzed. A formation system consisting of n UAVs is used. represents n×m-dimensional Euclidean space, where the dynamic model of the i-th UAV is written as follows: and Represent the position and linear velocity of the UAV, m i is the mass of the drone, g is the gravitational acceleration constant, and in formula (1) is the attitude rotation matrix from the drone body coordinate system to the inertial coordinate system at time t, f i is the total lift generated by the UAV rotor, e3 = [0 0 1] T , is the drag coefficient of the drone, Define auxiliary control input signals for external disturbances to the drone for The system dynamics model can be written as Secondly, we use graph theory to describe the UAV formation system: the connection relationship between the UAVs in the UAV formation is represented by an undirected graph G = (V, E), where the node set V = {1, 2, ..., n} represents the set of individual UAVs, and the boundary set Represents the connection relationship between drones, and (i, j)∈E represents the position information g of drone i relative to drone j. ij , and UAV j is its neighbor, that is, N i ={j∈V:(i,j)∈E}, according to the connection characteristics of the undirected graph, The position information g ij Defined as: In the above formula, ||e ij || represents the Euclidean distance between two bodies; is a unit vector, actually measured by a camera or wireless array; Defining the orthographic projection matrix as follows: in is the unit matrix; Sequentially put the first n l The nodes are set as the nodes corresponding to the Leader drone, and the remaining n f =nn l Set the nodes as Follower drone nodes, and record the sequence number of the first Follower node as n ff , that is, n ff =n l +1, and the first n l Nodes belong to the Leader set After n f Nodes belong to the Follower set In order to ensure the uniqueness of the azimuth formation, the following azimuth Laplace array is introduced: Arrange the machines in the order of Leader to Follower. Divided into the following forms: when When it is a non-singular matrix, the azimuth formation p*(t) is determined by the expected azimuth With Leader position The only decision is to design the controller and define the potential field function between UAV i and its neighbors as: Along it ij Taking partial derivative in the direction, we get: The total potential field function of UAV i is defined as: Introducing variables Then the system model (3) is transformed into: Design a virtual speed input v for each Follower drone si (t) are as follows: Define auxiliary error variables Then the system model of (11) can be rewritten as: where v i =diag([v ix (t),v iy (t),v iz (t)]), v ix (t),v iy (t),v iz (t) is the speed of the drone in the x, y and z directions, D i =[D ix ,D iy ,D iz ] T , where D ix ,D iy ,D iz is the drag coefficient of the UAV in the x, y, and z directions; For the above system (13), the control input of the i-th UAV is for: in, v* is the expected speed of the formation as a whole; sgn(*) is a sign function, and its specific form is as follows: In the formula D is the wind resistance parameter i The estimated value of Given in formula (16), the formation control rate is proposed: is a positive parameter, for The update law, under the action of the above control law (16), makes the azimuth angles between the UAVs in the multi-UAV system converge to the expected value, and the speed of each UAV converges to the expected speed, thus completing the formation, maintenance and tracking tasks of the target formation.

2. The nonlinear robust tracking control method for a multi-UAV system based on azimuth measurement information as claimed in claim 1 is characterized in that: The stability of the system is proved based on the Lyapunov analysis method.

3. The nonlinear robust tracking control method for a multi-UAV system based on azimuth measurement information as claimed in claim 1 is characterized in that: The specific steps of verification are to combine the Lyapunov analysis method to prove that the azimuth deviation and velocity deviation converge globally asymptotically, that is, The control method is experimentally verified by controlling the UAV using only the orientation information to enable it to complete the formation flight function, so as to verify the rationality and effectiveness of the designed control strategy.

Images