An Autonomous Navigation Method for Unmanned Aerial Vehicles Based on Attractive Manifolds

Abstract

This invention relates to an autonomous navigation method for unmanned aerial vehicles (UAVs) based on attractive manifolds, belonging to the field of UAV navigation and automatic control technology. It solves the problems of incomplete functionality, low control accuracy in unstable states, slow response speed, high implementation difficulty, large computational and information requirements, low scalability, low robustness, and low flexibility in existing technologies. This invention adjusts the operating mode of the control system based on the real-time detected relative distance between the UAV and obstacles, while considering path tracking, collision avoidance, and the UAV's dynamic characteristics. It can adapt to various complex environments and improve the safety and reliability of UAVs in practical applications. It features faster response speed, lower implementation difficulty, less computational load, and fewer information requirements; it also has better adaptability to changes in model parameters, exhibiting scalability, robustness, and flexibility, and a smoother motion trajectory.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) navigation and automatic control technology, and specifically to an autonomous navigation method for UAVs based on an attractive manifold. Background Technology

[0002] With their real-time and efficient data acquisition capabilities and flexible maneuverability, drones are widely used in environmental monitoring, search and rescue, and reconnaissance. However, due to the diverse mission requirements and complex working environments, drones face threats from the environment and obstacles during flight.

[0003] Researchers have proposed local navigation methods for real-time perception and decision-making. These methods utilize the UAV's own sensors to acquire real-time environmental information and perform path planning based on this information. The aim is to overcome the challenge of comprehensively acquiring environmental information and improve the UAV's adaptability to dynamic obstacles or environmental changes. Generally, local navigation methods can be categorized into artificial potential field methods, optimization methods, geometric methods, and artificial intelligence methods. Some typical examples include the steering vector field method, which uses a composite steering vector field to enable the aircraft to move along a smooth, desired path of arbitrary shape, even under static or moving obstacle occlusion. Another example is the optimal mutual collision avoidance method, which transforms other UAVs or obstacles in the environment into velocity obstacles to determine the permissible speed range of the controlled UAV. Then, the optimal speed and velocity direction of the controlled object are determined through optimization criteria, achieving reactive local navigation. This method is based on an algorithmic framework and uses simple geometric methods to ensure that the UAV can autonomously resolve conflicts. This method has shown good application results in practice. However, many of these studies use simplified UAV models and do not consider the UAV's dynamic performance constraints and the closed-loop stability of the system. This may lead to inappropriate control strategies or parameter selection, resulting in instability of the UAV control system and an inability to achieve precise flight control. Therefore, during UAV navigation, it is necessary to consider the kinematics, dynamics, and performance constraints of the navigation controller, as well as the impact of inertial factors introduced by the actuators on the actual UAV system.

[0004] One approach to solving UAV navigation tasks is to treat it as a position control problem with state constraints caused by obstacles. To address this, researchers have proposed a chance-constrained linear optimization method based on Gaussian and Gaussian mixture models. This method models relative position and velocity errors and effectively handles control inputs and system state constraints through model predictive control. Furthermore, researchers have combined obstacle function control and information gap theory to propose a robust and safe controller for collision-free trajectory planning between multiple UAVs or between UAVs and obstacles. However, most current optimization methods based on model predictive control or obstacle function control exhibit low scalability when faced with variations in UAV parameters or number, thus limiting their applicability. Therefore, it is necessary to further improve these methods to enhance their scalability and to explore other solutions to address the challenges posed by variations in UAV parameters or models.

[0005] In summary, existing technologies suffer from problems such as incomplete functionality, low flight control accuracy under unstable conditions, slow response speed, high implementation difficulty, large computational and information requirements, low scalability, low robustness, and low flexibility. Summary of the Invention

[0006] In view of the above problems, the present invention provides an autonomous navigation method for unmanned aerial vehicles based on an attractive manifold, which solves the problems of incomplete functions, low flight control accuracy under unstable conditions, slow response speed, high implementation difficulty, large amount of computation and information requirements, low scalability, low robustness and low flexibility in the prior art.

[0007] This invention provides an autonomous navigation method for unmanned aerial vehicles (UAVs) based on an attractive manifold, comprising the following steps:

[0008] Step S1. Set the desired trajectory of the UAV, obtain the desired trajectory parameters, and construct the attractive manifold of the UAV by combining the kinematic and dynamic performance constraints of the UAV.

[0009] Step S2. Obtain the distance information between the drone and the obstacle, determine the collision risk between the drone and the obstacle, and obtain the desired working mode of the drone control system based on the collision risk;

[0010] Step S3. Based on the attracting manifold and the desired working mode of the UAV control system, establish a closed-loop control system for autonomous navigation of the UAV and set the parameters of the UAV control system.

[0011] Step S4. Construct a mathematical model of the UAV, substitute it into the closed-loop control system of the UAV autonomous navigation to obtain the control input of the UAV, which is used to control the UAV to avoid obstacles and track the desired trajectory to complete the autonomous navigation task.

[0012] Furthermore, step S1 specifically includes:

[0013] Step S1-1. Set the desired trajectory of the UAV and obtain the desired trajectory parameters; wherein, the desired trajectory of the UAV consists of a target point, a straight line, and a quadratic curve that can be represented by a system of equations;

[0014] Step S1-2. Obtain the kinematic performance constraints of the UAV, including maximum speed, maximum acceleration, maximum pitch angle, and maximum roll angle;

[0015] Steps S1-3. Construct the attracting manifold of the UAV based on its desired trajectory parameters and kinematic performance constraints, including the positional manifold ψ. tr and velocity manifold ψ sp .

[0016] Further, in steps S1-3, the position manifold ψ tr and velocity manifold ψ sp They are represented as follows:

[0017]

[0018] ;

[0019] Where P is the position of the UAV, Φ(P,ψ,t) is the constraint function of the UAV's heading angle ψ, V is the actual velocity of the UAV, and A1, A2, A3, J... s and J t It is a coefficient matrix related to the desired trajectory parameters, where t is time; It is the difference between the actual speed of the drone and its expected speed.

[0020] Furthermore, step S2 specifically includes:

[0021] Step S2-1. Obtain the location information of the UAV, and then use the UAV's visual sensor to perform distance measurement on the obstacle to obtain the distance r between the UAV and the obstacle. c , will r c Compare with the minimum safe distance r and obtain the comparison result;

[0022] Step S2-2. Based on the comparison results of step S2-1, determine the collision risk between the drone and the obstacle, and obtain the desired working mode of the drone control system;

[0023] Step S2-2A. If the comparison result of step S2-1 is r c >r indicates that the distance between the drone and the obstacle is far and the risk of collision between the drone and the obstacle is low. Therefore, the desired working mode of the drone control system is the asymptotically stable mode, which is used to control the drone to track the desired trajectory.

[0024] Step S2-2B. If the comparison result in Step S2-1 is r c < r, and r c (k) at time k is less than r c (k-1) at time k-1, it indicates that the UAV is approaching an obstacle and the risk of collision between the UAV and the obstacle is high. Then the expected working mode of the UAV control system is an unstable mode, which is used to control the movement trajectory of the UAV to diverge away from the obstacle near the obstacle, so as to avoid collision with the obstacle;

[0025] Step S2-2C. If the comparison result in Step S2-1 is r c < r, and r c (k) at time k is greater than r c (k-1) at time k-1, it indicates that the UAV is moving away from the obstacle and the risk of collision between the UAV and the obstacle is low. Then the expected working mode of the UAV control system is an asymptotically stable mode, and the UAV is controlled to track the desired trajectory.

[0026] Furthermore, in Step S2-1, the position information of the UAV is obtained through the UAV's autonomous positioning system; the position information of the obstacle is obtained through the UAV's environmental perception system.

[0027] Furthermore, Step S3 specifically includes:

[0028] Step S3-1. Construct a closed-loop control system according to the working mode of the UAV control system;

[0029] If the expected working mode of the UAV control system is an asymptotically stable mode, then construct a closed-loop control system in the following form:

[0030]

[0031] ]] ;

[0032] where T1, T2 and T3 are parameter matrices in the asymptotically stable mode respectively;

[0033] Set T1, T2 and T3 as positive definite matrices, the closed-loop control system has asymptotic stability, and set the working mode of the UAV control system as an asymptotically stable mode;

[0034] If the working mode of the UAV control system is an unstable mode, then construct a closed-loop control system in the following form:

[0035] ψ = ψ tr + Aψ sp ;

[0036]

[0037] Where A and T are the coefficient matrices in the unstable mode; by setting both A and T to negative definite matrices, the closed-loop control system is unstable, and the working mode of the UAV control system is set to the unstable mode.

[0038] When the closed-loop control system exhibits instability, parameter β is used to set the parameter matrices A and T in the unstable mode, letting T = A = diag(S). i If i = 1, ..., v, then we have:

[0039]

[0040]

[0041] Where j is the number of obstacles in the UAV's reaction area, and s0 is a constant greater than zero.

[0042] Furthermore, step S4 specifically includes:

[0043] Step S4-1. Establish the mathematical model of the UAV, represented in matrix form as follows:

[0044]

[0045]

[0046]

[0047]

[0048] The position of the drone in space is P = [p x p y p z ] T The attitude is Θ=[φ θ ψ] T Where φ, θ, and ψ are the roll angle, pitch angle, and yaw angle, respectively; g e F is the projection of gravitational acceleration in the Earth coordinate system. b It is the projection of the total propeller thrust onto the airframe coordinate system, τ represents the torque acting on the airframe axis, J is the moment of inertia of the UAV, and G... a This refers to the gyroscopic torque.

[0049] Step S4-2. Substitute the mathematical model of the UAV into the closed-loop control system constructed in S3-1, solve for the control input of the UAV, and thus realize the motion control of the UAV.

[0050] Step S4-3. Repeat steps S2-1 to S4-2 until the UAV safely and without collision completes the autonomous navigation task.

[0051] Compared with the prior art, the present invention has at least the following beneficial effects:

[0052] (1) The UAV autonomous navigation method based on attracting manifold of the present invention adjusts the working mode of the control system by the UAV according to the real-time detected relative distance between itself and obstacles. When the distance between the UAV and the obstacle is greater than the minimum safe distance, the control system will enter an asymptotically stable mode, at which time the UAV's trajectory will gradually converge to the desired path. However, when the distance between the UAV and the obstacle is less than the safe distance, the control system will switch to an unstable mode, at which time the UAV's trajectory will diverge outward from the obstacle to avoid collision. Once the collision risk is eliminated, the control system will switch back to the asymptotically stable mode. This method has more comprehensive functions and can simultaneously consider path tracking tasks, collision avoidance tasks, and the dynamic characteristics of the UAV, adapting to various complex environments and improving the safety and reliability of UAVs in practical applications.

[0053] (2) The UAV autonomous navigation method based on attracting manifold of the present invention addresses the path tracking and collision avoidance control problems in UAV autonomous navigation from the perspective of control system design. Compared with the optimization-based model predictive control method and the geometric feature-based guiding vector field method, the method proposed in this invention has a faster response speed, lower implementation difficulty, less computation and less information requirement.

[0054] (3) The UAV autonomous navigation method based on attracting manifold of the present invention integrates the planning and control problems in UAV navigation, thereby enabling the method to have better adaptability to changes in model parameters, that is, it has scalability, robustness and flexibility; this integrated design not only improves the overall performance of the system, but also makes the UAV motion trajectory obtained by the method smoother. Attached Figure Description

[0055] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0056] Figure 1 This is a flowchart of the UAV autonomous navigation method based on attracting manifold proposed in this invention;

[0057] Figure 2 This is a structural diagram and coordinate definition of a quadcopter unmanned aerial vehicle according to an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of stable and unstable states according to an embodiment of the present invention;

[0059] Figure 4This is a schematic diagram of the desired path according to an embodiment of the present invention;

[0060] Figure 5 This is a diagram showing the path tracking results of a UAV under complex conditions using an UAV autonomous navigation method based on an attractive manifold according to an embodiment of the present invention.

[0061] Figure 6 The diagram shows the path tracking results of a UAV in a complex environment, simplified to an unstable point form, based on the UAV autonomous navigation method based on an attractive manifold according to an embodiment of the present invention.

[0062] Figure 7 This is a diagram showing the path tracking results of a UAV using the artificial potential field method according to an embodiment of the present invention;

[0063] Figure 8 This is a diagram showing the path tracking results of a UAV using the vector field method according to an embodiment of the present invention;

[0064] Figure 9 This is a diagram showing the path tracking results of a UAV using a model predictive control method according to an embodiment of the present invention. Detailed Implementation

[0065] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0066] This invention discloses an autonomous navigation method for unmanned aerial vehicles (UAVs) based on attracting manifolds, aiming to enhance the motion planning and obstacle avoidance capabilities of UAVs in complex environments. This method comprehensively addresses multiple issues in UAV autonomous navigation, including path tracking, collision avoidance, and dynamic constraints, integrating them into a unified control system design problem. In this method, the UAV adjusts the operating mode of the control system based on the real-time detected relative distance between itself and obstacles. When the distance between the UAV and the obstacle is greater than the minimum safe distance, the control system enters an asymptotically stable mode, where the UAV's trajectory gradually converges to the desired path. However, when the distance between the UAV and the obstacle is less than the safe distance, the control system switches to an unstable mode, where the UAV's trajectory diverges outwards from the obstacle to avoid collisions. Once the collision risk is eliminated, the control system switches back to the asymptotically stable mode. In summary, this method provides a more comprehensive solution for UAV autonomous navigation, adaptable to various complex environments, and improves the safety and reliability of UAVs in practical applications.

[0067] The UAV autonomous navigation method based on attraction manifolds includes the following steps:

[0068] Step S1. Set the desired trajectory of the UAV, obtain the desired trajectory parameters, and construct the attractive manifold of the UAV by combining the kinematic and dynamic performance constraints of the UAV.

[0069] Step S1 specifically includes:

[0070] Step S1-1. Set the desired trajectory of the UAV and obtain the desired trajectory parameters; wherein, the desired trajectory of the UAV consists of a target point, a straight line and a quadratic curve that can be represented by a system of equations.

[0071] Step S1-2. Obtain the kinematic performance constraints of the UAV, including maximum speed, maximum acceleration, maximum pitch angle, and maximum roll angle.

[0072] Steps S1-3. Construct the attracting manifold of the UAV based on its desired trajectory parameters and kinematic performance constraints, including the positional manifold ψ. tr and velocity manifold ψ sp .

[0073] Position manifold ψ tr and velocity manifold ψ sp They are represented as follows:

[0074]

[0075] ;

[0076] Where P is the position of the UAV; Φ(P,ψ,t) is the constraint function for the UAV's heading angle ψ, which is determined by the UAV's mission requirements. The UAV has four degrees of freedom, but only three are needed for position control. The remaining degree of freedom (heading angle) can be a fixed value or controlled according to actual mission requirements; V is the actual speed of the UAV; A1, A2, A3, J s and J t It is a coefficient matrix related to the desired trajectory parameters; t is time; It is the difference between the actual speed of the drone and its expected speed.

[0077] Step S2. Obtain the distance information between the drone and the obstacle, determine the collision risk between the drone and the obstacle, and obtain the desired working mode of the drone control system based on the collision risk.

[0078] Step S2 specifically includes:

[0079] Step S2-1. Obtain the position information of the drone, and then use the drone's vision sensor to measure the distance to the obstacle, obtaining the distance r between the drone and the obstacle. c , and compare r c with the minimum safe distance r to obtain the comparison result.

[0080] The position information of the drone is obtained through the drone's autonomous positioning system, optionally GPS or an inertial navigation system.

[0081] Step S2-2. Judge the collision risk between the drone and the obstacle according to the comparison result of Step S2-1, obtaining the desired working mode of the drone control system.

[0082] Step S2-2A. If the comparison result of Step S2-1 is r c > r, indicating that the distance between the drone and the obstacle is far and the collision risk between the drone and the obstacle is low, then the desired working mode of the drone control system is the asymptotically stable mode, which is used to control the drone to track the desired trajectory.

[0083] Step S2-2B. If the comparison result of Step S2-1 is r c < r, and r c (k) at time k is less than r c (k - 1) at time k - 1, then the drone is approaching the obstacle and the collision risk between the drone and the obstacle is high. Then the desired working mode of the drone control system is the unstable mode, which is used to control the movement trajectory of the drone to diverge away from the obstacle near the obstacle, thus avoiding collision with the obstacle.

[0084] Step S2-2C. If the comparison result in Step S2-1 is r c < r, and r c (k) at time k is greater than r c (k - 1) at time k - 1, indicating that the drone is moving away from the obstacle and the collision risk between the drone and the obstacle is low. Then the desired working mode of the drone control system is the asymptotically stable mode, controlling the drone to track the desired trajectory.

[0085] Step S3. According to the attracting manifold and the desired working mode of the drone control system, establish a closed-loop control system for the drone's autonomous navigation and set the parameters of the drone control system.

[0086] Step S3 specifically includes:

[0087] Step S3-1. Construct a closed-loop control system according to the working mode of the drone control system.

[0088] If the desired operating mode of the UAV control system is asymptotically stable, then a closed-loop control system of the following form can be constructed:

[0089]

[0090] ;

[0091] Where T1, T2 and T3 are the parameter matrices in the asymptotically stable mode;

[0092] By setting T1, T2, and T3 as positive definite matrices, the closed-loop control system exhibits asymptotic stability, and the operating mode of the UAV control system is set to asymptotically stable mode.

[0093] If the drone control system operates in an unstable mode, then a closed-loop control system of the following form should be constructed:

[0094] ψ=ψ tr +Aψ sp ;

[0095]

[0096] Where A and T are the coefficient matrices in the unstable mode; by setting both A and T to negative definite matrices, the closed-loop control system is unstable, and the working mode of the UAV control system is set to the unstable mode.

[0097] When the closed-loop control system exhibits instability, parameter β is used to set the parameter matrices A and T in the unstable mode, letting T = A = diag(S). i If i = 1, ..., v, then we have:

[0098]

[0099]

[0100] Where j is the number of obstacles in the UAV's reaction area, and s0 is a constant greater than zero.

[0101] Step S4. Construct a mathematical model of the UAV based on its structure and parameters, substitute it into the closed-loop control system of the UAV's autonomous navigation to obtain the control input of the UAV, which is used to control the UAV to avoid obstacles and track the desired trajectory, thus completing the autonomous navigation task.

[0102] Step S4 specifically includes:

[0103] Step S4-1. Establish the mathematical model of the UAV, represented in matrix form as follows:

[0104]

[0105]

[0106]

[0107]

[0108] The position of the drone in space is P = [p x p y p z ] T The attitude is Θ=[φ θ ψ] T Where φ, θ, and ψ are the roll angle, pitch angle, and yaw angle, respectively; g e F is the projection of gravitational acceleration in the Earth coordinate system. b It is the projection of the total propeller thrust onto the airframe coordinate system, τ represents the torque acting on the airframe axis, J is the moment of inertia of the UAV, and G... a This is the gyroscopic torque.

[0109] Step S4-2. Substitute the mathematical model of the UAV into the closed-loop control system constructed in S3-1, solve for the control input of the UAV, and thus realize the motion control of the UAV.

[0110] Step S4-3. Repeat steps S2-1 to S4-2 until the UAV safely and without collision completes the autonomous navigation task.

[0111] To illustrate the effectiveness of the method proposed in this invention, the following detailed description of the above technical solution is provided through a specific embodiment:

[0112] Example 1

[0113] The embodiments of this invention utilize a quadcopter UAV to verify path tracking and collision avoidance control, thereby demonstrating the feasibility and effectiveness of the proposed UAV autonomous navigation method based on attraction manifold in complex environments.

[0114] The flowchart of the UAV autonomous navigation method based on attracting manifolds in this embodiment is as follows: Figure 1 As shown.

[0115] In complex environments, we selected a quadcopter drone as the controlled object to demonstrate an example of autonomous drone navigation. Figure 2 This section provides a structural diagram and coordinate definition for a quadcopter drone. Figure 3 This diagram illustrates the stable and unstable states of a quadcopter drone. During the experiment, the quadcopter drone was placed in a complex environment with obstacles and made to track... Figure 4 The desired path is shown. This desired trajectory can be represented by the following system of equations:

[0116]

[0117] Among them, the expected trajectory parameters are a = b = 16 and c = 8. Assume that the position of the quadrotor UAV can be represented by P = [p x p y p z T Then the position constraint of the UAV can be expressed as:

[0118]

[0119] Furthermore, for the speed constraint of the UAV, we always expect the speed direction of the UAV to be consistent with the tangent direction of the expected path, and the magnitude of the speed to always remain near the expected speed V * = 3 m / s. Then the speed constraint of the UAV can be expressed as:

[0120]

[0121] According to the ranging information of the UAV's vision sensor, we can obtain the distance r c between the UAV and the obstacle, and compare it with the minimum safety distance r. When the UAV is far from the obstacle in the environment, that is, r c > r, the UAV control system will switch to the asymptotically stable mode to control the UAV to track the expected path. At this time, we can construct the following closed-loop control system according to the position constraint ψ tr and the speed constraint ψ sp :

[0122]

[0123] ;

[0124] Let the parameter matrices T1 = T2 = 2 and T3 = 3 among them, then the above closed-loop control system is asymptotically stable.

[0125] When the UAV detects an obstacle within the safe distance, that is, r c < r, we need to determine whether there is a collision risk between the UAV and the obstacle. If the r c (k) detected by the UAV at the current moment, i.e., k moment, is greater than the r c (k - 1) at the previous moment, i.e., k - 1 moment, it means that the UAV is moving away from the obstacle and there is no collision risk, then the UAV motion control system operates in the asymptotically stable mode. On the contrary, if the r c (k) detected by the UAV at the current moment, i.e., k moment, is less than the r cIf (k-1) indicates that the drone is approaching an obstacle and collision avoidance control is needed, i.e., switching the drone control system to unstable mode. In unstable mode, to transform the obstacle into an unstable point in the drone control system, consider the following form of drone motion constraint equations:

[0126] ψ=ψ tr +Aψ sp ;

[0127] Where, ψ tr and ψ sp Let A represent the path constraint (position manifold) and the velocity constraint (velocity manifold), respectively, and let A be the coefficient matrix in diagonal form. Based on the UAV's motion constraint equations, a closed-loop control system can be defined:

[0128]

[0129] Here, T is the coefficient matrix in diagonal form under unstable mode. When both coefficient matrices A and T are negative definite matrices, the closed-loop control system will be unstable, and the UAV's trajectory will diverge outwards near the obstacle.

[0130] Considering that the drone may face multiple obstacles simultaneously, we considered both single-obstacle and multi-obstacle scenarios in the unstable mode. For the single-obstacle case, we let T = A = diag(S) i ), i = 1, ..., v, where S i The function parameters are given by the following expression:

[0131]

[0132] Where, ε 2 =(xx) o ) 2 +(yy o ) 2 +(zz o ) 2 a oi and r i Both are constants greater than zero, determining the obstacle's repulsion range and the drone's reaction radius, respectively. From the above expressions, it can be seen that in areas relatively far from the obstacle, S... i The value is basically determined by parameter a oi Decision, that is In the area near the obstacle, there are If matrices A and T are negative definite, then the obstacle is transformed into an unstable point in the UAV control system.

[0133] In the case of facing multiple obstacles simultaneously, assume the distance between the drone and the feature points of these obstacles is r. c The minimum safe distance is r. When multiple obstacles are located within the drone's reaction zone, a new variable, the bifurcation parameter β, is introduced to handle the relationship between the drone and the feature points of these obstacles. Its specific expression is:

[0134]

[0135] Where j represents the number of obstacle feature points within the UAV's perception range. Based on the above bifurcation parameters, we can define the function parameters:

[0136]

[0137] Here, s0 is a positive constant.

[0138] Based on the above steps, the simulation results of the UAV's autonomous navigation in complex environments are as follows: Figure 5 As shown. By Figure 5 As can be seen, the drone successfully bypassed multiple obstacles and consistently tracked the desired trajectory with minimal deviation, demonstrating the effectiveness and reliability of the method provided by this invention. Figure 6 The simulation results of UAV path tracking and collision avoidance control in the form of obstacle feature points are shown. From Figure 6 It makes it easier to see the divergence process of the drone's trajectory near obstacles and the convergence process in the safe area.

[0139] To better demonstrate the superiority of the unsteady state-based UAV autonomous navigation method provided by this invention, we compared it with the potential field method, vector field method, and model predictive control method. Figure 7 , Figure 8 and Figure 9 Simulation results of the UAV under the potential field method, vector field method, and model predictive control method are presented respectively. Figure 7 It can be seen that the artificial potential field method can control UAVs to complete path tracking and collision avoidance control tasks in complex environments. However, there is always a large deviation between the UAV and the desired path in this method, and it is difficult to quickly converge back to the desired path after a collision avoidance maneuver. Furthermore, observations... Figure 8 It is easy to see that the guided vector field method has good tracking performance in obstacle-free areas. However, judging from the UAV's trajectory, although it successfully avoids all obstacles in the environment, it always bypasses them from the outside of the desired path, which increases the collision avoidance cost for the UAV. Compared with the UAV autonomous navigation method based on unstable modes and the artificial potential field method provided in this invention, the guided vector field method has a very tortuous trajectory when avoiding obstacles, which undoubtedly increases the difficulty of UAV motion control. Figure 9 Simulation results of the model predictive control (MMCC) method are presented. Clearly, the MMCC method performs exceptionally well in tracking the desired path, ensuring the UAV follows the expected trajectory for almost all time periods. However, its collision avoidance performance is inferior to other methods, and in some cases, it may even collide with obstacles. In summary, the unstable mode-based autonomous UAV navigation method provided in this invention offers superior performance compared to other methods, enabling efficient, reliable, and safe UAV navigation.

[0140] The comparison of several UAV navigation methods above demonstrates that the unstable mode-based autonomous navigation method provided by this invention effectively balances path tracking, collision avoidance, and the UAV's dynamic characteristics. Among the methods compared, only Model Predictive Control (MMCC) achieves this; other methods fail to integrate UAV dynamics into trajectory planning, and MMCC's collision avoidance performance is unsatisfactory. Secondly, the method provided by this invention produces a smoother UAV trajectory, avoiding the jitter seen in artificial potential field and guided vector field methods when avoiding obstacles. Finally, the method of this invention exhibits high stability, ensuring that the UAV's trajectory converges to the desired path while maintaining its speed near a desired value, thus improving flight stability while reducing the difficulty of UAV motion control.

[0141] Compared to existing technologies, the UAV autonomous navigation method based on attracting manifolds in this invention adjusts the operating mode of the control system based on the real-time detected relative distance between the UAV and obstacles. When the distance between the UAV and the obstacle is greater than the minimum safe distance, the control system enters an asymptotically stable mode, where the UAV's trajectory gradually converges to the desired path. However, when the distance between the UAV and the obstacle is less than the safe distance, the control system switches to an unstable mode, where the UAV's trajectory diverges outward from the obstacle to avoid collision. Once the collision risk is eliminated, the control system switches back to the asymptotically stable mode. This method possesses more comprehensive functionality, simultaneously considering path tracking, collision avoidance, and the dynamic characteristics of the UAV. It can adapt to various complex environments and improve the safety and reliability of UAVs in practical applications. From the perspective of control system design, it addresses the path tracking and collision avoidance control problems in UAV autonomous navigation. Compared to optimization-based model predictive control methods and geometric feature-based guiding vector field methods, the proposed method has faster response speed, lower implementation difficulty, less computation, and less information requirements. By integrating the planning and control problems in UAV navigation, this method possesses better adaptability to changes in model parameters, exhibiting scalability, robustness, and flexibility. This integrated design not only improves the overall performance of the system but also makes the UAV trajectory obtained by this method smoother.

[0142] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. An autonomous navigation method for unmanned aerial vehicles based on an attractive manifold, characterized in that, Includes the following steps: Step S1. Set the desired trajectory of the UAV, obtain the desired trajectory parameters, and construct the attractive manifold of the UAV by combining the kinematic and dynamic performance constraints of the UAV. Step S2. Obtain the distance information between the drone and the obstacle, determine the collision risk between the drone and the obstacle, and obtain the desired working mode of the drone control system based on the collision risk; Step S3. Based on the attracting manifold and the desired working mode of the UAV control system, establish a closed-loop control system for autonomous navigation of the UAV and set the parameters of the UAV control system. Step S4. Construct a mathematical model of the UAV, substitute it into the closed-loop control system of the UAV autonomous navigation to obtain the control input of the UAV, which is used to control the UAV to avoid obstacles and track the desired trajectory to complete the autonomous navigation task. Step S1 specifically includes: Step S1-1. Set the desired trajectory of the UAV and obtain the desired trajectory parameters; wherein, the desired trajectory of the UAV consists of a target point, a straight line, and a quadratic curve that can be represented by a system of equations; Step S1-2. Obtain the kinematic performance constraints of the UAV, including maximum speed, maximum acceleration, maximum pitch angle, and maximum roll angle; Steps S1-3. Construct the attractive manifold of the UAV, including the position manifold, based on the UAV's desired trajectory parameters and kinematic performance constraints. and velocity manifold ; Steps S1-3, position manifold and velocity manifold They are represented as follows: ； ； in, It's the location of the drone. It is the heading angle of the drone. constraint functions, It is the actual speed of the drone. , , , and It is a coefficient matrix related to the desired trajectory parameters. It is time; It is the difference between the actual speed and the expected speed of the drone; Step S2 specifically includes: Step S2-1. Obtain the drone's position information, and then use the drone's visual sensor to perform distance measurement on the obstacle to obtain the distance between the drone and the obstacle. ,Will minimum safe distance Compare and obtain the comparison results; Step S2-2. Based on the comparison results of step S2-1, determine the collision risk between the drone and the obstacle, and obtain the desired working mode of the drone control system; Step S3 specifically includes: Step S3-1. Construct a closed-loop control system based on the working mode of the UAV control system; If the desired operating mode of the UAV control system is asymptotically stable, then a closed-loop control system of the following form can be constructed: ； in, , and These are the parameter matrices in the asymptotically stable mode; Will , and All matrices are set to positive definite matrices, and the closed-loop control system has asymptotic stability. The working mode of the UAV control system is set to the asymptotically stable mode. If the drone control system operates in an unstable mode, then a closed-loop control system of the following form should be constructed: ； ； in, and These are the coefficient matrices under unstable modes; and All are set to negative definite matrices. The closed-loop control system is unstable. Therefore, the working mode of the UAV control system is set to the unstable mode. When the closed-loop control system exhibits instability, parameters are used. For the parameter matrix in unstable mode and Configure it so that Then we have: ； ； in, The number of obstacles within the drone's reaction area. It is a constant greater than zero.

2. The UAV autonomous navigation method based on attracting manifolds according to claim 1, characterized in that, Step S2 also includes: Step S2-2A. If the comparison result of step S2-1 is... If the distance between the drone and the obstacle is large and the risk of collision between the drone and the obstacle is low, then the desired drone control system operating mode is the asymptotically stable mode, which is used to control the drone to track the desired trajectory. Step S2-2B. If the comparison result of step S2-1 is... ,and Moment Less than Moment This indicates that the drone is approaching an obstacle and the risk of collision between the drone and the obstacle is high. Therefore, the desired operating mode of the drone control system is the unstable mode, which controls the drone's trajectory to diverge away from the obstacle near the obstacle, thereby avoiding a collision with the obstacle. Step S2-2C. If the result of the comparison in step S2-1 is... ,and Moment Greater than Moment This indicates that the drone is moving away from the obstacle, and the risk of collision between the drone and the obstacle is low. Therefore, the desired operating mode of the drone control system is the asymptotically stable mode, which controls the drone to track the desired trajectory.

3. The UAV autonomous navigation method based on attracting manifolds according to claim 2, characterized in that, The drone's location information is obtained through the drone's autonomous positioning system; the location information of obstacles is obtained through the drone's environmental perception system.

4. The UAV autonomous navigation method based on attracting manifolds according to claim 3, characterized in that, Step S4 specifically includes: Step S4-1. Establish the mathematical model of the UAV, represented in matrix form as follows: ； ； ； The location of the drone in space is The posture is ,in , and These are roll angle, pitch angle, and yaw angle, respectively. This is the projection of gravitational acceleration onto the Earth's coordinate system. It is the projection of the total propeller thrust onto the fuselage coordinate system. This represents the torque acting on the machine body axis. Let the moment of inertia of the drone be... This refers to the gyroscopic torque. Step S4-2. Substitute the mathematical model of the UAV into the closed-loop control system constructed in S3-1, solve for the control input of the UAV, and thus realize the motion control of the UAV. Step S4-3. Repeat steps S2-1 to S4-2 until the UAV safely and without collision completes the autonomous navigation task.

Images