Unmanned aerial vehicle anti-interference autonomous navigation method in weak communication environment and related device

Abstract

The invention discloses an anti-interference autonomous navigation method for an unmanned aerial vehicle in a weak communication environment and a related device. According to the method, a VINS-Fusion algorithm is adopted to fuse odometer IMU data and camera visual data, and the attitude and the motion state of the unmanned aerial vehicle are obtained; constructing a raster map based on camera visual data, and generating an initial path by using a jump point search algorithm; converting the initial path into a B spline curve, and dynamically adjusting a control point in the B spline curve to obtain an optimized path; and inputting the pose and the motion state as real-time states, and outputting a control instruction by using a linear quadratic regulator based on the optimized path to realize the attitude control and trajectory tracking of the aircraft in the disturbance environment. According to the method, LQR control, JPS path planning, B spline trajectory optimization and a VINS-Fusion algorithm are integrated, flight stability can be guaranteed, and reliable navigation is supported.

Description

Technical Field

[0001] This invention relates to the field of autonomous navigation and control technology for unmanned aerial vehicles (UAVs), and in particular to an anti-interference autonomous navigation method and related device for UAVs in weak communication environments. Background Technology

[0002] With the continuous development of drone technology, its application in areas such as maritime infrastructure inspection, communication support, and search and rescue is becoming increasingly widespread. In complex environments such as the ocean, the autonomy and intelligence of drones become key factors in ensuring the successful execution of missions. To achieve safe flight and efficient mission completion, it is necessary to build more advanced and efficient control and perception systems.

[0003] Positioning accuracy is crucial during drone flight. While GPS signals are widely used for positioning, they are easily interfered with in obstacle-filled environments. To address this issue, Visual Inertial Navigation Systems (VINS) have been increasingly adopted to replace pure GPS navigation in recent years. VINS systems achieve stable positioning and navigation even when GPS fails by fusing visual information with inertial measurement data. Since visual navigation places high demands on system storage and computing power, and drone processors typically have limited performance, GPUs have been increasingly used to accelerate image processing tasks. Deploying OpenCV on GPU platforms has become an effective way to overcome computational bottlenecks. The efficient utilization of GPU resources significantly improves the system's real-time perception capabilities.

[0004] While ensuring positioning accuracy, the design of the control method is also crucial. Existing research has attempted to combine traditional PID controllers with deep learning methods, optimizing PID parameters through dynamic deep echo state networks to improve the stability and responsiveness of UAVs when actuators malfunction. Other researchers have used deep reinforcement learning to achieve autonomous control of UAVs under wind disturbance conditions, or introduced adaptive fuzzy control strategies to address system uncertainties. However, these methods generally suffer from strong environmental dependence, high computational complexity, and difficulty in real-time response to environmental changes.

[0005] To overcome the aforementioned shortcomings, Model Predictive Control (MPC) has emerged as a viable alternative. This type of method enhances adaptability to dynamic environments by generating predictive models and evaluating control strategies in real time. However, it inherently relies on solving complex optimization problems, resulting in high computational overhead and limiting its practical deployment on resource-constrained platforms. In contrast, the Linear Quadratic Regulator (LQR) control method offers lower computational complexity while maintaining control performance, making it suitable for resource-limited unmanned systems. It employs a linear feedback control strategy, calculating the control input based on the current system state to minimize the cost function and achieve stable system control.

[0006] Furthermore, to ensure the reliability of autonomous flight of UAVs, an effective controller alone is insufficient to handle complex real-world scenarios; an efficient path planning mechanism is also required. For decades, the A* algorithm, as a classic path search method, has been widely used in UAV path planning. Some studies have proposed a hierarchical adaptive resolution A* algorithm, which improves the path adaptability of UAVs in complex environments by optimizing the accuracy and efficiency of trajectory planning. Other studies have designed a robot path planning method based on a spatiotemporal grid model for indoor scenarios, providing a new approach to path generation in enclosed spaces. In addition, the EGO-Planner framework calculates the optimal flight trajectory in real time using direct gradient optimization technology, achieving efficient and smooth path planning, and has become a commonly used reference solution in the industry. However, these path planning methods generally suffer from a common problem: they require handling a large number of redundant nodes, which not only significantly increases the computational burden but also reduces search efficiency. This efficiency bottleneck may lead to real-time decision delays, which in turn affect flight performance due to lag in trajectory generation. Other studies are also exploring ways to improve the technology. For example, some solutions combine global planning with local optimization of RRT-star to form a hybrid path planning technique. The initial path search is completed through a global algorithm, and then the path is refined through local optimization, which enhances the flexibility of UAVs in obstacle-dense environments. Summary of the Invention

[0007] To address the aforementioned technical issues, this invention proposes an anti-interference autonomous navigation method and related devices for unmanned aerial vehicles (UAVs) in weak communication environments. By integrating LQR control, JPS path planning, B-spline trajectory optimization, and VINS-Fusion algorithm, it ensures flight stability and supports reliable navigation, making it particularly suitable for collaborative operations of unmanned systems in complex marine environments such as maritime inspection and disaster relief.

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

[0009] An anti-jamming autonomous navigation method for UAVs in weak communication environments includes the following steps:

[0010] The VINS-Fusion algorithm is used to fuse odometry IMU data and camera vision data to obtain the attitude and motion state of the UAV;

[0011] A raster map based on camera visual data was constructed, and an initial path was generated using a jump point search algorithm.

[0012] The initial path is converted into a B-spline curve, and the control points in the B-spline curve are dynamically adjusted to obtain an optimized path.

[0013] Using the pose and motion state as real-time state inputs, and based on the optimized path, a linear quadratic regulator is used to output control commands to achieve attitude control and trajectory tracking of the aircraft in a disturbed environment.

[0014] Preferably, the path control points are dynamically adjusted to obtain an optimized path, including the following steps:

[0015] For each control point of the B-spline curve, calculate the tangent vector that is aligned with the tangent direction of the control point. Define a tangent vector Vertical plane The plane intersects the extension of the initial path, forming an intersection line. ;

[0016] The intersection line Intersects with the surface of an obstacle in the environment at a point Calculate control points Intersection with the surface of the obstacle distance The distance formula is used to assess collision risk:

[0017] (7)

[0018] in, From point to , unit vector;

[0019] like This indicates the control point. If no obstacle is encountered, there is no risk of collision at this point; otherwise, it indicates a control point. There is a risk of collision; adjust the path.

[0020] Preferably, the speed at the control point acceleration and accelerometer Represented as:

[0021] (8)

[0022] By adjusting the control points, the shape and curvature of the trajectory can be optimized to meet flight requirements. The objective function for trajectory optimization is defined as:

[0023] (9)

[0024] in, For smoothing terms, For collision terms, For a kinetic feasible term, , , These are the weighting coefficients for each item.

[0025] Preferably, the smoothing term The definition is as follows:

[0026] (10)

[0027] in, for The square of the L2 norm, for The squared L2 norm, N c This represents the total number of control points for the B-spline curve.

[0028] Preferably, the collision item The definition is as follows:

[0029] (11)

[0030] in,

[0031] (12)

[0032] in, , For safe distance threshold, Control points The collision cost generated at point (p,v), where p represents the point... The position vector at point v, where v represents the point. The velocity vector at that location.

[0033] Preferably, the kinetic feasible term The dynamic feasibility of the trajectory is ensured by constraining the higher-order derivatives of the control points, as defined below:

[0034] (13)

[0035] in, , , As weight, It is a twice-continuously differentiable metric function with higher-order derivatives.

[0036] Based on the above, the present invention also discloses a dynamic trajectory planning system for unmanned aerial vehicles (UAVs), comprising:

[0037] The state awareness module is used to fuse odometry IMU data and camera vision data using the VINS-Fusion algorithm to obtain the attitude and motion state of the UAV.

[0038] The path generation module is used to construct a raster map based on camera visual data and uses a jump point search algorithm to generate an initial path.

[0039] The path optimization module is used to convert the initial path into a B-spline curve, dynamically adjust the control points in the B-spline curve, and obtain an optimized path.

[0040] The flight control module is used to take the attitude and motion state as real-time status input, and output control commands based on the optimized path using a linear quadratic regulator to realize the attitude control and trajectory tracking of the aircraft in a disturbed environment.

[0041] Based on the above, the present invention also discloses a drone, comprising:

[0042] The fuselage itself;

[0043] A control system that communicates with the fuselage body, the control system including a memory and a processor, the memory for storing computer programs; the processor for executing the computer programs to implement any of the methods described above.

[0044] Based on the foregoing, the present invention also discloses a readable storage medium storing a computer program, which, when executed by a processor, implements any of the methods described above.

[0045] Based on the above technical solution, the beneficial effects of this invention are as follows: This invention discloses an anti-interference autonomous navigation method and related device for unmanned aerial vehicles (UAVs) in weak communication environments. This method uses the VINS-Fusion algorithm to fuse odometry IMU data and camera visual data to obtain the UAV's attitude and motion state; it constructs a grid map based on camera visual data and uses a jump point search algorithm to generate an initial path; it converts the initial path into a B-spline curve and dynamically adjusts the control points in the B-spline curve to obtain an optimized path; it uses the attitude and motion state as real-time state input, and based on the optimized path, it uses a linear quadratic regulator to output control commands, realizing attitude control and trajectory tracking of the aircraft in disturbed environments. Facing challenges such as unstable communication links, external wind disturbances, and trajectory instability caused by dense obstacles common in maritime scenarios, this invention integrates LQR control, JPS path planning, B-spline trajectory optimization, and the VINS-Fusion algorithm to ensure flight stability and support reliable navigation. It is particularly suitable for collaborative operations of unmanned systems in complex maritime environments such as maritime inspection and disaster relief. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating an anti-interference autonomous navigation method for unmanned aerial vehicles (UAVs) in a weak communication environment, as shown in one embodiment.

[0047] Figure 2 This is a convex hull characteristic diagram of a B-spline in one embodiment. Red dots represent control points, and arrows indicate directions;

[0048] Figure 3 This is a diagram of the experimental environment in one embodiment;

[0049] Figure 4 This is a test diagram of the wind resistance performance of a drone flying in a narrow space in one embodiment;

[0050] Figure 5 This is a schematic diagram of obstacle avoidance and path planning in one embodiment;

[0051] Figure 6 This is a schematic diagram of an embodiment where a drone passes through obstacles A, B, and C in sequence in an outdoor environment;

[0052] Figure 7 This is a schematic diagram illustrating part of the path and obstacles of a drone in one embodiment. The drone passes through the left side of the tree to reach the designated location;

[0053] Figure 8 One embodiment involves a takeoff test conducted in the same outdoor environment, wherein... Figure 8 (a): Takeoff test using the method of the present invention; Figure 8 (b): Takeoff testing using the Ego-Planner framework;

[0054] Figure 9 This is a comparison of different path planning algorithms in one embodiment;

[0055] Figure 10 One embodiment compares accelerations, where, Figure 10 (a): Comparison of acceleration on the X-axis between Ego-Planner and the method of this invention; Figure 10 (b): Comparison of acceleration on the Y-axis between Ego-Planner and the method of the present invention;

[0056] Figure 11 One embodiment compares angular velocities, where, Figure 11 (a): Comparison of angular velocity on the X-axis between Ego-Planner and the method of this invention; Figure 11 (b): Comparison of angular velocity on the Y-axis between Ego-Planner and the method of the present invention. Detailed Implementation

[0057] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0058] like Figure 1 As shown in the figure, this embodiment provides an anti-interference autonomous navigation method for UAVs in weak communication environments, including the following steps.

[0059] Step 100: The VINS-Fusion algorithm is used to fuse odometry IMU data and camera vision data to obtain the attitude and motion state of the UAV.

[0060] In this embodiment, considering that GPS signals are often missing in environments with dense obstacles, when GPS is unavailable, a Realsense D435 stereo camera is used to collect camera visual data (RGB images and depth images) in real time. The flight controller uses a PX4 autopilot to collect odometer IMU data in real time. A vision-inertial navigation system (VINS-Fusion algorithm) is used to fuse odometer IMU data and camera visual data information, extract feature points, and complete image reprojection and attitude estimation. This enables accurate perception of the UAV's pose and motion state, providing real-time status input for subsequent path planning and controller feedback.

[0061] Step 200: Construct a raster map based on camera visual data and use a jump point search algorithm to generate an initial path.

[0062] In this embodiment, the Jump Point Search (JPS) algorithm is used to plan the path for the UAV during the initial path generation stage. The core advantage of the JPS algorithm is its accelerated path planning, making it particularly suitable for UAVs with limited computing resources. Its core lies in identifying and utilizing jump points j to bypass redundant intermediate nodes and directly focus on key nodes, significantly reducing the search space and lowering computational overhead and path search time. By leveraging jump techniques, the JPS algorithm reduces the number of nodes that need to be traversed, generating high-quality paths while significantly reducing computational complexity and achieving rapid path planning, making it particularly suitable for UAV systems with limited computing resources.

[0063] Step 300: Convert the initial path into a B-spline curve, and dynamically adjust the control points in the B-spline curve to obtain an optimized path.

[0064] In this embodiment, the initial path only satisfies the terminal constraints and does not consider obstacle avoidance. To address this issue, the UAV employs an iterative obstacle avoidance process: it detects obstacles using a depth camera and constructs a map, ultimately generating a collision-free trajectory. In obstacle detection and avoidance, such as... Figure 2 As shown, the UAV utilizes B-spline curves Describe the trajectory.

[0065] The initial path is converted into a B-spline curve, and the path control points are dynamically adjusted to avoid obstacles. Spatial relationships between control points and obstacles are established using depth map information, collision detection metrics and constraint cost functions are defined, and the trajectory is optimized through gradient descent. This allows the path to achieve obstacle avoidance and energy consumption optimization while satisfying smoothness and dynamic feasibility.

[0066] The initial path is generated by the JPS algorithm and iteratively adjusted through the following steps:

[0067] (1) Calculation of control points and tangent vectors: Define control points on the B-spline curve. For each control point, calculate the tangent vector that is aligned with its tangent direction. .

[0068] (2) Definition of perpendicular plane and intersection line: for each control point Define a tangent vector Vertical plane The plane intersects the extension of the original path, forming an intersection line. .

[0069] (3) Obstacle intersection detection: intersection line Intersects with the surface of an obstacle in the environment at a point This point is the projection of the obstacle's surface onto the line of intersection. The obstacle is detected by a depth camera. Let be the intersection point of the straight line on the surface of the obstacle.

[0070] (4) Obstacle distance calculation: Calculate control points Intersection with the surface of the obstacle distance This is used to assess collision risk. The distance formula is:

[0071] (7)

[0072] in, From point to The unit vector. If Explain the control points There is no risk of collision at this point as no obstacle has been encountered.

[0073] (5) Trajectory adjustment and obstacle avoidance: If a collision risk is detected, the UAV adjusts its path according to the calculation results to ensure that the control point avoids the surface of the obstacle, thereby effectively avoiding collision.

[0074] Through the above process, the drone can dynamically adjust its trajectory based on real-time obstacle detection and mapping to ensure a collision-free flight path and achieve efficient obstacle avoidance.

[0075] The smoothing properties of B-spline curves ensure a smooth transition between control points, improving flight stability. Uniform B-splines and their convex hull properties are used to measure the velocity at the control points. acceleration and accelerometer It can be represented as:

[0076] (8)

[0077] By adjusting the control points, the shape and curvature of the trajectory can be optimized to meet flight requirements. The objective function for trajectory optimization is defined as:

[0078] (9)

[0079] in, For smoothing terms, For collision terms, This is a kinetically feasible term. , , These are the weighting coefficients for each item.

[0080] Smoothing Term The definition is as follows:

[0081] (10)

[0082] in, for The square of the L2 norm, for The square of the L2 norm, This represents the total number of control points for the B-spline curve.

[0083] Collision items The definition is as follows:

[0084] (11)

[0085] in,

[0086] (12)

[0087] Here, , For safe distance threshold, Control points The collision cost generated at point (p,v) is given, where p represents the point... The position vector at point v, where v represents the point. The velocity vector at that location.

[0088] Dynamically feasible terms The dynamic feasibility of the trajectory is ensured by constraining the higher-order derivatives of the control points, as defined below:

[0089] (13)

[0090] in, , , For each weight, It is a twice-continuously differentiable metric function with higher-order derivatives.

[0091] The UAV utilizes the convex hull property of B-splines to constrain velocity, acceleration, and jerk within preset thresholds. Numerical optimization is performed using the finite memory Broyden-Fletcher-Goldfarb-Shanno algorithm to ensure fast convergence and achieve efficient path planning in obstacle-dense environments.

[0092] Step 400: The pose and motion state are used as real-time state inputs. Based on the optimized path, control commands are output using a linear quadratic regulator to achieve attitude control and trajectory tracking of the aircraft in a disturbed environment.

[0093] In this embodiment, the overall state equation of the UAV can be expressed as:

[0094] (1)

[0095] in, yes The state vector at any given time describes the motion state of the UAV; It represents the state transition pattern of the system at each time step (including updates of position and velocity); It is the control input matrix, which reflects the influence of the control quantity on the state; yes Real-time control inputs (such as thrust and acceleration commands).

[0096] To design the control strategy for the UAV, the state error equation is defined as follows:

[0097] (2)

[0098] in, This is the desired state for the drone. This is the actual state. This is the desired control quantity.

[0099] To optimize system performance, the goal of a linear quadratic regulator (LQR) is to minimize the following cost function:

[0100] (3)

[0101] in, yes transpose, yes transpose, It is the state error penalty matrix. The control input penalty matrix and the control cost are weighted to reflect the relative importance of the state error and the control cost. Solving this objective requires deriving the optimal gain matrix using the Discrete-Time Algebraic Riccati Equation (DARE). Finally, the LQR control law is obtained:

[0102] (4)

[0103] In the formula, the gain matrix The formula for calculation is:

[0104] (5)

[0105] in, for transpose, It is the solution to DARE, representing the weight matrix of the system state; This is the cross-weight matrix, used to penalize the coupling effect between state error and control input. for The transpose. For drones, assuming a small angle approximation, the roll angle can be adjusted via attitude control. and pitch angle The formula is defined as:

[0106] (6)

[0107] in, It is gravitational acceleration. It is the rotation matrix from the body coordinate system to the ground coordinate system (used for coordinate transformation of acceleration); It is the acceleration vector in the UAV's body coordinate system (mainly composed of the roll angle). and pitch angle Decide).

[0108] During flight, the system calculates LQR control output in real time based on state feedback and path tracking error, and adjusts thrust and attitude angle to ensure flight stability under external disturbances such as wind disturbance and obstacle changes, while dynamically updating the path to cope with environmental changes.

[0109] It should be understood that although the steps in the flowchart above are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart above may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0110] experiment

[0111] The method of this invention was compared and verified with the EGO-Planner framework. UAV flight tests were conducted in both indoor and outdoor environments, with the experimental platform and hardware configuration as follows: Figure 3 As shown in the table, the UAV is built on an F450 frame and equipped with a Jetson Xavier NX onboard computer, responsible for the real-time operation of perception and planning algorithms. The entire software environment is built on ROS1, and the UAV status is monitored in real time through a ground control station, continuously providing feedback on flight performance (including telemetry data and system diagnostics), facilitating the tracking of key operating parameters and adjustments as needed. Table 1 shows the core parameters of the UAV (weight, maximum speed, control frequency, etc.), which have a significant impact on flight performance and stability.

[0112]

[0113] By integrating hardware and software components, along with real-time monitoring and security mechanisms, the method proposed in this invention has been effectively validated in complex indoor and outdoor environments. During the experiment, this configuration ensured that the drone operated autonomously and stably at all times.

[0114] 1) Indoor experiments

[0115] The indoor flight experiment was conducted in a 5m×5m enclosed space. To verify the stability and path planning efficiency of the method of this invention, a fan was placed next to the UAV to create airflow disturbance (see...). Figure 4 The wind speed reached approximately 5 m / s. Compared to EGO-Planner, this method significantly improves the stability of the UAV during takeoff, enhances its anti-interference capability, and highlights the advantages of the method in maintaining the initial dynamic stability of the UAV (especially when dealing with external disturbances).

[0116] Furthermore, the drone exhibited clear obstacle avoidance maneuvers when encountering obstacles in flight (see...). Figure 5Furthermore, the system can quickly regain stability after execution. These avoidance maneuvers are achieved through real-time adjustment of trajectory control points: the control points generated by the trajectory guide the UAV to avoid obstacles, and when the UAV approaches an obstacle, the JPS algorithm dynamically generates new control points, enabling the UAV to quickly change course and avoid collisions without sacrificing trajectory smoothness. This optimization is particularly prominent in high-frequency trajectory planning: the method of this invention can generate a 100Hz collision-free smooth trajectory, ensuring that the UAV successfully avoids obstacles and quickly returns to stable flight.

[0117] Indoor comparative experiments show that the advantage of the method of this invention stems from the combination of two factors: LQR's fine optimization of dynamic control and JPS's high efficiency in path planning. This synergy enables the UAV to quickly recover stability even when disturbed in obstacle-dense environments or when autonomously performing rapid and smooth obstacle avoidance, outperforming EGO-Planner.

[0118] 2) Outdoor experiments

[0119] In the outdoor crossing experiment, the distribution of obstacles along some of the drone's flight paths was as follows: Figure 6 As shown. Compared to EGO-Planner, the method of this invention exhibits stronger anti-interference capabilities during takeoff, with a measured wind speed of 7.4 m / s. With the help of an optimized LQR controller, the system quickly adjusts the control input, enabling the UAV to maintain a balanced and stable takeoff attitude even under wind disturbances, avoiding lateral drift caused by oscillations and external factors.

[0120] During obstacle avoidance and path planning, the method of this invention utilizes depth information to construct an occupied grid map in real time, combined with the JPS efficient path search algorithm, to generate a smooth, collision-free flight path for the UAV in outdoor environments. When encountering obstacles, the method of this invention does not directly avoid obstacles by tilting the fuselage, but instead uses a visual depth sensor to collect environmental data, replans the path, and adjusts the UAV's heading by adjusting the nose direction (see...). Figure 7 This precise adjustment, with its environmental awareness, significantly reduces lateral drift and instability, ensuring a smoother flight. In contrast, the EGO-Planner framework performs poorly under the same wind conditions. We measured the takeoff time of both methods, and the results are shown in Table 2. Due to limitations in its controller and path planning capabilities, the EGO-Planner system is more susceptible to wind interference during takeoff, resulting in longer takeoff times and more pronounced fuselage oscillations. Furthermore, EGO-Planner relies on a simpler obstacle avoidance mechanism and cannot efficiently replan the global path as the method described in this invention, making it more prone to lateral drift and a less stable flight trajectory.

[0121]

[0122] Figure 8The performance of the proposed method versus the EGO-Planner framework was compared under the same conditions for the same UAV. During takeoff, despite external wind interference, the proposed method achieved a fast and stable takeoff; while the EGO-Planner's takeoff process was relatively shaky, with a slower flight speed and significant lateral drift.

[0123] The JPS algorithm is compared with A*, RRT, and D* algorithms in terms of cumulative path cost and computation time, such as Figure 9 As shown in the figure. Analysis shows that, in terms of planning speed, the UAV equipped with the method of this invention plans the path faster, while EGO-Planner experiences a significant delay when encountering obstacles.

[0124] To further evaluate the control accuracy and stability of the two systems, a detailed analysis of the acceleration in the X and Y directions in the body coordinate system was conducted (see...). Figure 10 In the X direction, the UAV equipped with the method of this invention (LQR-JPS) exhibits a significantly smaller acceleration fluctuation range, indicating better stability during horizontal flight and a smaller acceleration adjustment range;

[0125] In the Y direction, the EGO-Planner exhibits more pronounced acceleration changes, but the method of this invention still stably controls the acceleration changes within the range of -0.2 m / s² to +0.2 m / s², while the EGO-Planner's acceleration changes exceed this range. After comparing specific data depths, the results show that the method of this invention only causes minor turbulence during takeoff and landing, and maintains stable displacement control throughout the entire process.

[0126] In addition, such as Figure 11 As shown, the angular velocity variations near the X and Y axes of the UAV were also analyzed. Under the method of this invention, the angular velocity fluctuations of the UAV in the X and Y directions are relatively small, especially when approaching obstacles, the UAV can achieve smooth path tracking through stable attitude adjustments. The globally optimal control provided by LQR ensures precise control during attitude changes, further improving flight stability. In contrast, the EGO-Planner system experiences more drastic angular velocity fluctuations during flight, leading to abrupt attitude changes and poorer flight stability. With the help of globally optimized control, the method of this invention (LQR-JPS) can finely control the angular velocity in each direction during the attitude adjustment phase, ensuring stability throughout the flight. In addition, this framework can also evaluate the current state in real time and dynamically adjust, minimizing abrupt angular velocity changes in obstacle-dense environments and maintaining smoother flight performance.

[0127] In summary, this invention proposes an anti-interference autonomous navigation method for UAVs in weak communication environments. It integrates LQR control, JPS path planning, B-spline trajectory optimization, and a VINS-Fusion algorithm. The design aims to address challenges common in maritime scenarios, such as GPS signal loss and obstacle interference. By fusing perception and control strategies, it ensures flight stability and supports reliable navigation. Experimental results show that the system can quickly generate smooth, collision-free trajectories and rapidly recover flight stability after being disturbed. Flight tests and comparative evaluations have verified the reliability and efficiency of the framework, providing a practical solution for UAV applications in similar constrained scenarios.

[0128] Based on the same inventive concept, this application also provides a system for implementing the above-described method for anti-interference autonomous navigation of unmanned aerial vehicles (UAVs) in weak communication environments. The solution provided by this system is similar to the solution described in the above method, and therefore will not be repeated here.

[0129] In one embodiment, an anti-jamming autonomous navigation system for unmanned aerial vehicles (UAVs) in a weak communication environment is also provided, comprising:

[0130] The state awareness module is used to fuse odometry IMU data and camera vision data using the VINS-Fusion algorithm to obtain the attitude and motion state of the UAV.

[0131] The path generation module is used to construct a raster map based on camera visual data and uses a jump point search algorithm to generate an initial path.

[0132] The path optimization module is used to convert the initial path into a B-spline curve, dynamically adjust the control points in the B-spline curve, and obtain an optimized path.

[0133] The flight control module is used to take the attitude and motion state as real-time status input, and output control commands based on the optimized path using a linear quadratic regulator to realize the attitude control and trajectory tracking of the aircraft in a disturbed environment.

[0134] In the above embodiments, the various modules of the UAV anti-jamming autonomous navigation system under weak communication environments can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0135] In one embodiment, a drone is also provided, comprising: a fuselage body; and a control system communicating with the fuselage body, the control system including a memory and a processor, the memory for storing a computer program; and the processor for executing the computer program to implement the steps as described in all the above method embodiments.

[0136] In one embodiment, a readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps as described in all the above method embodiments.

[0137] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0138] The embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0139] The above are merely preferred embodiments of the present application and are not intended to limit the scope of the present application. Various modifications and variations can be made to the embodiments of the present application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of the present application should be included within the protection scope of the embodiments of the present application.

Claims

1. An anti-interference autonomous navigation method for unmanned aerial vehicles (UAVs) in weak communication environments, characterized in that, Includes the following steps: The VINS-Fusion algorithm is used to fuse odometry IMU data and camera vision data to obtain the attitude and motion state of the UAV; A raster map based on camera visual data was constructed, and an initial path was generated using a jump point search algorithm. The initial path is converted into a B-spline curve, and the control points in the B-spline curve are dynamically adjusted to obtain an optimized path. Using the pose and motion state as real-time state inputs, and based on the optimized path, a linear quadratic regulator is used to output control commands to achieve attitude control and trajectory tracking of the aircraft in a disturbed environment.

2. The anti-interference autonomous navigation method for UAVs in weak communication environments according to claim 1, characterized in that, Dynamically adjusting path control points to obtain an optimized path includes the following steps: For each control point of the B-spline curve, calculate the tangent vector that is aligned with the tangent direction of the control point. Define a tangent vector Vertical plane The plane intersects the extension of the initial path, forming an intersection line. ; The intersection line Intersects with the surface of an obstacle in the environment at a point Calculate control points Intersection with the surface of the obstacle distance The distance formula is used to assess collision risk: (7) in, From point to , unit vector; like This indicates the control point. If no obstacle is encountered, there is no risk of collision at this point; otherwise, it indicates a control point. There is a risk of collision; adjust the path.

3. The anti-interference autonomous navigation method for UAVs in weak communication environments according to claim 2, characterized in that, The speed at the control point acceleration and accelerometer Represented as: (8) By adjusting the control points, the shape and curvature of the trajectory can be optimized to meet flight requirements. The objective function for trajectory optimization is defined as: (9) in, For smoothing terms, For collision terms, For a kinetic feasible term, , , These are the weighting coefficients for each item.

4. The anti-interference autonomous navigation method for UAVs in weak communication environments according to claim 3, characterized in that, The smoothing item The definition is as follows: (10) in, for The square of the L2 norm, for The squared L2 norm, N c This represents the total number of control points for the B-spline curve.

5. The anti-interference autonomous navigation method for UAVs in weak communication environments according to claim 3, characterized in that, The collision item The definition is as follows: (11) in, (12) in, , For safe distance threshold, Control points The collision cost generated at point (p,v), where p represents the point... The position vector at point v, where v represents the point. The velocity vector at that location.

6. The anti-interference autonomous navigation method for UAVs in weak communication environments according to claim 2, characterized in that, The kinetic feasible terms The dynamic feasibility of the trajectory is ensured by constraining the higher-order derivatives of the control points, as defined below: (13) in, , , As weight, It is a twice-continuously differentiable metric function with higher-order derivatives.

7. A dynamic trajectory planning system for unmanned aerial vehicles (UAVs), characterized in that, include: The state awareness module is used to fuse odometry IMU data and camera vision data using the VINS-Fusion algorithm to obtain the attitude and motion state of the UAV. The path generation module is used to construct a raster map based on camera visual data and uses a jump point search algorithm to generate an initial path. The path optimization module is used to convert the initial path into a B-spline curve, dynamically adjust the control points in the B-spline curve, and obtain an optimized path. The flight control module is used to take the attitude and motion state as real-time status input, and output control commands based on the optimized path using a linear quadratic regulator to realize the attitude control and trajectory tracking of the aircraft in a disturbed environment.

8. A drone, characterized in that, include: The fuselage itself; A control system that communicates with the fuselage body, the control system including a memory and a processor, the memory for storing computer programs; A processor for executing the computer program to implement the method as described in any one of claims 1 to 6.

9. A readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images