Air-ground switching and control method of air-ground amphibious unmanned aerial vehicle based on variable configuration

Abstract

The invention discloses an air-ground switching and control method of an air-ground amphibious unmanned aerial vehicle based on a variable configuration, and relates to the technical field of unmanned aerial vehicle control, and the method comprises the steps: obtaining attitude output data and height detection data through calculation based on detection data, and obtaining an optimal estimation value of an unmanned aerial vehicle angle through fusion; and performing noise reduction based on the optimal estimated value to obtain an output signal, further obtaining an output signal of real-time error control and total disturbance compensation through a cascade improved ADRC controller, and performing stepping landing flight control of dynamic error adjustment on the unmanned aerial vehicle based on the output signal of real-time error control and total disturbance compensation. And then a multi-condition combined ground mode is adopted for safe switching judgment, a servo steering engine control signal is obtained based on synchronous movement of a servo steering engine, and then the angle of a vehicle arm of the unmanned aerial vehicle is changed according to the control signal so as to control the unmanned aerial vehicle to carry out air-ground switching. According to the invention, the unmanned aerial vehicle can realize the anti-interference and stable attitude while changing the frame type autonomous landing and air-ground switching of the unmanned aerial vehicle.

Description

Technical Field

[0001] The present application relates to the field of UAV control technology, and in particular to an air-to-ground switching and control method for an amphibious UAV based on a variable configuration. Background Art

[0002] With the rapid development of drone technology, its application scenarios have expanded from single-use aerial operations to multimodal missions combining land and air. Currently, traditional drone control still relies primarily on manual intervention or simple preset programs to enable the drone to autonomously adjust its flight attitude or switch to ground driving mode. However, the drone control process may be affected by complex factors such as multiple interference sources and frequent unknown disturbances. This can lead to deviations in the drone's sensor data, poor synchronization of multi-servo control, and the inability to autonomously adjust its flight attitude or switch to ground driving mode based on real-time environmental information. This can result in the drone losing attitude control and failing to maintain its autonomous operation capabilities.

[0003] Multimodal switching can present numerous challenges for autonomous landing in areas with complex terrain and changing environments, leading to attitude loss or switching failure during multimodal switching. Therefore, UAV control requires simultaneous consideration of flight attitude, ground attitude, and the ability to autonomously adjust flight attitude based on real-time environmental information during transitions between the two. Recent advances in sensor technology have provided more accurate data support for UAV attitude measurement and environmental perception. How to effectively integrate this sensor data to enable autonomous landing during UAV configuration changes while also resisting interference and stabilizing attitude to improve attitude estimation accuracy has long been a key concern. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose an air-to-ground switching and control method for an amphibious drone based on a variable configuration, so as to enable the drone to change its frame configuration and autonomously land while being able to resist interference and stabilize its posture.

[0005] To achieve the above objectives, the present invention provides an air-to-ground switching and control method for an amphibious drone with a variable configuration, the method comprising the following steps:

[0006] The attitude output data and height detection data are obtained based on the detection data after the drone sensor is initialized;

[0007] Fusing the attitude output data and the height detection data to obtain an optimal estimate of the drone angle;

[0008] Noise reduction is performed based on the optimal estimate of the drone angle to obtain a full-band, noise-free output signal;

[0009] Based on the full-band, noise-free output signal, an output signal for real-time error control and total disturbance compensation is obtained through a cascade improved ADRC controller;

[0010] Based on the output signals of the real-time error control and total disturbance compensation, the UAV is subjected to a step-by-step landing flight control with dynamic error adjustment;

[0011] Based on the ground mode safety switching judgment using multiple conditions combined under landing flight control, a judgment result is obtained;

[0012] According to the judgment result, PID closed-loop control is adopted to coordinate the synchronous movement of the servo steering gear to obtain a servo steering gear control signal, and then the arm angle of the UAV is changed according to the control signal to control the UAV to perform air-to-ground switching.

[0013] In some embodiments, the step of calculating the attitude output data and the altitude detection data based on the detection data after initialization of the drone sensor includes the following steps:

[0014] Obtaining the attitude output data by calculating the initialization detection data of the gyroscope, magnetometer, and accelerometer in the drone;

[0015] The initialized detection data is used to calculate and obtain two different digital distance values of the ultrasonic module and the infrared ranging sensor as the height detection data;

[0016] The ultrasonic module's calculation method converts the distance calculated based on the speed of sound into the original signal, and the air density change is compensated by the ambient temperature. The formula is:

[0017] v=331.4+0.6·T ℃ (m / s);

[0018] Among them, T ℃ is the current ambient temperature;

[0019] The infrared ranging sensor calculation method is based on converting the output analog voltage or digital signal into distance through a calibration curve, and the conversion formula established by linear fitting is:

[0020] d infrared =k·v+b;

[0021] Among them, k and b are calibration coefficients.

[0022] In some embodiments, fusing the attitude output data and the altitude detection data to obtain an optimal estimate of the drone angle comprises the following steps:

[0023] Optimizing the quaternion solution to calculate the attitude output data based on the gradient descent method so that the error between the predicted acceleration and the measured acceleration is minimized;

[0024] The method for solving the attitude output data is to construct a gravity vector model by calculating the three attitude angles obtained by the rotation matrix of the body coordinate system and the geographic coordinate system, which are the angles of rotation around the X, Y, and Z axes of the geographic coordinate system:

[0025]

[0026] The gravity vector in the geographic coordinates is g earth =[0,0,1], quaternion q = [q0,q1,q2,q3] rotated to the body coordinate system;

[0027] The objective function is used to obtain the partial derivative of each component of the quaternion based on the objective function to obtain the gradient of the objective function E with respect to the quaternion q, wherein the objective function is the sum of squares of the errors between the accelerometer measurement value and the theoretical gravity vector;

[0028] The gradient of the objective function E with respect to the quaternion q The calculation formula is:

[0029]

[0030] Among them, e x ,e y ,e z The error between the accelerometer measurement and the theoretical gravity on each axis is calculated by adjusting the quaternion and normalizing the gradient to the following formula:

[0031]

[0032] In some embodiments, the method further comprises the following steps:

[0033] The angular motion equations between the pitch, roll, and yaw motions and the system are obtained by solving the attitude output data:

[0034]

[0035] The θ, ψ are the three attitude angles of the UAV, namely the roll angle, pitch angle and yaw angle, which are the angular motions of the rotation around the X, Y and Z axes of the geographic coordinate system;

[0036] Based on the air-ground coordinate system conversion, the system motion control formula in the ground coordinate system is obtained as follows:

[0037]

[0038] are the accelerations on the x, y, and z axes in the system's ground coordinate system, and g is the gravity coefficient.

[0039] In some embodiments, the method further comprises the following steps:

[0040] Based on the Kalman filter, real-time sensor data is obtained through model prediction and sensor measurement to provide optimal state estimation by minimizing the covariance of the estimation process noise;

[0041] The prediction formula for the covariance of the prior estimation process noise is:

[0042]

[0043] The update formula for the covariance of the prior estimation process noise is:

[0044]

[0045] Among them, Q = 0.1 is the process noise variance, R = 0.2 is the measurement noise variance, K K is the Kalman gain, is the optimal estimate.

[0046] In some embodiments, the step of fusing the height detection data includes the following steps:

[0047] Based on stage division and weight distribution, the advantages of ultrasound and infrared are combined to improve the robustness of height measurement, and the fusion formula is constructed as follows:

[0048] d funsion =w u ·d fultrasonic +w i ·d infrared ;

[0049] The ultrasonic weight is w u , the infrared weight is w i , dual sensor fusion is d funsion .

[0050] In some embodiments, performing noise reduction on the optimal estimate based on the drone angle to obtain a full-band, noise-free output signal includes the following steps:

[0051] The accelerometer is low-pass filtered to suppress high-frequency noise based on the first-order complementary filtering algorithm. The correction amount of the high-frequency noise and the integral amount of the low-frequency noise are expressed as follows:

[0052]

[0053] Dynamically adjust the trust coefficient α∈[0.01,0.1], where T is the unit conversion coefficient. The unit conversion coefficient is 1 when it is normally complementary and 100 when the accelerometer is converted from meters to centimeters.

[0054] The low-pass filter is obtained by dynamically fusing the attitude observation value and the fused distance value based on the first-order complementary filter, and the high-frequency noise of the sensor is suppressed by the second-order Butterworth low-pass filter. The cutoff frequency of the low-pass filter can be dynamically configured, including:

[0055] The formula for first-order complementary filtering is:

[0056] y k =(1-α)(x k-1 +B k x k );

[0057] Among them, y k is the final output value, x k is the sensor measurement value with high-frequency noise, B k is the differential of the sensor measurement value with high-frequency noise, dt is the data processing period;

[0058] The formula for suppressing high-frequency noise using a second-order Butterworth low-pass filter is:

[0059] y k =b0x k +b1x k*-1 +b2x k-2 -a1y k-1 -a2y k-2 ;

[0060] Among them, the coefficients b0, b1, b2, a1, and a2 are calculated by bilinear transformation and normalized to the cutoff frequency, and x k is the current input, y k is the current output, x k-1 is the input of the previous moment value, y k-1 The output of the previous moment value.

[0061] In some embodiments, obtaining an output signal for real-time error control and total disturbance compensation based on the full-band, noise-free output signal through a cascaded improved ADRC controller comprises the following steps:

[0062] The input signal and output signal of the system are processed based on the tracking differentiator and the extended state observer respectively, and the nonlinear state error feedback control law controls the error between the processed signals and compensates for the total disturbance;

[0063] The dynamic equations of the tracking differentiator and extended state observer are:

[0064]

[0065] The observer dynamically adjusts the state estimate z through e(t) i (t), so that z1(t)→y1(t), z2(t) estimates the velocity, z3(t) estimates the acceleration and so on, with each order of dynamics being scaled by a parameter ε n-1 Adjust the nonlinear gain to make the cascade stable and estimate the total disturbance z n+1 (t) is expanded to additional states and the perturbation dynamics is passed through ε -1 Amplify the feedback gain to quickly track disturbance changes.

[0066] In some embodiments, the ground mode safety switching judgment based on multiple conditions combined under landing flight control to obtain a judgment result includes the following steps:

[0067] Monitor and judge the ground flatness based on the attitude output data of the drone's gyroscope, accelerometer, and magnetometer;

[0068] The composite inclination angle is calculated using the Euclidean norm θ x 2 +θ y 2 is the total tilt degree. If the composite tilt angle exceeds the threshold θ max , if it is determined that the ground is uneven, a switching prohibition signal is sent;

[0069] Ultrasonic sensors are installed on the left and right sides of the drone to measure the minimum distance between the two sides in real time. left ,d right There is enough space on both sides to expand. If the minimum value is less than d min If it is determined that there is insufficient space, a switching prohibition signal is sent.

[0070] In some embodiments, the step of using PID closed-loop control to coordinate the synchronous motion of the servo steering gear according to the judgment result to obtain a servo steering gear control signal includes the following steps:

[0071] The control variable is generated based on the error between the target angle and the current angle, and is output through the discretized PID algorithm; the control variable output by the discretized PID algorithm is:

[0072]

[0073] Among them, μ[k] is the control quantity of the kth cycle, T s is the sampling cycle time, e[k] is the angle error of the kth cycle, and the angle error calculation formula is:

[0074] e(t)=θ d -θ t ;

[0075] Among them, θ d is the target angle of the retracted arm in land mode, θ t is the current angle fed back by the servo angle sensor;

[0076] Convert the PID control quantity into a servo duty cycle signal to coordinate the synchronous movement of multiple servos; the duty cycle calculation formula of the PWM conversion module is:

[0077] D[k]=clip(D[k-1]+K csale ·u[k],D min ,D max );

[0078] Among them, K scale is the scale factor, clip(x,a,b) means limiting x to the interval [a,b].

[0079] The embodiments of the present application include at least the following beneficial effects:

[0080] The present application can obtain attitude output data and altitude detection data based on the detection data after initialization of the drone sensor; fuse the attitude output data and altitude detection data to obtain the optimal estimate of the drone angle; perform noise reduction based on the optimal estimate of the drone angle to obtain a full-band, noise-free output signal; based on the full-band, noise-free output signal, obtain the output signal of real-time error control and total disturbance compensation through the cascade improved ADRC controller; based on the output signal of real-time error control and total disturbance compensation, adopt dynamic error adjustment for step-by-step landing flight control of the drone; based on the ground mode safety switching judgment using multiple conditions under landing flight control, obtain the judgment result; according to the judgment result, use PID closed-loop control to coordinate the synchronous movement of the servo servo to obtain the servo servo control signal, and then change the drone's arm angle according to the control signal to control the drone to switch between air and ground. The present application uses the detection data of the drone sensor to realize the anti-interference and stable attitude of the drone while changing the frame configuration and autonomous landing, significantly improving the autonomous operation capability in complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0082] Figure 1A schematic flow chart of an air-to-ground switching and control method for an amphibious drone with a variable configuration provided in an embodiment of the present application;

[0083] Figure 2 An example flow chart of an air-to-ground switching and control method for an amphibious drone with a variable configuration provided in an embodiment of the present application;

[0084] Figure 3 A schematic diagram of the angular motion between the pitch, roll, and yaw movements and the system is obtained by performing data solution on the attitude data provided in an embodiment of the present application;

[0085] Figure 4 A calibration curve diagram of an infrared ranging sensor provided in an embodiment of the present application;

[0086] Figure 5 A schematic diagram of a complementary filter provided in an embodiment of the present application adding accelerometer data after low-pass filtering and gyroscope data after high-pass filtering;

[0087] Figure 6 A schematic diagram of calculating an environment adaptation threshold D1 provided in an embodiment of the present application;

[0088] Figure 7 This is a diagram of the air-to-ground switching structure of a variable-configuration amphibious drone provided in an embodiment of the present application. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are merely examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0090] It will be appreciated that the terms "first", "second" (if any) etc. used herein may be used to describe various concepts herein, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0091] The terms "at least one", "plurality", "each", "any", etc. used in this application include "at least one", "two" or more, "plurality" or "each", "any" or "any one", "each" or "any one" as used herein.

[0092] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0093] Reference Figure 1 The embodiment of the present application provides an air-to-ground switching and control method for an amphibious drone with a variable configuration. The method may include but is not limited to S1 to S7, as follows:

[0094] S1: Calculate the attitude output data and height detection data based on the detection data after the drone sensor is initialized;

[0095] S2: fusing the attitude output data and the height detection data to obtain an optimal estimate of the drone angle;

[0096] S3: performing noise reduction based on the optimal estimated value of the drone angle to obtain a full-band, noise-free output signal;

[0097] S4: Based on the full-band, noise-free output signal, an output signal for real-time error control and total disturbance compensation is obtained through a cascade improved ADRC controller;

[0098] S5: Based on the output signals of the real-time error control and the total disturbance compensation, the UAV is subjected to a step-by-step landing flight control with dynamic error adjustment;

[0099] S6: Obtaining a judgment result based on a ground mode safety switch judgment using multiple conditions combined under landing flight control;

[0100] S7: Based on the judgment result, PID closed-loop control is adopted to coordinate the synchronous movement of the servo steering gear to obtain a servo steering gear control signal, and then the arm angle of the UAV is changed according to the control signal to control the UAV to perform air-to-ground switching.

[0101] Optionally, the step of calculating the attitude output data and the height detection data based on the detection data after initialization of the drone sensor comprises the following steps:

[0102] Obtaining the attitude output data by calculating the initialization detection data of the gyroscope, magnetometer, and accelerometer in the drone;

[0103] The initialized detection data is used to calculate and obtain two different digital distance values of the ultrasonic module and the infrared ranging sensor as the height detection data;

[0104] The ultrasonic module's calculation method converts the distance calculated based on the speed of sound into the original signal, and the air density change is compensated by the ambient temperature. The formula is:

[0105] v=331.4+0.6·T ℃ (m / s);

[0106] Among them, T ℃ is the current ambient temperature;

[0107] The infrared ranging sensor calculation method is based on converting the output analog voltage or digital signal into distance through a calibration curve, and the conversion formula established by linear fitting is:

[0108] d infrared =k·v+b;

[0109] Among them, k and b are calibration coefficients.

[0110] Optionally, fusing the attitude output data and the altitude detection data to obtain an optimal estimate of the drone angle comprises the following steps:

[0111] Optimizing the quaternion solution to calculate the attitude output data based on the gradient descent method so that the error between the predicted acceleration and the measured acceleration is minimized;

[0112] The method for solving the attitude output data is to construct a gravity vector model by calculating the three attitude angles obtained by the rotation matrix of the body coordinate system and the geographic coordinate system, which are the angles of rotation around the X, Y, and Z axes of the geographic coordinate system:

[0113]

[0114] The gravity vector in the geographic coordinates is g earth =[0,0,1], quaternion q = [q0,q1,q2,q3] rotated to the body coordinate system;

[0115] The objective function is used to obtain the partial derivative of each component of the quaternion based on the objective function to obtain the gradient of the objective function E with respect to the quaternion q, wherein the objective function is the sum of squares of the errors between the accelerometer measurement value and the theoretical gravity vector;

[0116] The gradient of the objective function E with respect to the quaternion q The calculation formula is:

[0117]

[0118] Among them, ex ,e y ,e z The error between the accelerometer measurement and the theoretical gravity on each axis is calculated by adjusting the quaternion and normalizing the gradient to the following formula:

[0119]

[0120] Optionally, the method further comprises the following steps:

[0121] The angular motion equations between the pitch, roll, and yaw motions and the system are obtained by solving the attitude output data:

[0122]

[0123] The θ, ψ are the three attitude angles of the UAV, namely the roll angle, pitch angle and yaw angle, which are the angular motions of the rotation around the X, Y and Z axes of the geographic coordinate system;

[0124] Based on the air-ground coordinate system conversion, the system motion control formula in the ground coordinate system is obtained as follows:

[0125]

[0126] are the accelerations on the x, y, and z axes in the system's ground coordinate system, and g is the gravity coefficient.

[0127] Optionally, the method further comprises the following steps:

[0128] Based on the Kalman filter, real-time sensor data is obtained through model prediction and sensor measurement to provide optimal state estimation by minimizing the covariance of the estimation process noise;

[0129] The prediction formula for the covariance of the prior estimation process noise is:

[0130]

[0131] The update formula for the covariance of the prior estimation process noise is:

[0132]

[0133] Among them, Q = 0.1 is the process noise variance, R = 0.2 is the measurement noise variance, K K is the Kalman gain, is the optimal estimate.

[0134] Optionally, the step of fusing the height detection data comprises the following steps:

[0135] Based on stage division and weight distribution, the advantages of ultrasound and infrared are combined to improve the robustness of height measurement, and the fusion formula is constructed as follows:

[0136] d funsion =w u ·d fultrasonic +w i ·d infrared ;

[0137] The ultrasonic weight is w u , the infrared weight is w i , dual sensor fusion is d funsion .

[0138] Optionally, performing noise reduction on the optimal estimated value based on the drone angle to obtain a full-band, noise-free output signal comprises the following steps:

[0139] The accelerometer is low-pass filtered to suppress high-frequency noise based on the first-order complementary filtering algorithm. The correction amount of the high-frequency noise and the integral amount of the low-frequency noise are expressed as follows:

[0140]

[0141] Dynamically adjust the trust coefficient α∈[0.01,0.1], where T is the unit conversion coefficient. The unit conversion coefficient is 1 when it is normally complementary and 100 when the accelerometer is converted from meters to centimeters.

[0142] The low-pass filter is obtained by dynamically fusing the attitude observation value and the fused distance value based on the first-order complementary filter, and the high-frequency noise of the sensor is suppressed by the second-order Butterworth low-pass filter. The cutoff frequency of the low-pass filter can be dynamically configured, including:

[0143] The formula for first-order complementary filtering is:

[0144] y k =(1-α)(x k-1 +B k x k );

[0145] Among them, y k is the final output value, x k is the sensor measurement value with high-frequency noise, B k is the differential of the sensor measurement value with high-frequency noise, dt is the data processing period;

[0146] The formula for suppressing high-frequency noise using a second-order Butterworth low-pass filter is:

[0147] y k =b0x k +b1x k-1 +b2x k-2-a1y k-1 -a2y k-2 ;

[0148] Among them, the coefficients b0, b1, b2, a1, and a2 are calculated by bilinear transformation and normalized to the cutoff frequency, and x k is the current input, y k is the current output, x k-1 is the input of the previous moment value, y k-1 The output of the previous moment value.

[0149] Optionally, obtaining an output signal for real-time error control and total disturbance compensation based on the full-band, noise-free output signal through a cascade improved ADRC controller comprises the following steps:

[0150] The input signal and output signal of the system are processed based on the tracking differentiator and the extended state observer respectively, and the nonlinear state error feedback control law controls the error between the processed signals and compensates for the total disturbance;

[0151] The dynamic equations of the tracking differentiator and extended state observer are:

[0152]

[0153] The observer dynamically adjusts the state estimate z through e(t) i (t), so that z1(t)→y1(t), z2(t) estimates the velocity, z3(t) estimates the acceleration and so on, with each order of dynamics being scaled by a parameter ε n-1 Adjust the nonlinear gain to make the cascade stable and estimate the total disturbance z n+1 (t) is expanded to additional states and the perturbation dynamics is passed through ε -1 Amplify the feedback gain to quickly track disturbance changes.

[0154] Optionally, the ground mode safety switching judgment based on multiple conditions combined under landing flight control to obtain a judgment result includes the following steps:

[0155] Monitor and judge the ground flatness based on the attitude output data of the drone's gyroscope, accelerometer, and magnetometer;

[0156] The composite inclination angle is calculated using the Euclidean norm θ x 2 +θ y 2 is the total tilt degree. If the composite tilt angle exceeds the threshold θ max , if it is determined that the ground is uneven, a switching prohibition signal is sent;

[0157] Ultrasonic sensors are installed on the left and right sides of the drone to measure the minimum distance between the two sides in real time. left ,d right There is enough space on both sides to expand. If the minimum value is less than d min If it is determined that there is insufficient space, a switching prohibition signal is sent.

[0158] Optionally, the step of using PID closed-loop control to coordinate the synchronous movement of the servo steering gear according to the judgment result to obtain a servo steering gear control signal comprises the following steps:

[0159] The control variable is generated based on the error between the target angle and the current angle, and is output through the discretized PID algorithm; the control variable output by the discretized PID algorithm is:

[0160]

[0161] Among them, μ[k] is the control quantity of the kth cycle, T s is the sampling cycle time, e[k] is the angle error of the kth cycle, and the angle error calculation formula is:

[0162] e(t)=θ d -θ t ;

[0163] Among them, θ d is the target angle of the retracted arm in land mode, θ t is the current angle fed back by the servo angle sensor;

[0164] Convert the PID control quantity into a servo duty cycle signal to coordinate the synchronous movement of multiple servos; the duty cycle calculation formula of the PWM conversion module is:

[0165] D[k]=clip(D[k-1]+K csale ·u[k],D min ,D max );

[0166] Among them, K scale is the scaling factor, and clip(x,a,b) means limiting x to the interval [a,b].

[0167] Next, the solution of the embodiment of the present application will be introduced and explained in detail with reference to specific application examples.

[0168] Specifically, this embodiment may include the following steps:

[0169] Obtain the detection data after the drone sensor is initialized and judge the safety of the air mode. The detection data after initialization is the basic data such as the attitude, speed, and altitude detected by the current drone sensor. The safety of the air mode is judged by detecting the attitude stability coefficient C θ , power system health coefficient C M , Deformation mechanism locking coefficient C S The safety judgment formula after logical multiplication and merging is determined as follows:

[0170] S safe4 =C θ +C M +C S ;

[0171] The attitude output data and height detection data are calculated based on the initialized detection data. The attitude output data is calculated using the three sensors of gyroscope, accelerometer and magnetometer. The height detection data is calculated using the ultrasonic module and infrared ranging sensor.

[0172] The attitude output data and height detection data are fused to obtain the optimal estimate of the drone angle and the fusion distance value. The optimal estimate of the drone angle is to estimate the attitude quaternion by minimizing the gravity direction error using the accelerometer measurement value. Combining model prediction and sensor measurement, the optimal state estimate is provided by minimizing the error.

[0173] Based on the optimal estimate of the drone's angle, noise reduction is performed to obtain a full-band, noise-free output signal. The full-band, noise-free output signal is obtained by adding the low-pass filtered accelerometer data and the high-pass filtered gyroscope data through a complementary filter.

[0174] The formula for adding accelerometer data and gyroscope data is:

[0175] x k =αA k +(1-α)(x k-1 +B k dt);

[0176] where x k is the final output value, A k is the sensor measurement value with high-frequency noise, B k is the differential of the sensor measurement value with high-frequency noise, dt is the data processing period, α is the filter coefficient and its value is between [0,1] to control the trust of high-frequency and low-frequency signals;

[0177] Based on the full-band, noise-free output signal, the output signal of real-time error control and total disturbance compensation is obtained through the cascade improved ADRC controller. The real-time estimation of system disturbance and error is achieved through the tracking differentiator, extended state observer and nonlinear state error feedback control law to achieve adaptive disturbance control of the system.

[0178] Based on the output signal of real-time error control and total disturbance compensation, a step-by-step landing flight control with dynamic error adjustment is adopted. The step-by-step landing flight control algorithm with dynamic error adjustment adopts a segmented uniform speed variable landing flight control algorithm to land in a step-by-step manner close to the landing point.

[0179] The entire flight is performed at a fixed speed. When the distance error between the drone and the landing point in the x or y axis is less than or equal to a threshold, the horizontal speed of the drone is zero. The mathematical expression is:

[0180]

[0181] V k Indicates the fixed speed of horizontal flight of the drone during landing, D1 indicates the landing allowable error distance threshold, Δ i (i=x,y) represents the absolute value of the difference between the real-time position of the drone (x,y,z) and the landing origin (0,0,0). The mathematical expression is:

[0182]

[0183] When the errors in the X and Y axis directions are both less than or equal to the allowable error, the drone slowly descends to the ground;

[0184] The ground mode safety switching judgment is based on the multi-condition combination under landing flight control; the multi-condition combination judgment is to realize the ground mode safety switching judgment by combining the ground flatness judgment and the obstacle distance safety margin judgment formula to ensure that the UAV can safely switch the arm shape after landing.

[0185] The ground mode safety switching judgment formula based on multiple conditions is:

[0186]

[0187] θ x ,θ y are the roll and pitch angles of the drone, θ max is the maximum allowable servo inclination angle threshold, d left ,d right is the shortest distance from the left and right arms of the drone to obstacles, d min The minimum lateral clearance required for the drone's arms to deploy;

[0188] The ground mode safety switching judgment based on multiple conditions is based on the use of PID closed-loop control to coordinate the synchronous movement of the servo servo to obtain the servo servo control signal; the servo servo is driven by multiple servo servos to drive the foldable arm to realize the switching of the UAV's air mode and land mode.

[0189] The servo control signal is to convert the PID output value into the target duty cycle. Suppose the servo angle range θ min ~θ max Corresponding PWM duty cycle range D min ~D max The conversion formula is:

[0190]

[0191] D[t] is the PWM duty cycle that controls the angle of the servo.

[0192] Obtain detection data after drone sensor initialization and determine the safety of the aerial mode, including:

[0193] The drone sensor data is repeatedly sampled multiple times at a sampling frequency greater than 100 Hz within the set delay time through a delay function and cross-validated with multiple sensors to ensure data accuracy;

[0194] The drone's attitude output data is obtained from the drone's gyroscope, magnetometer, and accelerometer, and the altitude detection data is obtained from the drone's barometer, infrared ranging sensor, and ultrasonic sensor.

[0195] Obtaining detection data after the drone sensor is initialized and determining the safety of the aerial mode, including:

[0196] The attitude stability coefficient C is constructed based on the current attitude angle initialized by the drone sensor θ Determine the rationality of the current attitude tilt and attitude stability coefficient C θ for:

[0197]

[0198] γ is the pitch angle, roll angle, and yaw angle, θ max is the preset maximum allowable attitude deviation;

[0199] Construct the power system health coefficient C based on the motor status M Detect the motor's response to the command, the power system health coefficient C M for:

[0200]

[0201] M j is the state of the jth motor, M j =1 is the normal output high level signal, M j =0 is the fault output low level signal, k is the total number of motors;

[0202] Deformation mechanism locking coefficient C based on servo motor encoder S The difference between the actual angle and the target angle is compared with the preset angle value to determine the angle error and the deformation mechanism locking coefficient C S for:

[0203]

[0204] α actual The actual rotation angle of the servo is fed back by the servo encoder, α target is the preset target locking angle value, Δα tol To allow angular error.

[0205] The posture output data is obtained based on the initial detection data, including:

[0206] The height detection data is solved to obtain two different digital distance values of the ultrasonic module and the infrared ranging sensor;

[0207] The ultrasonic solution method converts the distance calculated based on the speed of sound into the original signal, and compensates for the change in air density through the ambient temperature. The formula is:

[0208] v=331.4+0.6·T ℃ (m / s);

[0209] Among them, T ℃ is the current ambient temperature;

[0210] The infrared ranging sensor calculation method is based on converting the output analog voltage or digital signal into distance through the calibration curve, and the conversion formula is established through linear fitting:

[0211] d infrared =k·v+b;

[0212] Among them, k and b are calibration coefficients.

[0213] The posture output data is obtained based on the initialization detection data, and also includes:

[0214] The attitude data is solved by data to obtain the pitch, roll, yaw motion and the angular motion equation between the system:

[0215]

[0216] θ、 ψ are the three attitude angles of the UAV, namely the roll angle, pitch angle and yaw angle, which are the angular motions of the rotation around the X, Y and Z axes of the geographic coordinate system;

[0217] Based on the air-ground coordinate system conversion, the system motion control formula in the ground coordinate system is obtained as follows:

[0218]

[0219] are the accelerations on the x, y, and z axes in the system's ground coordinate system, and g is the gravity coefficient.

[0220] The optimal estimated value of the drone angle and the fused distance value are obtained by data fusion based on the attitude output data and the height detection data, including:

[0221] The quaternion-based attitude output data is optimized based on the gradient descent method to minimize the error between the predicted acceleration and the measured acceleration.

[0222] The solution is to calculate the three attitude angles obtained by the rotation matrix of the body coordinate system and the geographic coordinate system, which are the angles of rotation around the X, Y, and Z axes of the geographic coordinate system to construct a gravity vector model:

[0223]

[0224] The gravity vector in the Earth coordinate system is g earth =[0,0,1], quaternion q = [q0,q1,q2,q3] rotated to the body coordinate system;

[0225] Based on the objective function, the partial derivatives of each component of the quaternion are calculated to obtain the gradient of the objective function E with respect to the quaternion q. The objective function is the sum of squares of the errors between the accelerometer measurement value and the theoretical gravity vector.

[0226] Gradient of the objective function E with respect to the quaternion q The calculation formula is:

[0227]

[0228] e x ,e y ,e z The error between the accelerometer measurement and the theoretical gravity on each axis is calculated by adjusting the quaternion and normalizing the gradient to the following formula:

[0229]

[0230] The attitude output data is fused with the altitude detection data to obtain the optimal estimate of the drone angle and the fused distance value. It also includes:

[0231] Based on the Kalman filter, real-time sensor data is obtained through model prediction and sensor measurement to provide optimal state estimation by minimizing the covariance of the estimation process noise;

[0232] The prediction formula for the covariance of the prior estimation process noise is:

[0233]

[0234] The update formula for the covariance of the prior estimation process noise is:

[0235]

[0236] Q = 0.1 is the process noise variance, R = 0.2 is the measurement noise variance, K K is the Kalman gain, is the optimal estimate.

[0237] High-level detection data fusion also includes:

[0238] Based on stage division and weight distribution, the advantages of ultrasound and infrared are combined to improve the robustness of height measurement, and the fusion formula is constructed as follows:

[0239] d funsion =w u ·d fultrasonic +w i ·d infrared ;

[0240] The ultrasonic weight is w u , the infrared weight is w i , dual sensor fusion is d funsion .

[0241] Based on the optimal estimate of the drone's angle, noise reduction is performed to obtain a full-band, noise-free output signal, including:

[0242] Based on the first-order complementary filtering algorithm, the accelerometer low-pass filter suppresses high-frequency noise. The correction amount of high-frequency noise and the integral amount of low-frequency noise are expressed as follows:

[0243]

[0244] Dynamically adjust the trust coefficient α∈[0.01,0.1], where T is the unit conversion coefficient. The unit conversion coefficient is 1 when it is normally complementary and 100 when the accelerometer is converted from meters to centimeters.

[0245] The low-pass filter is obtained by dynamically fusing the attitude observation value and the fused distance value based on the first-order complementary filter. The high-frequency noise of the sensor is suppressed by the second-order Butterworth low-pass filter. The cutoff frequency can be dynamically configured, including:

[0246] The formula for first-order complementary filtering is:

[0247] y k =(1-α)(x k-1 +B k x k );

[0248] y k is the final output value, x k is the sensor measurement value with high-frequency noise, B k is the differential of the sensor measurement value with high-frequency noise, dt is the data processing period;

[0249] The formula for suppressing high-frequency noise using a second-order Butterworth low-pass filter is:

[0250] y k =b0x k +b1x k-1 +b2x k-2 -a1y k-1 -a2y k-2 ;

[0251] The coefficients b0, b1, b2, a1, a2 are calculated by bilinear transformation and normalized to the cutoff frequency, x k is the current input, y k is the current output, x k-1 is the input of the previous moment value, y k-1 The output of the previous moment value.

[0252] Based on the full-band, noise-free output signal, the cascaded improved ADRC controller is used to obtain the output signal for real-time error control and total disturbance compensation, including:

[0253] The input signal and output signal of the system are processed based on the tracking differentiator and the extended state observer respectively, and the nonlinear state error feedback control law controls the error between the processed signals and compensates for the total disturbance;

[0254] The dynamic equations of the tracking differentiator and extended state observer are:

[0255]

[0256] The observer dynamically adjusts the state estimate z through e(t) i (t) ensures that z1(t)→y1(t), z2(t) estimates the velocity, z3(t) estimates the acceleration and so on, with each order of dynamics being scaled by a scaling parameter ε n-1 Adjust nonlinear gains, ensure cascade stability, and estimate the total disturbance z n+1 (t) is expanded to additional states and the perturbation dynamics is passed through ε -1Amplify the feedback gain to quickly track disturbance changes.

[0257] The output signals of real-time error control and total disturbance compensation obtained by the cascade improved ADRC controller also include:

[0258] The selection of the tracking differentiator parameter ε depends on the actual response speed requirement;

[0259] The tracking differentiator passes z1(t)→y1(t), etc. can track the status of drones step by step;

[0260] Total disturbance z n+1 (t) include external disturbances and model uncertainty factors;

[0261] Nonlinear state error feedback control law u(t)=u0(t)-z n+1 (t), by compensating z n+1 (t) suppresses the influence of disturbance on the system, and u0(t) is the nominal control output value.

[0262] Dynamic error adjustment step-down flight control, including:

[0263] According to the real-time distance error Δ between the UAV and the landing point i With dynamic adjustment of horizontal speed V k , the specific rules are:

[0264]

[0265] D1 is the environmental adaptive threshold, and the calculation formula is D1=0.2·H+0.1·V wind , H is the current height, V wind The wind speed is set to ensure smoothness and safety of the landing process by adjusting the threshold.

[0266] The ground mode safety switching judgment based on multiple conditions combined under landing flight control includes:

[0267] Monitor and judge the ground flatness based on the attitude output data of the drone's gyroscope, accelerometer, and magnetometer;

[0268] The composite inclination angle is calculated using the Euclidean norm θ x 2 +θ y 2 is the total tilt degree. If the composite tilt angle exceeds the threshold θ max , it is determined that the ground is uneven and a switching prohibition signal is sent;

[0269] Ultrasonic sensors are installed on the left and right sides of the drone to measure the minimum distance between the two sides in real time. left,d right Ensure that there is enough space on both sides to expand. If the minimum value is less than d min It is determined that there is insufficient space and a switching prohibition signal is sent.

[0270] PID closed-loop control coordinates the synchronous movement of the servo actuator to obtain the servo actuator control signal, including:

[0271] The control variable is generated based on the error between the target angle and the current angle, and is output through the discretized PID algorithm. The output control variable of the discretized PID algorithm is:

[0272]

[0273] μ[k] is the control quantity of the kth cycle, T s is the sampling cycle time, e[k] is the angle error of the kth cycle, and the angle error calculation formula is:

[0274] e(t)=θ d -θ t ;

[0275] θ d is the target angle of the retracted arm in land mode, θ t is the current angle fed back by the servo angle sensor;

[0276] Convert the PID control quantity into a servo duty cycle signal to coordinate the synchronous movement of multiple servos; the duty cycle calculation formula of the PWM conversion module is:

[0277] D[k]=clip(D[k-1]+K csale ·u[k],D min ,D max );

[0278] where K scale is the scale factor, clip(x,a,b) means limiting x to the interval [a,b].

[0279] Next, the present application will be described in more detail. Figure 2 This embodiment provides an example flow chart of an air-to-ground switching and control method for an amphibious drone with a variable configuration. This embodiment includes the following steps:

[0280] Step S100: Obtain detection data after the drone sensor is initialized and determine the safety of the aerial mode.

[0281] The initialization state detection data is the basic data such as attitude, speed, and altitude detected by the current drone sensor, and the safety of the air mode is judged by detecting the attitude stability coefficient C θ, power system health coefficient C M , Deformation mechanism locking coefficient C S Sure.

[0282] Specifically, the drone is equipped with a gyroscope, an accelerometer, a magnetometer, an ultrasonic module, an infrared ranging sensor and a barometer. 2 The sensor is connected to the main control unit via a C or SPI interface. Initialization state detection data is performed by repeatedly sampling raw sensor data at a sampling frequency exceeding 100 Hz. A delay function is used to average multiple samples of the same sensor to reduce transient noise. Data accuracy is ensured through cross-validation of multiple sensors. Gyroscope and accelerometer data verify attitude angle consistency. Resampling is triggered when the difference between ultrasonic and infrared ranging sensor data exceeds a set threshold.

[0283] Furthermore, by constructing the detection attitude stability coefficient C θ , power system health coefficient C M , Deformation mechanism locking coefficient C S , the safety judgment formula is constructed by logical multiplication and merging to judge the safety of the current air mode. Among them, the current attitude angle constructs the attitude stability coefficient C θ Based on the UAV sensor to obtain the initialization of the current attitude angle to judge the rationality of the current attitude tilt to prevent the loss of control due to the initial attitude tilt, and then the power system health coefficient C M The motor's response to the command is detected according to the current state of the motor to ensure that all motors can respond to the control command normally, and the deformation mechanism locking coefficient C is constructed based on the servo actuator encoder. S The difference between the actual angle and the target angle is compared with the preset angle value to determine the angle error to ensure that the arm is deformed into place and the pneumatic structure is reliable.

[0284] For example, the attitude angle data can be detected to obtain the pitch angle φ, roll angle ψ, and yaw angle γ. The total number of motors j = 4 is in normal state, M1 = M2 = M3 = M4 = 1, and the target lock α is set. target and the permissible angle error Δα tol Based on this, the actual steering angle α of the servo fed back by the encoder can be obtained by detecting the deformation mechanism data. actual , the safety judgment formula can be

[0285] Step S101: Calculate and obtain attitude output data and height detection data based on the initialization detection data.

[0286] Specifically, attitude output data is calculated using a gyroscope, accelerometer, and magnetometer, while altitude detection data is calculated using an ultrasonic module and an infrared ranging sensor. Attitude data is calculated to determine pitch, roll, and yaw motions, as well as the angular motion between the system and the system.

[0287] In some embodiments of the present application, the process of obtaining the posture output data and the height detection data by calculating the above step S101 based on the initialization detection data is introduced.

[0288] like Figure 3 As shown, Figure 3 An embodiment of the present application provides a posture data that is obtained by data solution to obtain a schematic diagram of the angular motion between each motion of pitch, roll, and yaw and the system. The method may include the following steps:

[0289] The attitude data is solved to obtain the angular motion equations between the pitch, roll, and yaw movements and the system.

[0290] Specifically, the gyroscope can obtain three angles through integration, and the accelerometer and magnetometer are calculated through the rotation matrix of the body coordinate system and the geographic coordinate system to obtain the angular motion equations between the pitch, roll, and yaw movements and the system respectively. Through the air-ground coordinate system conversion, the UAV system's air motion control in the ground coordinate system can be obtained.

[0291] For example, pitching motion involves the drone rotating around its lateral axis, with the head tilting upward or downward, affecting vertical attitude stability and control. This is regulated by elevators or thrust distribution. Rolling motion involves the drone rotating around its longitudinal axis, with a sideways tilt affecting lateral balance and steering coordination. This is dynamically adjusted by ailerons or differential thrust. Yawing motion involves the drone rotating around its vertical axis, with horizontal steering determining the system's orientation and course correction, relying on rudder or differential torque control.

[0292] Furthermore, in the above step S101, after obtaining the angular motion equations between the pitch, roll, and yaw movements and the system, two different digital distance values of the ultrasonic module and the infrared ranging sensor are obtained through data solution.

[0293] Specifically, by transmitting sound waves and receiving echoes, the distance is calculated based on the time difference. The formula is: Dynamic correction is achieved by compensating for changes in air density based on ambient temperature. The infrared distance sensor calculation method is based on converting the output analog voltage or digital signal into distance through a calibration curve.

[0294] For example, Figure 4 This is a calibration curve diagram of an infrared ranging sensor provided in the embodiment of the present application, refer to Figure 4, we can get the mapping relationship between the pre-calibrated sensor digital signal output value ADC and the actual distance, and reversely solve the actual distance formula through the linear fitting equation:

[0295]

[0296] Step S102: fusing the attitude output data and the height detection data to obtain the optimal estimated value of the drone angle and the fused distance value.

[0297] Specifically, the system fuses gyroscope, accelerometer, and magnetometer data using the quaternion method. Accelerometer measurements, combined with model predictions and sensor measurements, minimize the gravity direction error to estimate the attitude quaternion. This attitude quaternion is then optimized using gradient descent to minimize the error and provide the optimal state estimate. The model prediction constructs a gravity vector model, resulting in an objective function that is the sum of the squared errors between the accelerometer measurements and the theoretical gravity vector. The quaternion is then iteratively updated using gradient descent.

[0298] After iteratively updating the quaternion through gradient descent, a Kalman filter is used to minimize the covariance of the estimated process noise to provide an optimal state estimate. This process noise covariance minimization provides the optimal state estimate, including a prediction formula for the prior estimated process noise covariance. The Kalman gain is then used to update the prior estimated process noise covariance to obtain the optimal estimate of the drone's angle. Finally, a fusion formula is constructed based on stage division and weight allocation to obtain the fused distance value.

[0299] Step S103: noise reduction is performed based on the optimal estimated value of the drone angle to obtain a full-band, noise-free output signal.

[0300] Specifically, the attitude observation value and the fused distance value are dynamically fused through the first-order complementary filter to obtain the filtered low-pass filter, and then the sensor high-frequency noise is suppressed through the second-order Butterworth low-pass filter, and the cutoff frequency can be dynamically configured. Finally, the low-pass filtered accelerometer data and the high-pass filtered gyroscope data are added through the complementary filter to obtain the output value.

[0301] For example, Figure 5 A schematic diagram of adding low-pass filtered accelerometer data and high-pass filtered gyroscope data by a complementary filter provided in an embodiment of the present application, with reference to Figure 5 , we can get x k is the final output value, A k is the sensor measurement value with high-frequency noise, B k is the differential of the sensor measurement value with high-frequency noise, dt is the data processing period, α is the filter coefficient and its value is between [0,1] to control the trust of high-frequency and low-frequency signals. The complementary filter adds the low-pass filtered accelerometer data and the high-pass filtered gyroscope data. The formula can be:

[0302] x k =αA k +(1-α)(x k-1 +B k dt).

[0303] Step S104: Based on the full-band, noise-free output signal, an output signal for real-time error control and total disturbance compensation is obtained through a cascade improved ADRC controller.

[0304] Specifically, the observer first dynamically adjusts the state estimate z through e(t) i (t) ensures that z1(t)→y1(t), z2(t) estimates the velocity, z3(t) estimates the acceleration and so on, with each order of dynamics being scaled by a scaling parameter ε n-1 Adjust nonlinear gains, ensure cascade stability, and estimate the total disturbance z n+1 (t) is expanded to additional states and the perturbation dynamics is passed through ε -1 Amplify the feedback gain to quickly track disturbance changes.

[0305] In one feasible implementation, assuming that the UAV position model is a second-order system, the error is defined as For real-time error control, the observer design can be:

[0306]

[0307] Where z3 is the estimated total disturbance, which is compensated to the control input value in real time as the disturbance compensation output signal.

[0308] The following describes an air-to-ground autonomous switching control method for an amphibious drone with a variable configuration provided in an embodiment of the present application. The air-to-ground autonomous switching control method for an amphibious drone with a variable configuration described below and the self-anti-interference method for an amphibious drone with a variable configuration described above can be referenced to each other.

[0309] Step S105: adopting a step-by-step landing flight control with dynamic error adjustment based on the output signal of the real-time error control and the total disturbance compensation.

[0310] Specifically, according to the real-time distance error Δ between the UAV and the landing point i With dynamic adjustment of horizontal speed V k By adjusting the environmental adaptation threshold, when the distance error between the drone and the landing point in the x or y axis is less than or equal to a threshold, the drone's horizontal speed is zero. Finally, when the errors in both the x and y axis are less than or equal to the allowable error, the drone slowly descends to the ground.

[0311] For example, Figure 6This is a schematic diagram of calculating an environmental adaptive threshold D1 provided in an embodiment of the present application. The height distance value is measured by a micro anemometer carried by a drone. The height and wind speed data are updated every 100ms and a safety judgment is made. The judgment is based on:

[0312]

[0313] Step S106: Ground mode safety switching judgment based on multiple conditions combined under landing flight control.

[0314] Specifically, the Euclidean norm θ is used x 2 +θ y 2 With the threshold value θ max A comparison is made to ensure that the landing tilt angle of the drone is within a safe range, and the minimum lateral space margin required for the arm to be deployed is compared with the shortest distances to obstacles on the left and right sides measured by the drone's ultrasonic sensor to ensure that the arms will not collide with obstacles when deployed to both sides.

[0315] For example, by placing the drone on a slope and gradually increasing the inclination angle until it rolls over, the limit inclination angle is obtained, and the safety factor is taken as the threshold, and the threshold θ is set max =Limiting inclination angle·Safety factor.

[0316] In addition, the maximum lateral width of the boom after deployment is measured and a safety margin is added. Based on this, the minimum lateral space margin d required for boom deployment is calculated. min The formula can be:

[0317] d min =L arm ×(1+k safe );

[0318] Step S107: Based on the ground mode safety switching judgment of multiple conditions, PID closed-loop control is adopted to coordinate the synchronous movement of the servo steering gear to obtain a servo steering gear control signal.

[0319] Specifically, the control quantity is generated by the discretized PID algorithm, which is converted into a PWM duty cycle signal according to the proportional factor. The strategy of unified target angle, independent control and limiting protection is adopted to achieve high-precision synchronous motion of multiple servos.

[0320] For example, Figure 7 The air-to-ground switching structure diagram of a variable configuration amphibious drone provided in the embodiment of the present application is shown in FIG. Figure 7The variable-configuration land-air amphibious drone uses two power systems to provide power for land mode and air mode respectively. To switch from air mode to land mode, the drone retracts the expansion plate through the servo steering gear, and then retracts the arm to make the land wheels touch the ground.

[0321] It can be seen from the above technical solution that the embodiment of the present application provides an air-to-ground switching method for an amphibious unmanned aerial vehicle with a variable configuration. Since the present application can use PID closed-loop control to coordinate the synchronous movement of the servo servo, the PID control quantity is converted into a servo duty cycle signal to change the tilt angle of the servo, and then the arm is retracted to make the land wheels touch the ground, thereby realizing the air-to-ground switching of the amphibious unmanned aerial vehicle with a variable configuration.

[0322] The target angle setting is the arm expansion angle in the air mode to θ d =90°, the drone can fly normally after unfolding, and the arm retracts to the angle θ in land mode d =0°, after retraction, the drone switches to ground movement state.

[0323] In the above embodiment, the PID closed-loop control is used to coordinate the synchronous movement of the servo steering gear. After the drone lands, the four arms need to be retracted from 90° to 0°, which can be adjusted by multiple control cycles. In this embodiment, the control process is described using steering gear No. 1 as an example, as follows:

[0324] By the current initial angle θ d =90°, target angle θ d =0°, the PWM duty cycle corresponding to 90° is D[0]=7.5%, and the error calculation is e(1)=0°-90°=-90°. Based on this, the PID output is The duty cycle is then obtained as D[1] = clip(7.5% + 0.01% -927.18, 5%, 10%) = 5%. In the subsequent control cycle, when the angle approaches the target value, such as θ = 5°, the error e(k) = -5°, the PID output decreases, the duty cycle adjustment amplitude decreases, and smooth convergence is achieved.

[0325] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0326] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0327] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0328] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0329] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. An air-to-ground switching and control method for an amphibious UAV with a variable configuration, characterized in that: The method comprises the following steps: The attitude output data and height detection data are obtained based on the detection data after the drone sensor is initialized; Fusing the attitude output data and the height detection data to obtain an optimal estimate of the drone angle; Noise reduction is performed based on the optimal estimate of the drone angle to obtain a full-band, noise-free output signal; Based on the full-band, noise-free output signal, an output signal for real-time error control and total disturbance compensation is obtained through a cascade improved ADRC controller; Based on the output signals of the real-time error control and total disturbance compensation, the UAV is subjected to a step-by-step landing flight control with dynamic error adjustment; Based on the ground mode safety switching judgment using multiple conditions combined under landing flight control, a judgment result is obtained; According to the judgment result, PID closed-loop control is adopted to coordinate the synchronous movement of the servo steering gear to obtain a servo steering gear control signal, and then the arm angle of the UAV is changed according to the control signal to control the UAV to perform air-to-ground switching.

2. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The method of calculating the attitude output data and the height detection data based on the detection data after the initialization of the drone sensor includes the following steps: Obtaining the attitude output data by calculating the initialization detection data of the gyroscope, magnetometer, and accelerometer in the drone; The initialized detection data is used to calculate and obtain two different digital distance values of the ultrasonic module and the infrared ranging sensor as the height detection data; The ultrasonic module's calculation method converts the distance calculated based on the speed of sound into the original signal, and the air density change is compensated by the ambient temperature. The formula is: v=331.4+0.6·T ℃ (m / s); Among them, T ℃ is the current ambient temperature; The infrared ranging sensor calculation method is based on converting the output analog voltage or digital signal into distance through a calibration curve, and the conversion formula established by linear fitting is: d infrared =k·v+b; Among them, k and b are calibration coefficients.

3. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The step of fusing the attitude output data and the height detection data to obtain an optimal estimate of the drone angle comprises the following steps: Optimizing the quaternion solution to calculate the attitude output data based on the gradient descent method so that the error between the predicted acceleration and the measured acceleration is minimized; The method for solving the attitude output data is to construct a gravity vector model by calculating the three attitude angles obtained by the rotation matrix of the body coordinate system and the geographic coordinate system, which are the angles of rotation around the X, Y, and Z axes of the geographic coordinate system: The gravity vector in the geographic coordinates is g earth =[0,0,1], quaternion q = [q0,q1,q2,q3] rotated to the body coordinate system; The objective function is used to obtain the partial derivative of each component of the quaternion based on the objective function to obtain the gradient of the objective function E with respect to the quaternion q, wherein the objective function is the sum of squares of the errors between the accelerometer measurement value and the theoretical gravity vector; The gradient of the objective function E with respect to the quaternion q The calculation formula is: Among them, e x ,e y ,e z The error between the accelerometer measurement and the theoretical gravity on each axis is calculated by adjusting the quaternion and normalizing the gradient to the following formula:

4. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The method further comprises the following steps: The angular motion equations between the pitch, roll, and yaw motions and the system are obtained by solving the attitude output data: The θ, ψ are the three attitude angles of the UAV, namely the roll angle, pitch angle and yaw angle, which are the angular motions of the rotation around the X, Y and Z axes of the geographic coordinate system; Based on the air-ground coordinate system conversion, the system motion control formula in the ground coordinate system is obtained as follows: are the accelerations on the x, y, and z axes in the system's ground coordinate system, and g is the gravity coefficient.

5. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 4 is characterized in that: The method further comprises the following steps: Based on the Kalman filter, real-time sensor data is obtained through model prediction and sensor measurement to provide optimal state estimation by minimizing the covariance of the estimation process noise; The prediction formula for the covariance of the prior estimation process noise is: The update formula for the covariance of the prior estimation process noise is: Among them, Q = 0.1 is the process noise variance, R = 0.2 is the measurement noise variance, K K is the Kalman gain, is the optimal estimate.

6. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 3 is characterized in that: The step of fusing the height detection data comprises the following steps: Based on stage division and weight distribution, the advantages of ultrasound and infrared are combined to improve the robustness of height measurement, and the fusion formula is constructed as follows: d funsion =w u ·d fultrasonic +w i ·d infrared ; The ultrasonic weight is w u , the infrared weight is w i , dual sensor fusion is d funsion .

7. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The method of performing noise reduction based on the optimal estimated value of the drone angle to obtain a full-band, noise-free output signal includes the following steps: The accelerometer is low-pass filtered to suppress high-frequency noise based on the first-order complementary filtering algorithm. The correction amount of the high-frequency noise and the integral amount of the low-frequency noise are expressed as follows: Dynamically adjust the trust coefficient α∈[0.01,0.1], where T is the unit conversion coefficient. The unit conversion coefficient is 1 when it is normally complementary and 100 when the accelerometer is converted from meters to centimeters. The low-pass filter is obtained by dynamically fusing the attitude observation value and the fused distance value based on the first-order complementary filter, and the high-frequency noise of the sensor is suppressed by the second-order Butterworth low-pass filter. The cutoff frequency of the low-pass filter can be dynamically configured, including: The formula for first-order complementary filtering is: y k =(1-a)(x) k-1 +B k x k ); Among them, y k is the final output value, x k is the sensor measurement value with high-frequency noise, B k is the differential of the sensor measurement value with high-frequency noise, dt is the data processing period; The formula for suppressing high-frequency noise using a second-order Butterworth low-pass filter is: y k =b0x k +b1x k-1 +b2x k-2 -a1y k-1 -a2y k-2 ; Among them, the coefficients b0, b1, b2, a1, and a2 are calculated by bilinear transformation and normalized to the cutoff frequency, and x k is the current input, y k is the current output, x k-1 is the input of the previous moment value, y k-1 The output of the previous moment value.

8. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The method of obtaining an output signal for real-time error control and total disturbance compensation based on the full-band, noise-free output signal through a cascade improved ADRC controller comprises the following steps: The input signal and output signal of the system are processed based on the tracking differentiator and the extended state observer respectively, and the nonlinear state error feedback control law controls the error between the processed signals and compensates for the total disturbance; The dynamic equations of the tracking differentiator and extended state observer are: The observer dynamically adjusts the state estimate z through e(t) i (t), so that z1(t)→y1(t), z2(t) estimates the velocity, z3(t) estimates the acceleration and so on, with each order of dynamics being scaled by a parameter ε n-1 Adjust the nonlinear gain to make the cascade stable and estimate the total disturbance z n+1 (t) is expanded to additional states and the perturbation dynamics is passed through ε -1 Amplify the feedback gain to quickly track disturbance changes.

9. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The method of determining a safe ground mode switch based on multiple conditions combined under landing flight control to obtain a determination result includes the following steps: Monitor and judge the ground flatness based on the attitude output data of the drone's gyroscope, accelerometer, and magnetometer; The composite inclination angle is calculated using the Euclidean norm θ x 2 +θ y 2 is the total tilt degree. If the composite tilt angle exceeds the threshold θ max , if it is determined that the ground is uneven, a switching prohibition signal is sent; Ultrasonic sensors are installed on the left and right sides of the drone to measure the minimum distance between the two sides in real time. left ,d right There is enough space on both sides to expand. If the minimum value is less than d min If it is determined that there is insufficient space, a switching prohibition signal is sent.

10. The air-to-ground switching and control method of the amphibious UAV based on a variable configuration according to claim 1 is characterized in that: The method of using PID closed-loop control to coordinate the synchronous movement of the servo steering gear according to the judgment result to obtain a servo steering gear control signal includes the following steps: The control variable is generated based on the error between the target angle and the current angle, and is output through the discretized PID algorithm; the control variable output by the discretized PID algorithm is: Among them, μ[k] is the control quantity of the kth cycle, T s is the sampling cycle time, e[k] is the angle error of the kth cycle, and the angle error calculation formula is: e(t)=θ d -θ t ; Among them, θ d is the target angle of the retracted arm in land mode, θ t is the current angle fed back by the servo angle sensor; Convert the PID control quantity into a servo duty cycle signal to coordinate the synchronous movement of multiple servos; the duty cycle calculation formula of the PWM conversion module is: D[k]=clip(D[k-1]+K scale ·u[k],D min ,D max ); Among them, K scale is the scale factor, clip(x,a,b) means limiting x to the interval [a,b].

Images