Improved adrc control quadrotor unmanned aerial vehicle precision landing method and system

Abstract

The application discloses a kind of improved ADRC control quadrotor unmanned plane precision landing method and system, the system includes quadrotor aircraft, flight controller, ultrasonic ranging sensor, vision sensor, GPS module, landing landmark, improved ADRC controller. Quadrotor aircraft carries flight controller, ultrasonic distance meter, vision sensor, GPS module, flight controller is built-in ardupilot firmware, ultrasonic ranging sensor is used to provide landing height information for flight controller, vision sensor uses Openmv4, Openmv4 obtains the relative coordinates of landing center by identifying landing landmark, flight controller receives the relative coordinates of landing center, and uses improved ADRC controller to control unmanned aerial vehicle position and attitude, ensure that quadrotor aircraft can still complete precision landing in the case where wind field disturbance, GPS signal exists interference and precision is limited.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) landing control technology, specifically to an improved ADRC control method and system for precise landing of a quadcopter UAV. Background Technology

[0002] In recent years, drone technology has developed rapidly. Quadcopter drones, with their simple mechanical structure and unique flight mode, have become a research hotspot in the drone field and are widely used in industrial and agricultural scenarios, such as quadcopter drones for power line inspection and pesticide spraying. However, landing accidents account for the largest proportion of drone accidents, therefore, research on precision landing technology for quadcopter drones has profound prospects.

[0003] Currently, quadcopter drones extensively utilize GPS (Global Positioning System) navigation for pinpoint landing. However, GPS is susceptible to interference from various environmental factors. For example, in GPS-denied areas, drones cannot receive satellite signals and lose their positioning. Therefore, relying solely on GPS cannot meet the application requirements in complex environments. Furthermore, GPS alone offers low landing accuracy, failing to meet the demands of precise landing. With the rapid development of computer vision, vision-based methods have demonstrated significant advantages in drone navigation. Currently, many researchers apply vision technology to achieve precise landing by using a host computer mounted on the drone. However, the onboard platform is heavy and consumes a lot of power, which is detrimental to the drone's endurance. Therefore, designing a method to achieve precise landing of quadcopter drones using only lightweight, low-power sensors is of great engineering significance and research value.

[0004] Precise landing of quadcopter drones requires stable flight control. Wind disturbances are common in the flight environment of quadcopter drones. Continuously changing wind disturbances generate additional interference forces and torques on the drone, affecting its attitude and horizontal position, and reducing the accuracy of its landing. The most widely used flight control algorithm in quadcopter drone landing is PID (Proportional-Integral-Differential) control and its improved methods. However, when wind disturbances are present in the drone landing environment, traditional PID control has limited disturbance rejection capabilities and cannot meet the high-precision landing requirements of quadcopter drones. Active Disturbance Rejection Control (ADRC) technology, proposed by researcher Han Jingqing, is an active disturbance rejection control method. Compared with traditional PID control, it has stronger disturbance rejection capabilities and robustness, making it more suitable for systems with significant external disturbances. However, ADRC controllers have numerous parameters, making parameter tuning difficult and limiting its practicality in actual quadcopter drone control. Professor Gao Zhiqiang transformed the ADRC controller into a LADRC (Linear Active Disturbance Rejection Control) controller through linearization and bandwidth concepts, significantly reducing controller parameters, but also noticeably lowering ADRC performance. In summary, traditional ADRC control presents a trade-off between parameter tuning complexity and control performance. Therefore, designing an improved ADRC controller to resolve this contradiction and applying it to the precision landing of quadcopter UAVs is promising and significant.

[0005] AprilTag is a type of visual reference library that uses fast edge detection to obtain the relative position of the camera and the AprilTag, and is often used as a visual guide landmark for precise drone landing. However, using a single AprilTag as a landing guide landmark often results in low landing accuracy, and may even lead to the loss of navigation information. Therefore, using a combination of landmarks for landing is particularly important. Summary of the Invention

[0006] To address the low accuracy of using a single AprilTag as a landing guidance landmark, this invention aims to propose an improved ADRC-controlled method and system for precise landing of quadcopter drones, solving problems such as insufficient landing accuracy and poor landing performance of quadcopter drones under wind disturbances.

[0007] To achieve the above objectives, the technical solution proposed by this invention is as follows:

[0008] An improved ADRC-controlled method for precise landing of a quadcopter UAV includes the following steps:

[0009] Step S1: In return-to-home mode, the quadcopter drone uses GPS positioning to fly to the landing landmark above the nested AprilTag;

[0010] Step S2: The quadcopter drone searches for the nested AprilTag landing landmarks using OpenMV. If the central AprilTag is identified, proceed to step S3; if only the auxiliary AprilTag is identified but the central AprilTag is not, the quadcopter drone is guided towards the central AprilTag using the relative coordinates provided by the auxiliary AprilTag until the central AprilTag is identified, then proceed to step S3; if within the set time threshold T... th If no AprilTag is found, the quadcopter drone will automatically switch to stationary mode and wait for the pilot to take over.

[0011] Step S3, when the quadcopter drone repeatedly identifies N th After the secondary center AprilTag is applied, the landing controller of the quadcopter drone is switched to the improved ADRC controller, and the quadcopter drone automatically switches to landing mode.

[0012] Step S4: The quadcopter drone performs altitude control based on the altitude information obtained by the ultrasonic ranging sensor; the quadcopter drone performs horizontal position control based on the relative coordinates parsed by the OpenMV recognition center AprilTag; when a nested AprilTag is recognized, the navigation information of the nested AprilTag is used to guide the quadcopter drone to land.

[0013] Step S5, when the descent altitude is lower than the set altitude threshold H th Afterwards, the quadcopter drone shut down its motors and landed on the landing platform.

[0014] As an optimized solution of the present invention, in step S3, the improved ADRC landing controller used by the quadcopter UAV is specifically as follows:

[0015] For attitude control of quadcopter UAVs, cascaded PID control is used in the yaw channel, that is, PID control is used in both the inner loop of yaw rate and the outer loop of yaw angle; PID-improved ADRC cascade control is used in both the roll and pitch channels. Taking the roll channel as an example, PID control is used in the inner loop of roll rate and improved ADRC control is used in the outer loop of roll angle.

[0016] For the position control of the quadcopter UAV, a cascaded improved ADRC control is used, that is, improved ADRC control is used for both the horizontal x-channel velocity inner loop and the horizontal x-channel position outer loop, as well as the horizontal y-channel velocity inner loop and the horizontal y-channel position outer loop. In addition, during the visual-guided landing of the quadcopter UAV, disturbances such as OpenMV positioning drift can cause abrupt changes in the input signal, leading to system overshoot. Therefore, a tracking differentiator (TD) is added to the input of the horizontal position outer loop to smooth the x and y horizontal position command signals and reduce overshoot.

[0017] As an optimized solution of the present invention, the improved ADRC control consists of a square root P feedback control law, a linear extended state observer (LESO), and a total disturbance low-pass filter. Specifically, the improved ADRC control is as follows:

[0018] Based on the UAV dynamics model, the UAV's horizontal position outer loop, horizontal velocity inner loop, and attitude angular velocity inner loop are all first-order systems. Therefore, the design process for the first-order improved ADRC controller is similar for the horizontal x-channel velocity inner loop, horizontal x-channel position outer loop, horizontal y-channel velocity inner loop, horizontal y-channel position outer loop, roll angular velocity inner loop, pitch angular velocity inner loop, and yaw angular velocity inner loop. The following design of a first-order improved ADRC controller is based on the roll angular velocity inner loop of the roll attitude channel as an example:

[0019] Based on the UAV attitude dynamics equations and the small angle assumption, the model of the inner loop of the roll angular velocity in the roll attitude channel can be obtained as follows:

[0020]

[0021] Where p, q, and r are the roll rate, pitch rate, and yaw rate, respectively; M φ The rolling torque is determined by the rotational speed of the four propellers; J xx J yy J zz The moment of inertia is the rotational inertia of the three axes;

[0022] Let the control quantity U = M φ If the perturbation term and coupling term in the model are f, then the above model is transformed into:

[0023]

[0024] Where b = 1 / J xx Let x1 = y = p, then we have:

[0025]

[0026] If the total disturbance is treated as a new state variable x2, then the state-space form of the system can be written as:

[0027]

[0028] Establish a second-order linear extended state observer (LESO):

[0029]

[0030] In the formula, z1 and z2 are the observed values ​​of states x1 and x2, respectively;

[0031] The total disturbance observation value z2 is low-pass filtered to reduce high-frequency noise from digital computation, improve system stability, and prevent the total disturbance observation value from being submerged in noise. The total disturbance observation value z2 after low-pass filtering is expressed as follows:

[0032] The feedback control rate is "square root P" control:

[0033]

[0034] Where u is the output control quantity, r is the setpoint, and k is the input control quantity. p Let k be the control gain of the "square root P" controller. Considering that the visual positioning signal of a quadcopter drone may exhibit abrupt changes, which can easily lead to significant overshoot and reduce landing accuracy, a controller design principle of "small gain for large errors, large gain for small errors" is adopted, where k is the control gain. p The design is partially linear and partially nonlinear, satisfying the following formula:

[0035]

[0036] In the formula, K p is the control gain for the linear segment, the same as P in a traditional PID controller; max is used to control the segmental threshold between the linear and nonlinear parts of the "square root P"; error is the error.

[0037] A precision landing system for a rotary-wing UAV based on improved ADRC control includes a quadcopter, a flight controller, an ultrasonic rangefinder, a vision sensor, a GPS module, landing landmarks, and an improved ADRC controller. The quadcopter is equipped with the flight controller, ultrasonic rangefinder, vision sensor, and GPS module. The flight controller has built-in Ardupilot firmware. The ultrasonic rangefinder provides landing altitude information to the flight controller. The vision sensor uses OpenMV4, which obtains the relative coordinates of the landing center by identifying the landing landmarks. The flight controller receives the relative coordinates of the landing center and uses the improved ADRC controller to control the UAV's position and attitude, ensuring that the quadcopter can still achieve precise landing even under conditions of wind disturbance, GPS signal interference, and limited accuracy.

[0038] As an optimized solution of the present invention, the landing landmark adopts a combined nested AprilTag landing landmark, specifically as follows:

[0039] The nested AprilTag landing marker consists of 6 AprilTags. A large AprilTag from the TAG16H5 family is placed in the center of the landing marker, called the central AprilTag. A small AprilTag from the TAG36H11 family is nested in the center of the central AprilTag, called the nested AprilTag. Four AprilTags from the TAG25H9 family of the same size but different IDs are placed around the central AprilTag at equal intervals, called the auxiliary guiding AprilTags.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1) This invention uses machine vision to achieve precise landing, solving the problems of GPS having rejection zones and insufficient positioning accuracy;

[0042] 2) This invention designs a combined nested AprilTag landing landmark, which overcomes the problem of quadcopter drones losing targets during landing due to narrowed field of view or external disturbances;

[0043] 3) This invention designs an improved ADRC controller, replacing the nonlinear state error feedback (NLSEF) control in the traditional ADRC with "square root P" control, and replacing the nonlinear extended state observer (ESO) with a linear extended state observer (LESO). The number of parameters that need to be tuned is significantly reduced. This controller is applied to the attitude and position control of quadrotor UAVs, resulting in better stability and higher landing accuracy of quadrotor UAVs in windy environments. Attached Figure Description

[0044] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0045] Figure 1 This is a flowchart of the landing method of the present invention.

[0046] Figure 2 This invention is a combined nested AprilTag landing landmark map.

[0047] Figure 3 This is a control structure diagram of the attitude control section in the improved ADRC landing controller designed in this invention.

[0048] Figure 4 This is a control structure diagram of the position control section in the improved ADRC landing controller designed in this invention.

[0049] Figure 5 This is a structural diagram of the improved ADRC controller designed in this invention.

[0050] Figure 6 This is a comparison of simulation results of the improved ADRC control scheme and the traditional PID control scheme designed in this invention, controlling a quadcopter UAV to track a spiral curve; Figure 6 (a) is a comparison chart of 3D simulation effects. Figure 6 (b) is a top view of the 3D simulation effect.

[0051] Figure 7 This is a comparison of simulation results of the improved ADRC control scheme and the traditional PID control scheme for controlling the landing of a quadcopter drone, as designed in this invention. Figure 7 (a) is a comparison chart of landing errors in the x-direction; Figure 7 (b) is a comparison chart of descent error in the y-direction; Figure 7 (c) is a comparison of the height tracking effect in the z-direction. Detailed Implementation

[0052] This invention discloses a precision landing system for a quadcopter drone based on improved ADRC control, comprising a quadcopter, a flight controller, an ultrasonic ranging sensor, a visual sensor, a GPS module, landing landmarks, and an active disturbance rejection controller. The quadcopter is equipped with the flight controller, ultrasonic ranging sensor, visual sensor, and GPS module. The flight controller has built-in Ardupilot firmware. The ultrasonic ranging sensor provides landing altitude information to the flight controller. The visual sensor uses OpenMV4, which obtains the relative coordinates of the landing center by identifying landing landmarks. The flight controller receives the relative coordinates of the landing center and uses the active disturbance rejection controller to control the drone's position and attitude, ensuring that the quadcopter can still achieve precise landing even under conditions of wind disturbance, GPS signal interference, and limited accuracy.

[0053] The flight controller uses Pixhawk 2.4.8, and the flight controller firmware is Ardupilot 4.0.7. The ultrasonic ranging sensor uses a GY-US42 ultrasonic module, which is compatible with the Pixhawk flight controller and Ardupilot firmware and is directly connected to the I / O pin of the Pixhawk 2.4.8 flight controller. 2 It can be used with the C interface; the Openmv4 module is connected to the telem1 of the Pixhawk 2.4.8 flight controller and communicates via MavLink serial port with a baud rate of 115200.

[0054] The landing landmark uses the nested AprilTag landing landmark designed in this invention, such as... Figure 2 As shown. The nested AprilTag landing marker consists of six AprilTags. A 30cm x 30cm AprilTag from the TAG16H5 family is placed in the center of the landing marker, called the central AprilTag; a 5cm x 5cm AprilTag from the TAG36H11 family is nested in the center of the central AprilTag, called the nested AprilTag; four AprilTags from the TAG25H11 family, each with a different ID number and a size of 15cm x 15cm, are placed around the central AprilTag, called the auxiliary guidance AprilTags. The AprilTag dimensions provided in this invention are for reference only and can be adjusted according to actual conditions.

[0055] Combination Figure 1 The landing method of the quadcopter UAV precision landing system based on improved ADRC control includes the following steps:

[0056] Step S1: In return-to-home mode, the quadcopter drone uses GPS positioning to fly to the landing landmark above the nested AprilTag;

[0057] In step S1, the landing landmark of the quadcopter drone is set as the home point through the Mission Planner ground station. During the normal flight of the quadcopter drone, the flight mode of the quadcopter drone can be switched to return-to-home mode through the remote controller. The aircraft will automatically maintain a certain return-to-home altitude and fly towards the home point under the guidance of GPS. The return-to-home altitude can be manually set in the ground station. If you want to switch the flight mode of the quadcopter drone to return-to-home mode through the ground station, you need to install a data transmission module on the flight control terminal and the PC terminal.

[0058] Step S2: The quadcopter drone searches for the nested AprilTag landing landmarks using OpenMV. If the central AprilTag is identified, proceed to step S3; if only the auxiliary AprilTag is identified but the central AprilTag is not, the quadcopter drone is guided towards the central AprilTag using the relative coordinates provided by the auxiliary AprilTag until the central AprilTag is identified, then proceed to step S3; if within the set time threshold T... th If no AprilTag is found, the quadcopter drone will automatically switch to stationary mode and wait for the pilot to take over.

[0059] In step S2, if the quadcopter drone only recognizes the auxiliary guide April tag but not the center April tag, then the quadcopter drone is guided towards the center April tag using the relative coordinates provided by the auxiliary guide April tag, as follows:

[0060] Using the known distance between the auxiliary guidance AprilTag and the central AprilTag, and the relative coordinates between the quadcopter and the auxiliary guidance AprilTag as resolved by OpenMV, the relative coordinates between the quadcopter and the central AprilTag are calculated; for example... Figure 2 As shown, the four auxiliary guide AprilTags have different ID numbers. When OpenMV recognizes an auxiliary guide AprilTag, it sends its ID number to the flight controller via serial port. The flight controller determines the "up", "down", "left", and "right" positions of the auxiliary guide AprilTag relative to the central AprilTag based on the ID number, and then controls the quadcopter drone to fly towards the central AprilTag based on the relative coordinates. If multiple auxiliary guide AprilTags are found in the OpenMV view, the average of the calculated relative coordinates is used as the actual relative coordinates, and these relative coordinates are used as control information to control the horizontal movement of the drone.

[0061] This invention uses the time threshold T thSet to 5 seconds, the timer starts from the moment OpenMV recognizes any AprilTag. If the flight controller does not receive AprilTag information from OpenMV via serial port after 5 seconds, it will automatically switch the quadcopter drone's flight mode to stationary mode and perform stationary control based on GPS signals, waiting for the pilot to take over.

[0062] Step S3, when the quadcopter drone repeatedly identifies N th After the secondary center AprilTag, the landing controller of the quadcopter drone is switched, changing the PID landing controller in the initial ardupilot4.0.7 firmware to the improved ADRC landing controller, and the flight mode of the quadcopter drone is switched to landing mode;

[0063] In step S3, the improved ADRC landing controller for the quadcopter UAV switching is as follows:

[0064] like Figure 3 As shown, φ, θ, and ψ represent the roll angle, pitch angle, and yaw angle, respectively. These represent the roll rate, pitch rate, and yaw rate, respectively. For the attitude control of a quadcopter UAV, the initial Ardupilot 4.0.7 firmware uses cascaded PID control in the yaw, roll, and pitch loops. After the switch, the quadcopter UAV still maintains cascaded PID control in the yaw loop, while using PID-improved ADRC cascade control in the roll and pitch loops. That is, PID control is used in the outer loop of the angle, and first-order improved ADRC control is used in the inner loop of the angular velocity.

[0065] like Figure 4 As shown, for the position control of a quadcopter UAV, the initial Ardupilot 4.0.7 firmware uses cascaded PID control in both the horizontal x and y loops. After the switch, the quadcopter UAV uses cascaded improved ADRC control in both the horizontal x and y loops, that is, the inner speed loop and outer position loop of the horizontal x channel, and the inner speed loop and outer position loop of the horizontal y channel all use first-order improved ADRC control. In addition, a second-order TD tracking differentiator is added to the input of the outer position loop to smooth the x and y horizontal position command signals, reduce the overshoot that may be caused by OpenMV positioning jumps in the horizontal position control of the quadcopter UAV, and thus improve landing accuracy. The TD tracking differentiator uses a second-order discrete form, and the equation is as follows:

[0066]

[0067] In the formula, v1 and v2 are the expected values ​​of the horizontal position and horizontal velocity of the quadcopter UAV, respectively; for example Figure 4As shown, v1 is the smoothed result of the x and y horizontal position command signals. v1 is used as the input to the improved ADRC controller for the x and y horizontal position loops; h is the controller's execution cycle; fhan is the fastest control synthesis function, with the expression:

[0068]

[0069] In the formula, r TD As a fast factor, r TD The larger the value, the faster the system response speed, but the larger the overshoot; h is the filter factor, the larger h is, the smaller the system static error, the larger the overshoot, and the worse the speed; the parameters in fhan need to be adjusted according to the actual test results;

[0070] like Figure 5 As shown, the improved ADRC control used in the attitude and position control of a quadcopter UAV consists of a square root P feedback control law, a linearly extended state observer (LESO), and a total disturbance low-pass filter, specifically:

[0071] Based on the UAV dynamics model, the UAV's horizontal position outer loop, horizontal velocity inner loop, and angular velocity inner loop are all first-order systems. Therefore, the design process for the first-order improved ADRC controller is similar for the horizontal x-channel velocity inner loop, horizontal x-channel position outer loop, horizontal y-channel velocity inner loop, horizontal y-channel position outer loop, roll angular velocity inner loop, pitch angular velocity inner loop, and yaw angular velocity inner loop. The following design of the first-order improved ADRC controller is based on the roll angular velocity inner loop of the roll attitude channel as an example:

[0072] Based on the UAV attitude dynamics equations and the small angle assumption, the model of the inner loop of the roll angular velocity in the roll attitude channel can be obtained as follows:

[0073]

[0074] Where p, q, and r are the roll rate, pitch rate, and yaw rate, respectively; M φ The rolling torque is determined by the rotational speed of the four propellers; J xx J yy J zz The moment of inertia is the rotational inertia of the three axes;

[0075] Let the control quantity U = M φ If the perturbation term and coupling term in the model are f, then the above model is transformed into:

[0076]

[0077] Where b = 1 / J xx Let x1 = y = p, then we have:

[0078]

[0079] If the total disturbance is treated as a new state variable x2, then the state-space form of the system can be written as:

[0080]

[0081] Establish a second-order linear extended state observer (LESO):

[0082]

[0083] In the formula, z1 and z2 are the observed values ​​of states x1 and x2, respectively;

[0084] like Figure 5 As shown in the dashed box ①, the total disturbance observation value z2 of the second-order linear extended state observer LESO is low-pass filtered to reduce high-frequency noise from digital computation, improve system stability, and prevent the total disturbance observation value from being submerged in noise. The total disturbance observation value z2 after low-pass filtering is expressed as:

[0085] like Figure 5 As shown in the dashed box ②, the feedback control rate is "square root P" control:

[0086]

[0087] Where u is the output control quantity, r is the setpoint, and k is the input control quantity. p Let k be the control gain of the "square root P" controller. Considering that the visual positioning signal of a quadcopter drone may exhibit abrupt changes, which can easily lead to significant overshoot and reduce landing accuracy, a controller design principle of "small gain for large errors, large gain for small errors" is adopted, where k is the control gain. p The design is partially linear and partially nonlinear, satisfying the following formula:

[0088]

[0089] In the formula, K p K represents the control gain for the linear segment, the same as P in a traditional PID controller; max is used to control the segmental thresholds between the linear and nonlinear parts of the "square root P"; due to differences in parameters and operating conditions among different quadcopter drones, K... p The maximum and minimum values ​​need to be adjusted according to the actual situation.

[0090] Compared to the traditional ADRC controller, the improved ADRC controller significantly reduces the number of parameters and lowers the difficulty of parameter tuning by replacing the extended state observer ESO with the linear extended state observer LESO and the nonlinear feedback controller NLSEF with the square root P feedback controller. This improves its engineering practicality. In addition, compared to the LADRC controller, the square root P feedback controller adopted by the improved ADRC controller has the control characteristics of "small gain with large error and large gain with small error", which can reduce overshoot under large error and ensure sensitivity under small error. Therefore, its control performance is better than that of the LADRC controller.

[0091] Step S4: The quadcopter drone performs altitude control based on the altitude information obtained by the ultrasonic ranging sensor; the quadcopter drone performs horizontal position control based on the relative coordinates parsed by the OpenMV recognition center AprilTag; when a nested AprilTag is recognized, the navigation information of the nested AprilTag is used to guide the quadcopter drone to land.

[0092] In step S4, when no nested AprilTag is detected, the present invention sets the landing speed of the quadcopter drone to 10cm / s, and when a nested AprilTag is detected, the landing speed is set to 5cm / s.

[0093] Step S5, when the descent altitude is lower than the set altitude threshold H th Afterwards, the quadcopter drone shut down its motors and landed on the landing platform.

[0094] In step S5, based on the Openmv4 sensor focal length, the size of the AprilTag, and experimental results, the present invention sets a height threshold H. th It is 10cm.

[0095] To verify the effectiveness of the improved ADRC landing controller, this invention conducted spiral trajectory tracking and landing simulation experiments in the MATLAB / Simulink environment and compared them with the traditional PID control scheme.

[0096] Under the assumption of small-angle motion of the quadrotor UAV, the six-degree-of-freedom nonlinear mathematical model of the quadrotor UAV adopted in this invention is as follows:

[0097]

[0098] Where U1, U2, U3, and U4 are control variables, and k x and k y d represents the drag coefficients of the quadcopter UAV in the x and y directions, respectively; x d y and d zThese represent the disturbances experienced by the quadcopter UAV in the x, y, and z directions, respectively, including wind field disturbances and disturbances generated by parts not modeled in the model. Their values ​​are:

[0099] d x =d y =d z

[0100] =0.045sin(2πt-3)+0.09sin(2πt+7)

[0101] +0.135sin(0.5πt-9.5)+0.09sin(0.3πt)

[0102] +0.054sin(0.15πt+4.5)+0.045sin(0.05πt+2)

[0103] +0.135sin(0.01πt+3)+0.225

[0104] The simulation parameters of the quadcopter UAV of the present invention are shown in the table below:

[0105] Table 1: Simulation Parameters for Quadrotor UAVs

[0106]

[0107] The following simulation experiment was conducted to track the spiral trajectory:

[0108] The three-dimensional spiral ascent curve is designed as a reference signal for tracking the spiral trajectory of a quadcopter UAV, and the formula is:

[0109]

[0110] Set the simulation step size to 0.001s and the simulation time to 20s. Add a perturbation d at the 6th second. x and d y The simulation results are as follows Figure 6 As shown. From Figure 6 As can be seen, under the influence of disturbance, the traditional PID control scheme has a large deviation in tracking the spiral trajectory, while the improved ADRC control scheme can track the spiral trajectory better after a short period of oscillation.

[0111] The following landing simulation experiment was conducted:

[0112] Set the initial coordinates of the drone (x,y,z) to (0,0,10m), the descent speed to a constant 0.5m / s, and the landing point coordinates to (0,0,0).

[0113] Set the simulation step size to 0.001s and the simulation time to 20s. Add a perturbation d at the 6th second.x and d y The simulation results are as follows Figure 7 As shown. From Figure 7 As can be seen, under the influence of disturbances, the traditional PID control scheme exhibits significant drift in both the x and y directions during the landing of the quadcopter drone, while the improved ADRC control scheme shows almost no drift and achieves high landing accuracy. In the z-axis altitude change curve, both the traditional PID control scheme and the improved ADRC control scheme can effectively control the drone to descend from an initial altitude of 10m to 0m. However, compared to the traditional PID control scheme, the improved ADRC control scheme exhibits no overshoot during landing, while the improved ADRC control scheme shows overshoot, resulting in a smoother landing process.

[0114] In practical implementation, the system hardware includes a Pixhawk 2.4.8 flight controller, a 450mm frame and matching landing gear, Langyu A2212 motors, Lotte 20A ESCs, 9450 self-locking propellers, an M8N GPS module, shock absorbers, OpenMV4 sensors, a Ledi remote controller and receiver, a GY-US42 ultrasonic ranging module, a data transmission module, and AprilTag16H5, AprilTag36H11, and AprilTag25H11 tags. Experimental verification shows that using the system and method provided by this invention, a quadcopter UAV can achieve a landing accuracy of less than 10cm under conditions of weak GPS, normal GPS signal, and minimal wind disturbance.

[0115] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An improved ADRC-controlled method for precise landing of a quadcopter UAV, characterized in that, Includes the following steps: Step S1: In return-to-home mode, the quadcopter drone uses GPS positioning to fly to the landing landmark above the nested AprilTag; Step S2: The quadcopter drone searches for the nested AprilTag landing landmarks using OpenMV. If the central AprilTag is identified, proceed to step S3; if only the auxiliary AprilTag is identified but the central AprilTag is not, the quadcopter drone is guided towards the central AprilTag using the relative coordinates provided by the auxiliary AprilTag until the central AprilTag is identified, then proceed to step S3; if within the set time threshold... If no AprilTag is found, the quadcopter drone will automatically switch to stationary mode and wait for the pilot to take over. Step S3, when the quadcopter drone repeatedly identifies After the secondary center AprilTag is applied, the landing controller of the quadcopter drone is switched to the improved ADRC controller, and the quadcopter drone automatically switches to landing mode. Step S4: The quadcopter drone acquires altitude information and performs altitude control; the quadcopter drone performs horizontal position control based on the relative coordinates parsed by the OpenMV recognition center AprilTag; when a nested AprilTag is recognized, the navigation information of the nested AprilTag is used to guide the quadcopter drone to land. Step S5, when the descent altitude is lower than the set altitude threshold Afterwards, the quadcopter drone shut down its motors and landed on the landing platform; The nested AprilTag landing markers consist of six AprilTags: a large AprilTag from the TAG16H5 family is placed in the center of the landing markers, called the central AprilTag; a small AprilTag from the TAG36H11 family is nested in the center of the central AprilTag, called the nested AprilTag; and four AprilTags from the TAG25H9 family of the same size but different IDs are placed at equal intervals around the central AprilTag, called the auxiliary guiding AprilTags. The improved ADRC controller includes the following control functions: The attitude control of the quadcopter UAV uses cascaded PID control in the yaw channel, that is, PID control is used in both the inner loop of yaw rate and the outer loop of yaw angle; PID-improved ADRC cascade control is used in both the roll channel and the pitch channel. The position control of the quadcopter drone uses cascaded improved ADRC control, i.e., horizontal... Channel speed inner loop and horizontal The outer ring of the passageway, and the horizontal Channel speed inner loop and horizontal The outer loop of the channel position uses improved ADRC control. During the visually guided landing of the quadcopter UAV, a tracking differentiator is added to the input of the horizontal position outer loop to smooth the landing. , The horizontal position command signal reduces overshoot.

2. The improved ADRC-controlled precision landing method for quadcopter UAVs according to claim 1, characterized in that, The improved ADRC controller is a first-order improved ADRC controller, which includes a square root P feedback control law, a linear extended state observer, and a total disturbance low-pass filter.

3. The improved ADRC-controlled precision landing method for quadcopter UAVs according to claim 2, characterized in that, The design of the first-order improved ADRC controller for the inner loop of the roll angular velocity in the roll attitude channel includes: Based on the UAV attitude dynamics equations and the small angle assumption, a model of the inner loop of the roll angular velocity of the roll attitude channel and the state-space equations of the system are established. Design a second-order linear extended state observer LESO based on the state-space equations; The total disturbance observations are processed using a total disturbance low-pass filter. The total disturbance observation after low-pass filtering. Represented as ; The design feedback control rate is the square root P control.

4. The improved ADRC-controlled precision landing method for quadcopter UAVs according to claim 3, characterized in that, The model for the inner loop of the roll angular velocity of the roll attitude channel, established based on the UAV attitude dynamics equations and the small angle assumption, specifically includes: The model for the inner loop of the roll angular velocity in the roll attitude channel is as follows: in These are roll rate, pitch rate, and yaw rate, respectively. The rolling torque is determined by the rotational speed of the four propellers; The moment of inertia is the rotational inertia of the three axes; Set control quantity The perturbation and coupling terms in the model are Then the model is transformed into: in ;make Then we have: The total disturbance is treated as a new state variable. The state-space equation of the system is: 。 5. The improved ADRC-controlled precision landing method for quadcopter UAVs according to claim 4, characterized in that, The second-order linear extended state observer LESO is: In the formula They are respectively states The observed values.

6. The improved ADRC-controlled precision landing method for quadcopter UAVs according to claim 3, characterized in that, The design feedback control rate is: in The output control quantity, For a given value, The control gain of the square root P controller is given.

7. The improved ADRC-controlled precision landing method for quadcopter UAVs according to claim 6, characterized in that, The control gain for: In the formula, The control gain for the linear segment; The segmentation threshold used to control the linear and nonlinear parts of the square root P, where error is the error value.

8. A precision landing system for a rotary-wing UAV employing the improved ADRC control precision landing method for a quadcopter UAV as described in any one of claims 1 to 7, characterized in that, The system includes a quadcopter, a flight controller, an ultrasonic ranging sensor, a vision sensor, a GPS module, landing markers, and an improved ADRC controller. The quadcopter is equipped with the flight controller, ultrasonic ranging sensor, vision sensor, and GPS module. The flight controller has built-in Ardupilot firmware. The ultrasonic ranging sensor provides landing altitude information to the flight controller. The vision sensor uses OpenMV4, which employs a nested AprilTag landing marker. By identifying the landing marker, the relative coordinates of the landing center are obtained. The flight controller receives the relative coordinates of the landing center and uses the improved ADRC controller to control the position and attitude of the drone, enabling the quadcopter to complete a precise landing.

Images