Unmanned aerial vehicle mechanical arm cooperative control system with dynamic sliding rail compensation

Abstract

The invention provides an unmanned aerial vehicle mechanical arm cooperative control system with dynamic sliding rail compensation, which relates to the technical field of unmanned aerial vehicle control and comprises a multi-source data acquisition module, a data fusion processing module, a dynamic pose matrix updating module and a sliding rail compensation control module. The multi-source data acquisition module is used for synchronously acquiring sensor data of an IMU (Inertial Measurement Unit), a visual odometer and a laser radar of the unmanned aerial vehicle; the data fusion processing module is used for fusing sensor data based on an extended Kalman filtering algorithm and outputting pose estimation of the unmanned aerial vehicle under a unified coordinate system; the dynamic pose matrix updating module is used for updating a target pose matrix in real time according to pose estimation and a target pose and calculating a pose error; according to the invention, fusion of IMU, visual odometer and laser radar data is realized through extended Kalman filtering, advantages of different sensors are complemented, perception precision and response speed of pose change of the unmanned aerial vehicle are improved, and a data basis is provided for compensation control.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a UAV manipulator arm collaborative control system with dynamic slide rail compensation. Background Art

[0002] In a UAV-mounted robotic arm operation system, the UAV's posture stability directly affects the positioning accuracy of the robotic arm's end effector. Existing UAVs have the following deficiencies:

[0003] In traditional systems, the data from sensors such as IMU, visual odometry, and lidar are not deeply integrated. The problems of IMU's cumulative error, visual sensitivity to light, and low lidar update frequency lead to insufficient posture perception accuracy and inability to quickly respond to the dynamic changes of the drone. In traditional technology, when only relying on the robot arm's own joint adjustment to compensate for the drone's posture error, there are problems of limited compensation range and slow response speed. In particular, when the drone is suddenly offset by external interference such as wind and airflow, the positioning accuracy of the robot arm end is significantly reduced. In traditional control strategies, a real-time updated target posture matrix is ​​not established, and the robot arm base coordinates cannot be dynamically adjusted according to the actual posture of the drone, resulting in a disconnect between the robot arm motion planning and the actual spatial relationship, making it difficult to meet high-precision operation requirements. Therefore, the present invention proposes a drone robot arm collaborative control system with dynamic slide rail compensation to solve the problems existing in the prior art. Summary of the Invention

[0004] In response to the above problems, the present invention proposes a UAV robotic arm collaborative control system with dynamic slide rail compensation. The UAV robotic arm collaborative control system with dynamic slide rail compensation realizes the fusion of IMU, visual odometer, and lidar data through extended Kalman filtering, complements the advantages of different sensors, improves the perception accuracy and response speed of UAV posture changes, and provides a reliable data basis for compensation control.

[0005] To achieve the objectives of the present invention, the present invention is implemented through the following technical solutions: a UAV manipulator collaborative control system with dynamic slide compensation, comprising a multi-source data acquisition module, a data fusion processing module, a dynamic pose matrix update module, and a slide compensation control module. The multi-source data acquisition module is used to synchronously collect sensor data from the UAV's IMU, visual odometer, and lidar; the data fusion processing module is used to fuse the sensor data based on the extended Kalman filter algorithm and output the UAV's pose estimate in a unified coordinate system;

[0006] The dynamic pose matrix update module is used to update the target pose matrix in real time according to the pose estimation and target pose, and calculate the pose error; the slide rail compensation control module is used to drive the slide rail to move through a closed-loop control strategy to compensate for the change in the drone's pose when the pose error exceeds a preset threshold.

[0007] Further improvements are: in the multi-source data acquisition module, the IMU collects acceleration and angular velocity data at a frequency greater than 100 Hz, the visual odometer collects environmental feature data at a frequency of 10-30 Hz, and the lidar collects three-dimensional point cloud data.

[0008] A further improvement is that the data fusion processing module adopts an extended Kalman filter algorithm, including:

[0009] Prediction stage: Calculate the predicted state through the kinematic equation:

[0010]

[0011] And propagate the covariance:

[0012] P k∣k-1 =F k P k-1∣k-1 F k T +Q k

[0013] in is the predicted state at time k, is the optimal state at time k-1, u k is the IMU input, F k is the state transfer matrix, Q k is the process noise covariance matrix;

[0014] Update phase: For visual odometry, calculate the new information:

[0015]

[0016] Calculate the Kalman gain:

[0017] K vis,k =P k∣k-1 H T vis,k (H vis,k P k∣k-1 H T vis,k +R vis,k ) -1

[0018] Calculate the updated status:

[0019]

[0020] Calculate the covariance:

[0021] P k∣k =(IK vis,k H vis,k )P k∣k-1

[0022] Among them, z vis,k is the visual observation value, h vis is the visual observation function, H vis,k is the visual observation matrix, R vis,k is the visual measurement noise covariance matrix;

[0023] For lidar, the same process is followed, using the corresponding measurement matrix and noise covariance.

[0024] A further improvement is that in the dynamic pose matrix update module, the target pose matrix is ​​expressed as:

[0025]

[0026] in, is the pose transformation matrix of the UAV in the world coordinate system, is the pose transformation matrix of the target relative to the UAV;

[0027] When the position error exceeds 1 mm or the attitude error exceeds 0.5 degrees, the slide compensation mechanism is triggered.

[0028] A further improvement is that the slide rail compensation control module adopts PID closed-loop control, and the control law is:

[0029]

[0030] Among them, u(k) is the control quantity, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(k) is the current position error, and Δt is the sampling time interval.

[0031] A further improvement is that the slide rail compensation control module obtains slide rail position feedback through a magnetic scale, the magnetic scale resolution is 0.1 micron, and the position calculation formula is:

[0032] Position=RawData×0.1μm / LSB;

[0033] Among them, RawData is the original 24-bit data of the magnetic scale.

[0034] A further improvement is that in the slide rail compensation control module, the linear drive unit drives the slide rail to move according to the target displacement, forms a closed loop through magnetic scale feedback, and adjusts the motor output in real time to ensure the accuracy of the slide rail movement.

[0035] A further improvement is that data synchronization is achieved between the multi-source data acquisition module, the data fusion processing module and the dynamic posture matrix update module through the ROS robot operating system, and the communication cycle is ≤10ms.

[0036] A further improvement is that it also includes a robot base coordinate system update module for updating the robot base coordinate system according to the actual displacement of the slide rail, so that the robot arm motion planning is based on the latest spatial relationship.

[0037] Further improvement is to include the following collaborative control steps:

[0038] Synchronously collect sensor data from IMU, visual odometry, and lidar;

[0039] The extended Kalman filter algorithm is used to fuse sensor data to obtain the UAV pose estimation;

[0040] Update the target pose matrix based on the pose estimate and the target pose, and calculate the pose error;

[0041] When the posture error exceeds the threshold, the slide rail is driven to move through PID closed-loop control to compensate for the posture change.

[0042] The beneficial effects of the present invention are:

[0043] 1. The present invention realizes the fusion of IMU, visual odometer and lidar data through extended Kalman filtering, complements the advantages of different sensors, improves the perception accuracy and response speed of drone posture changes, and provides a reliable data basis for compensation control.

[0044] 2. The present invention uses the slide rail to adjust the position of the robotic arm base in real time according to the drone's posture, dynamically offsetting the drone's offset. Compared with traditional joint compensation methods, it reduces the burden on the robotic arm, extends its service life, and achieves more precise end positioning.

[0045] 3. The present invention combines PID closed-loop control with magnetic scale feedback to achieve millimeter-level precise movement of the slide rail. Through the synergistic effect of proportional, integral and differential terms, it quickly suppresses errors, ensures that the robotic arm is always in the optimal operating position, expands the operating range and improves efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is the front view of the present invention. DETAILED DESCRIPTION

[0047] In order to deepen the understanding of the present invention, the present invention will be further described in detail below with reference to the examples. The examples are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0048] Example 1

[0049] according to Figure 1 As shown, this embodiment proposes a UAV manipulator collaborative control system with dynamic slide compensation, including a multi-source data acquisition module, a data fusion processing module, a dynamic pose matrix update module and a slide compensation control module. The multi-source data acquisition module is used to synchronously collect sensor data from the UAV's IMU, visual odometer and lidar; the data fusion processing module is used to fuse the sensor data based on the extended Kalman filter algorithm and output the UAV's pose estimation in a unified coordinate system;

[0050] The dynamic pose matrix update module is used to update the target pose matrix in real time according to the pose estimation and target pose, and calculate the pose error; the slide rail compensation control module is used to drive the slide rail to move through a closed-loop control strategy to compensate for the change in the drone's pose when the pose error exceeds a preset threshold.

[0051] High-precision posture perception is achieved through multi-source data fusion, and control targets are adjusted in real time based on a dynamic posture matrix update mechanism. Dynamic adjustment of the manipulator base is achieved using a rail compensation control strategy. The logical relationship between the various modules of the system is as follows: the multi-source data acquisition module provides raw sensor information to the data fusion processing module. The high-frequency dynamic perception capability of the IMU and the absolute positioning accuracy of the lidar are complemented by an extended Kalman filter. The output of the data fusion processing module serves as the input of the dynamic posture matrix update module. By calculating the difference between the current posture and the target posture in real time, it provides a precise control target for rail compensation. The rail compensation control module triggers closed-loop control based on the posture error. The combination of high-precision feedback from the magnetic scale and the PID algorithm ensures the accuracy of the rail movement. At the same time, the manipulator base coordinate system update module updates the motion planning benchmark based on the rail displacement, forming a complete control closed loop.

[0052] In the multi-source data acquisition module, the IMU collects acceleration and angular velocity data at a frequency greater than 100Hz, the visual odometer collects environmental feature data at a frequency of 10-30Hz, and the lidar collects three-dimensional point cloud data. The data fusion processing module adopts the extended Kalman filter algorithm, including:

[0053] Prediction stage: Calculate the predicted state through the kinematic equation:

[0054]

[0055] And propagate the covariance:

[0056] P k∣k-1 =F k P k-1∣k-1 F k T +Q k

[0057] in is the predicted state at time k, is the optimal state at time k-1, u k is the IMU input, F k is the state transfer matrix, Q k is the process noise covariance matrix;

[0058] Update phase: For visual odometry, calculate the new information:

[0059]

[0060] Calculate the Kalman gain:

[0061] K vis,k =P k∣k-1 H T vis,k (H vis,k P k∣k-1 H T vis,k +R vis,k ) -1

[0062] Calculate the updated status:

[0063]

[0064] Calculate the covariance:

[0065] P k∣k =(IK vis,k H vis,k )P k∣k-1

[0066] Among them, z vis,k is the visual observation value, h vis is the visual observation function, H vis,k is the visual observation matrix, R vis,k is the visual measurement noise covariance matrix;

[0067] For lidar, the same process is followed, using the corresponding measurement matrix and noise covariance.

[0068] The multi-source data acquisition module, the data fusion processing module and the dynamic posture matrix update module are synchronized with each other through the ROS robot operating system, and the communication cycle is ≤10ms.

[0069] In the dynamic pose matrix update module, the target pose matrix is ​​expressed as:

[0070]

[0071] in, is the pose transformation matrix of the UAV in the world coordinate system, is the pose transformation matrix of the target relative to the UAV;

[0072] When the position error exceeds 1 mm or the attitude error exceeds 0.5 degrees, the slide compensation mechanism is triggered.

[0073] The slide rail compensation control module adopts PID closed-loop control, and the control law is:

[0074]

[0075] Among them, u(k) is the control quantity, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(k) is the current position error, and Δt is the sampling time interval.

[0076] The slide rail compensation control module obtains slide rail position feedback through the magnetic scale. The resolution of the magnetic scale is 0.1 micron. The position calculation formula is:

[0077] Position=RawData×0.1μm / LSB;

[0078] Among them, RawData is the original 24-bit data of the magnetic scale.

[0079] In the slide rail compensation control module, the linear drive unit drives the slide rail to move according to the target displacement, forms a closed loop through magnetic scale feedback, and adjusts the motor output in real time to ensure the accuracy of the slide rail movement.

[0080] It also includes a robot base coordinate system update module, which is used to update the robot base coordinate system according to the actual displacement of the slide rail, so that the robot arm motion planning is based on the latest spatial relationship.

[0081] The collaborative control process includes the following steps: synchronously collecting sensor data from the IMU, visual odometry, and lidar; fusing the sensor data using the extended Kalman filter algorithm to obtain the drone's pose estimate; updating the target pose matrix based on the pose estimate and the target pose, and calculating the pose error; and when the pose error exceeds a threshold, driving the slide to move through PID closed-loop control to compensate for the pose change.

[0082] Extended Kalman Filter (EKF):

[0083] State vector: x = [p, v, q, b_a, b_g] (position, velocity, attitude quaternion, accelerometer bias, gyroscope drift);

[0084] Observation vector: z_vis = [Δp_vis, Δθ_vis] (visual relative displacement and angle), z_lidar = [x_lidar, y_lidar, z_lidar] (radar absolute coordinates);

[0085] State transition matrix: Contains the Jacobian term of the differential of the rotation matrix R(q) with respect to the attitude.

[0086] By fusing complementary data from IMU (high-frequency dynamic response), vision (relative posture), and lidar (absolute positioning) through EKF, the defects of a single sensor in terms of error accumulation, light sensitivity, and frequency limitations can be solved.

[0087] Data flow: IMU → Prediction phase → Covariance update → Vision / Radar → Update phase → Fusion pose → Pose matrix update → Triggering rail compensation → PID control → Magnetic scale feedback. When LiDAR data is lost, the system automatically switches to IMU + Vision tightly coupled mode, maintaining control stability through residual error monitoring. Dynamic operation process: UAV flies to the target area → Multi-source sensor initialization → Establish global coordinate system → Robotic arm executes the task → Pose error triggers rail compensation → Task completion.

[0088] Example 2

[0089] according to Figure 1 As shown, this embodiment proposes a UAV manipulator collaborative control system with dynamic slide rail compensation to address the impact of UAV posture deviation on the manipulator operation accuracy.

[0090] Multi-source data deep fusion technology is used to deeply fuse data from multiple sensors such as IMU, visual odometry, and lidar. IMU can quickly provide attitude and acceleration information, but there are cumulative errors; visual odometry obtains relative position information through image processing and is sensitive to light and texture; lidar can accurately measure distance, and the data update frequency is relatively low. This system uses extended Kalman filtering to linearize nonlinear systems and achieves data fusion through two steps: prediction and update. The prediction step predicts the current state and covariance based on the system model and the state estimate at the previous moment; the update step uses the current moment observation value to correct the predicted state to obtain a more accurate state estimate, thereby accurately unifying the data from different sensors into the same coordinate system, providing a reliable and comprehensive data basis for subsequent compensation control, and effectively improving the system's perception accuracy and response speed to changes in the drone's posture. The specific implementation steps are as follows:

[0091] First, data is collected synchronously using three sensors: an inertial measurement unit (IMU), a visual odometry system, and a lidar (lidar). The IMU provides the robot's acceleration and angular velocity information at a high frequency (>100Hz), enabling rapid perception of dynamic motion. The visual odometry captures environmental features at a lower frequency (10-30Hz), providing a relative displacement reference between adjacent frames. The lidar provides the robot's position in an absolute coordinate system through sparse but precise 3D point cloud data.

[0092] Next, the system uses the Extended Kalman Filter (EKF) algorithm to fuse these data, which is divided into two steps:

[0093] Prediction stage:

[0094] Based on the acceleration and angular velocity data from the IMU, the system calculates the robot's position, velocity, and posture at the next moment using kinematic equations (such as the Newton-Euler formula). The specific formula is:

[0095]

[0096] Where {p}{v}{θ} represent position, velocity, and attitude angle, respectively; acceleration a and angular velocity ω are the measurements of the IMU; and Δt is the sampling time interval.

[0097] At the same time, the system predicts the uncertainty of the state (covariance matrix) based on the noise characteristics of the IMU (such as the zero bias of the accelerometer and the drift of the gyroscope), providing a basis for subsequent corrections. The covariance propagation formula is: Pk|k-1=FPk-1FT+Q

[0098] Where F is the state transfer matrix (including the rotation Jacobian matrix), and Q is the process noise covariance matrix.

[0099] Update phase:

[0100] Visual correction: Compare the relative displacement measured by the visual odometry with the predicted result, calculate the difference between the two (innovation), adjust the predicted value through the Kalman gain (weight factor), and correct the accumulated error. The specific formula is:

[0101]

[0102] Among them, H vo is the visual observation matrix, R vo is the visual observation noise covariance.

[0103] LiDAR correction: The absolute position data from the LiDAR is used to further calibrate the robot's global coordinates to ensure positioning accuracy. The LiDAR correction process is similar to the visual correction process, but uses a different observation matrix Hlidar and noise covariance Rlidar.

[0104] Dynamic target pose matrix update mechanism: Based on the fused data, a dynamic target pose matrix update mechanism is designed. This mechanism calculates the current pose of the UAV in the world coordinate system in real time, and combines it with the target pose required by the mission to quickly and accurately update the target pose matrix. Through this dynamic update method, the system can always adjust the control strategy according to the actual pose of the UAV, providing precise target guidance for rail compensation. The specific implementation steps are as follows: The system calculates the difference between the current estimated pose x^ and the target pose xtarget in real time. If the position error exceeds 1 mm or the attitude error exceeds 0.5 degrees, the rail compensation mechanism is triggered. During the rail compensation process, the system updates the base coordinate system of the robot arm according to the actual displacement of the rail, ensuring that the robot arm's motion planning is always based on the latest spatial relationship.

[0105] High-precision slide rail compensation control strategy: A closed-loop control strategy is used for dynamic compensation of the slide rail. When it is detected that the drone position error exceeds the preset threshold, the linear drive unit quickly pushes the slide rail to move according to the target displacement. At the same time, high-precision sensors such as magnetic scales collect the slide rail position feedback information in real time and accurately compare it with the target displacement. The PID control algorithm is used to fine-tune the output of the linear motor. The proportional term responds quickly according to the size of the current error, the integral term is used to eliminate steady-state errors, and the differential term suppresses the error change in advance according to the rate of change of the error, ensuring that the slide rail can move to the target position accurately and stably, achieving efficient and accurate dynamic compensation and improving the positioning accuracy of the end effector of the robotic arm.

[0106] The specific steps are as follows: Finally, through the PID control algorithm and magnetic scale feedback, the system achieves high-precision control of the slide rail:

[0107] The PID algorithm calculates the control quantity (such as motor driving force) based on the posture error. The proportional term (P) quickly responds to the error, the integral term (I) eliminates long-term deviations, and the differential term (D) suppresses overshoot. The control law is:

[0108]

[0109] The magnetic scale measures the slide position in real time with a resolution of 0.1 micron, forming a closed-loop feedback loop to ensure accurate slide movement. The closed-loop control process includes reading the raw data from the magnetic scale (24-bit resolution), converting it to physical position, calculating the control error (e(k) = Target - Position), performing PID calculations, and outputting the control variable.

[0110] The introduction of slide rails brings the following advantages to the UAV robotic arm system:

[0111] First, the slide rail dynamically adjusts the base position of the robotic arm based on real-time changes in the drone's position. By accurately sensing the drone's position through multi-source data fusion technology, the slide rail quickly responds and moves, providing a more stable operating platform for the robotic arm. This helps offset the drone's deflection caused by external interference (such as wind and airflow), thereby improving the robotic arm's positioning accuracy and operational stability.

[0112] Secondly, the slide rail's high-precision movement capability enables quick and accurate compensation of the drone's positional errors. Compared to the traditional approach of relying solely on the robot arm's own joint adjustments to compensate for errors, the introduction of the slide rail shares some of the compensation burden, reducing the burden on the robot arm's joints and helping to extend the robot's service life. Furthermore, the slide rail's higher compensation accuracy improves the positioning accuracy of the robot arm's end effector, meeting the demands of high-precision operations.

[0113] Finally, because the slide rails are flexible, the system can quickly and accurately adjust their position based on the updated target pose matrix, ensuring the robotic arm is always in the optimal operating position. This dynamic adjustment capability expands the robotic arm's operating range, enabling it to complete complex operations in a wider range of spaces without requiring the drone to frequently adjust its flight attitude, thereby improving operational efficiency.

[0114] This UAV manipulator collaborative control system with dynamic slide rail compensation realizes the fusion of IMU, visual odometer, and lidar data through extended Kalman filtering, complementing the advantages of different sensors, improving the perception accuracy and response speed of UAV posture changes, and providing a reliable data basis for compensation control. In addition, the present invention adjusts the position of the manipulator base in real time according to the UAV posture by the slide rail, dynamically offsetting the UAV offset. Compared with the traditional joint compensation method, it reduces the burden on the manipulator arm, extends its service life, and achieves more accurate end positioning. At the same time, the present invention combines PID closed-loop control with magnetic scale feedback to achieve millimeter-level precise movement of the slide rail. Through the synergistic effect of proportional, integral, and differential terms, it quickly suppresses errors, ensures that the manipulator arm is always in the optimal operating position, expands the operating range, and improves efficiency.

[0115] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A UAV manipulator collaborative control system with dynamic slide rail compensation, comprising a multi-source data acquisition module, a data fusion processing module, a dynamic pose matrix update module, and a slide rail compensation control module, characterized in that: The multi-source data acquisition module is used to synchronously collect sensor data from the drone's IMU, visual odometer, and lidar; the data fusion processing module is used to fuse the sensor data based on the extended Kalman filter algorithm and output the drone's pose estimation in a unified coordinate system; The dynamic pose matrix update module is used to update the target pose matrix in real time according to the pose estimation and target pose, and calculate the pose error; the slide rail compensation control module is used to drive the slide rail to move through a closed-loop control strategy to compensate for the change in the drone's pose when the pose error exceeds a preset threshold.

2. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 1, characterized in that: In the multi-source data acquisition module, the IMU collects acceleration and angular velocity data at a frequency greater than 100 Hz, the visual odometer collects environmental feature data at a frequency of 10-30 Hz, and the lidar collects three-dimensional point cloud data.

3. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 2, characterized in that: The data fusion processing module adopts the extended Kalman filter algorithm, including: Prediction stage: Calculate the predicted state through the kinematic equation: x k∣k-1 =f(x k-1∣k-1 ,u k ) And propagate the covariance: P k∣k-1 =F k P k-1∣k-1 F k T +Q k where x k∣k-1 is the predicted state at time k, x k-1∣k-1 is the optimal state at time k-1, u k is the IMU input, F k is the state transfer matrix, Q k is the process noise covariance matrix; Update phase: For visual odometry, calculate the new information: v vis =z vis,k -h vis (x k∣k-1 ) Calculate the Kalman gain: K vis,k =P k∣k-1 H T vis,k (H vis,k P k∣k-1 H T vis,k +R vis,k ) -1 Calculate the updated status: Calculate the covariance: P k∣k =(I-K vis,k H vis,k )P k∣k-1 Among them, z vis,k is the visual observation value, h vis is the visual observation function, H vis,k is the visual observation matrix, R vis,k is the visual measurement noise covariance matrix; For lidar, the same process is followed, using the corresponding measurement matrix and noise covariance.

4. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 1, characterized in that: In the dynamic pose matrix update module, the target pose matrix is ​​expressed as: in, is the pose transformation matrix of the UAV in the world coordinate system, is the pose transformation matrix of the target relative to the UAV; When the position error exceeds 1 mm or the attitude error exceeds 0.5 degrees, the slide compensation mechanism is triggered.

5. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 4, characterized in that: The slide rail compensation control module adopts PID closed-loop control, and the control law is: Among them, u(k) is the control quantity, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(k) is the current position error, and Δt is the sampling time interval.

6. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 5, characterized in that: The slide rail compensation control module obtains slide rail position feedback through the magnetic scale. The resolution of the magnetic scale is 0.1 micron. The position calculation formula is: Position=RawData×0.1μm / LSB; Among them, RawData is the original 24-bit data of the magnetic scale.

7. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 6, characterized in that: In the slide rail compensation control module, the linear drive unit drives the slide rail to move according to the target displacement, forms a closed loop through magnetic scale feedback, and adjusts the motor output in real time to ensure the accuracy of the slide rail movement.

8. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 1, characterized in that: The multi-source data acquisition module, the data fusion processing module and the dynamic posture matrix update module are synchronized with each other through the ROS robot operating system, and the communication cycle is ≤10ms.

9. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to claim 1, characterized in that: It also includes a robot base coordinate system update module, which is used to update the robot base coordinate system according to the actual displacement of the slide rail, so that the robot arm motion planning is based on the latest spatial relationship.

10. The UAV manipulator arm collaborative control system with dynamic slide rail compensation according to any one of claims 1 to 9, characterized in that: The collaborative control steps include: Synchronously collect sensor data from IMU, visual odometry, and lidar; The extended Kalman filter algorithm is used to fuse sensor data to obtain the UAV pose estimation; Update the target pose matrix based on the pose estimate and the target pose, and calculate the pose error; When the posture error exceeds the threshold, the slide rail is driven to move through PID closed-loop control to compensate for the posture change.

Images