Laser radar-based jujube tree adaptive pruning method and system

Abstract

The application discloses a jujube tree self-adaptive pruning method and system based on a laser radar, and belongs to the technical field of agricultural automation. In order to solve the problem that the distance between a cutter head and a trunk is difficult to accurately keep due to disturbance of a machine tool posture in existing pruning operation, the method comprises the following steps: obtaining original distance data detected by a laser radar and machine tool posture data detected by a posture sensor, fusing and solving a real relative distance after elimination of the posture disturbance; obtaining a current position of the cutter head center through forward solution of a mechanical arm-push rod kinematic model according to a current swing angle of the mechanical arm; receiving a target distance input by a user; calculating a displacement vector according to the target distance, the real relative distance and the current position of the cutter head, and mapping the displacement vector into a target length of a servo push rod through inverse solution of the kinematic model; driving the push rod to move the cutter head to the target distance, and forming closed-loop control by monitoring the position of the cutter head in real time. The method is mainly used for self-adaptive pruning of jujube trees, and constant control of the distance between the cutter head and the trunk under dynamic disturbance is realized.

Description

Technical Field

[0001] This invention relates to the field of adaptive control technology. More specifically, this invention relates to a method and system for adaptive pruning of jujube trees based on lidar. Background Technology

[0002] As one of my country's important economic fruit trees, the quality of pruning directly affects the yield and fruit quality of the year. The central leader (CH) jujube tree shape, due to its regular tree structure, good ventilation and light penetration, and ease of mechanization, has become the mainstream cultivation shape in large-scale jujube orchards. Currently, pruning of CH jujube trees mainly relies on manual labor, resulting in low efficiency, high labor intensity, and inconsistent pruning quality, making it difficult to meet the timeliness requirements of large-scale planting. With the increasing prominence of agricultural labor shortages, the development of automated pruning equipment to replace manual labor has become an urgent need for the mechanization of jujube orchards.

[0003] In recent years, with the development of sensor and automatic control technologies, contour pruning technology based on lidar ranging has begun to be applied to fruit tree pruning equipment. Existing technologies, such as "a lidar-based method for precise pruning positioning in orchards," reconstruct a tree canopy model using three-dimensional point clouds and identify the positions of branches to be pruned, guiding the robotic arm in pruning operations. Other studies use lidar to detect the distance between the implement and the tree trunk in real time, controlling the robotic arm's swing to maintain a constant relative position between the cutter head and the tree. However, these technologies generally simplify the pruning problem to a two-dimensional distance tracking problem, assuming the lidar is always horizontally mounted and the implement is in a horizontal, stationary state, ignoring the posture changes of the pruning implement caused by uneven ground, tire sinking, and other factors during actual field operations.

[0004] In actual operation scenarios, pruning machines travel between rows in jujube orchards. Uneven ground, ditches, and varying hardness cause the machine to roll and pitch. At this time, the lidar fixed to the machine tilts with the machine, and its emitted laser beam is no longer horizontal, resulting in systematic errors introduced by the machine's attitude in the measured distance data. For example, when the machine tilts to one side, the laser beam shines obliquely onto the tree trunk, and the measured distance will be greater than the actual horizontal distance. If such errors are not corrected, the control system will adjust the cutter head position based on erroneous distance information, causing the actual pruning depth to deviate from the expected depth, potentially leading to missed pruning or damage to the main trunk. Current technology lacks an effective solution to address the impact of machine attitude disturbances on distance measurement accuracy.

[0005] Furthermore, most existing pruning control methods employ simple proportional control logic, directly adjusting the swing angle of the robotic arm based on distance deviation, lacking precise modeling of the kinematic characteristics of the robotic arm-push rod mechanism. There is a nonlinear mapping relationship between the swing angle of the robotic arm and the horizontal position of the cutter head center, and the extension / retraction of the push rod is also a nonlinear function of the robotic arm swing angle. Existing technologies typically ignore this nonlinearity, employing approximate linearization or open-loop control, resulting in low cutter head positioning accuracy and slow response speed. This makes it difficult to achieve precise constant-value control under dynamic disturbances such as changes in the machine's travel speed and trunk swaying. When the relative distance between the machine and the trunk fluctuates, the control system often exhibits a lag response, failing to compensate for positional deviations caused by disturbances in real time, severely impacting the consistency and reliability of pruning operations.

[0006] In summary, existing jujube tree pruning control technologies suffer from the following main deficiencies: First, they lack effective compensation methods for machine posture disturbances, and the lidar ranging data is affected by vehicle tilt, resulting in systematic errors. Second, they lack precise modeling of the kinematic characteristics of the robotic arm-push-rod mechanism, leading to insufficient precision in cutter head position control. Third, the control strategies are mostly open-loop or simple feedback, making them susceptible to dynamic disturbances during field operations and unable to achieve constant and precise control of the distance between the cutter head and the trunk. These three deficiencies result in inconsistent pruning depths and poor operational adaptability in the practical application of existing automated pruning equipment, hindering its large-scale promotion and application. Summary of the Invention

[0007] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0008] Another objective of this invention is to provide an adaptive pruning method for jujube trees based on lidar, which can eliminate the influence of machine attitude disturbances on distance measurement by fusing attitude data and kinematic models, and achieve closed-loop precise control of the cutter head position, significantly improving the accuracy and consistency of pruning operations under dynamic disturbances.

[0009] To achieve these objectives and other advantages according to the present invention, a lidar-based adaptive pruning method for jujube trees is provided, comprising: S1. Obtain the original distance data between the pruning tool and the jujube tree trunk detected by the lidar, and at the same time obtain the pruning tool attitude data detected by the attitude sensor. Calculate the real relative distance between the pruning tool and the trunk after eliminating the attitude disturbance of the pruning tool through the fusion algorithm. S2. Based on the current swing angle of the robotic arm relative to the pruning tool, the forward kinematics model of the robotic arm-push rod is used to determine the current actual position of the center of the cutter head on the robotic arm. S3. Receive the target distance between the tool turret and the main body, input by the user; S4. Based on the target distance, the relative distance between the actual pruning tool and the main trunk, and the current actual position of the cutter head center, calculate the required displacement vector of the cutter head center, and perform inverse kinematics solution through the robotic arm-puss kinematics model to map the displacement vector to the target length of the servo push rod. S5 drives the servo push rod to move according to the target length, so that the cutter head moves to the target distance, and monitors the position of the cutter head in real time to form a closed-loop control, so as to realize the constant value control of the distance between the cutter head and the trunk of the pruning machine under dynamic disturbance; The robotic arm, lidar, and attitude sensor are all mounted on the column of the pruning machine, and the robotic arm is hinged to the column, while the push rod is hinged between the column and the robotic arm.

[0010] Preferably, the step S1, which involves calculating the actual relative distance between the pruning tool and the trunk after eliminating the pruning tool's posture disturbance using a fusion algorithm, specifically includes: S11. Establish an inertial coordinate system {O} and a body coordinate system {B}. The origin of the body coordinate system {B} is located at the laser radar measurement center, the X-axis is along the forward direction of the pruning tool, the Y-axis is horizontal pointing to the side where the jujube tree is located, and the Z-axis is vertically downward. At the initial moment, the axes of the body coordinate system {B} and the inertial coordinate system {O} are parallel. S12. Obtain the roll angle of the pruning tool detected by the attitude sensor. And the pitch angle θ, where the roll angle is the angle of rotation of the pruning machine body about the X-axis and the pitch angle is the angle of rotation of the pruning machine body about the Y-axis, construct the rotation matrix from the body coordinate system {B} to the inertial coordinate system {O}: ; S13. Obtain the raw distance data d detected by the lidar, and the unit vector V of the direction of the laser beam in the body coordinate system {B} at the current moment. B =[v x ,v y ,v z ] T Then the position vector P of the measurement point on the main surface in the body coordinate system. B For: P B =d·V B ; S14. The position vector P of the main surface measurement points after transforming from the body coordinate system to the inertial coordinate system. O for: ; S15. Extract the horizontal component D of the measurement points on the main surface in the inertial coordinate system. true The true relative distance D after eliminating attitude disturbances is obtained. true : ; Wherein, the unit vector V of the laser beam direction B Determined by the internal scanning mechanism of the lidar; , is the position vector of the origin of the body coordinate system in the inertial coordinate system; P O,x P O,y And respectively P O The x and y coordinate components in the inertial coordinate system.

[0011] Preferably, the robotic arm-puller kinematic model described in step S2 is established through the following steps: S21. Define the geometric parameters of the robotic arm-pussor mechanism: Let point A be the hinge point between the pruning tool column and the robotic arm, and point C be the connection point between the robotic arm and the center of the cutter head. The robotic arm is a rigid link, and the length of the line AC is denoted as L. AC ; Let point D be the hinge point between the servo actuator and the column, and point B be the hinge point between the actuator and the robotic arm. Point B lies on the line AC, and the length of the line AB is denoted as L. AB And L AB <L AC ; Establish a Cartesian coordinate system with point A as the origin, the x-axis pointing horizontally towards the side where the jujube tree is located, and the y-axis pointing vertically downwards. Then the coordinates of point D are (0, -H). D ); S22. Establish the positive equation relationship between the swing angle β of the robotic arm and the center position of the cutter head: When the swing angle is β, the coordinates of point B are (x, y). B ,y B ) is represented as: x B =L AB ·sinβ,y B =-L AB ·cosβ; The coordinates of point C (x) C ,y C ) is represented as: x C =L AC ·sinβ,y C =-L AC ·cosβ; S23. Establish the constraint equations between the push rod length S and the robot arm swing angle β: The real-time length S of the push rod is the distance between point B and point D, which is: ; The inverse solution of the constraint equations gives the swing angle β of the robotic arm: ; Among them, H D Let be the perpendicular distance from point D to point A.

[0012] Preferably, step S2, which involves determining the current actual position of the cutter head center on the robotic arm by performing forward kinematics analysis using a pre-established robotic arm-push rod kinematic model, is as follows: Real-time detected swing angle Substituting the coordinates of point C from step S22, we obtain the current actual position of the tool turret center: x C =L AC ·sinβ,y C =-L AC ·cosβ.

[0013] Preferably, the specific steps for mapping the displacement vector to the target length of the servo actuator in step S4 are as follows: S41. Based on the target distance D set and the true relative distance D true Calculate the target position that the center of the cutter head needs to reach in the x-axis direction. : ; S42, based on the length L of the robotic arm AC The target swing angle β is obtained by inverse solving the geometric relationship in step S22. target : ; S43, Based on the target swing angle β target Calculate the target position that the center of the cutter head needs to reach in the y-axis direction. : ; S44. Solve for the target swing angle β target Substituting the constraint equations from step S23, we obtain the target length S of the corresponding push rod. target : .

[0014] Preferably, step S5 involves real-time monitoring of the tool turret position to form a closed-loop control, specifically including: S51. During each control cycle, the current actual swing angle β of the robotic arm is collected in real time by the angle sensor. actual ; S52. Calculate the swing angle deviation Δβ: Δβ = β target -β actual ; S53. Calculate the compensation angle Δβ for the swing angle deviation using the built-in PID control mechanism.comp : ; S54. Based on the compensation angle and constraint equations, recalculate the target length S of the servo actuator after compensation. comp : ; S55, the compensated target length S comp The new control command is sent to the servo push rod, which drives the robotic arm to swing, so that the cutter head approaches the target position; S56. Repeat steps S51 to S55 until the angle deviation Δβ converges within the preset error range, thereby achieving closed-loop control of the tool head position. K1, K2, and K3 are the proportional, integral, and differential coefficients, respectively, which are tuned using the Ziegler-Nichols method.

[0015] Preferably, the step S14 described The distance traveled by the pruning machine is calculated by integrating the odometer and updated in real time based on the initial position.

[0016] Preferably, the laser beam direction unit vector V in step S13 is... B The method for determining it is as follows: A single-point lidar fixedly mounted on the column emits a laser beam horizontally along the positive Y-axis. Then V B =[0,1,0] T .

[0017] This invention further claims a lidar-based adaptive pruning system for jujube trees, used to implement the lidar-based adaptive pruning method for jujube trees, comprising: A multi-source information sensing module, installed on the upright of the pruning machine, includes: The lidar, installed near the bottom of the column, is used to collect raw distance data between the pruning equipment and the main trunk of the jujube tree; An attitude sensor, mounted on the column near the lidar, is used to collect real-time attitude data of the pruning machine, including roll angle. and pitch angle θ; An angle sensor, installed at the hinge point between the robotic arm and the column, is used to collect the real-time swing angle β of the robotic arm relative to the trimming tool. The data fusion and calculation module is connected to the multi-source information sensing module. It is used to receive the original distance data and attitude data, and calculate the real relative distance Dtrue between the pruning machine and the trunk after eliminating the attitude disturbance of the pruning machine through the fusion algorithm. A kinematic model storage module is used to store a pre-established kinematic model of a robotic arm-push rod, which includes the geometric parameters of the robotic arm-push rod mechanism, the forward kinematics relationship between the center position of the cutter head and the swing angle of the robotic arm, and the constraint equations between the push rod length and the swing angle of the robotic arm. The central control module, connected to the data fusion and calculation module, the kinematic model storage module, and the angle sensor, is used for: Receive the target distance D between the tool turret and the main body, input by the user. set ; Based on the current swing angle β of the robotic arm actual The kinematic model is called to perform a forward kinematics solution to determine the current actual position of the cutter head center. , ); According to the target distance D set True relative distance D true And the current actual position of the cutter head center, calculate the displacement vector required for the cutter head center; The kinematic model is called to perform inverse kinematics solution, and the displacement vector is mapped to the target length S of the servo actuator. target ; The servo drive execution module is installed between the column and the robotic arm of the pruning machine. It includes a servo push rod and a servo driver. One end of the servo push rod is hinged to the column and the other end is hinged to the robotic arm. The servo driver receives the target length command issued by the central control module and drives the servo push rod to move, so that the cutter head moves to the target distance. The closed-loop feedback control module, connected to the angle sensor and servo drive execution module, is used to monitor the actual position of the cutter head in real time and compare it with the target position. It calculates the compensation angle through a built-in PID control mechanism and generates the compensated push rod target length S. comp The data is then sent to the servo drive execution module to achieve closed-loop precise control of the tool head position; The human-computer interaction module, connected to the central control module, includes a touchscreen for receiving user input of the target distance D. set It also displays system operation parameters and working status in real time.

[0018] The present invention has at least the following beneficial effects: Firstly, this invention integrates raw distance data from lidar with roll and pitch angles detected by attitude sensors to construct a rotation matrix from the body coordinate system to the inertial coordinate system, accurately calculating the true relative distance after eliminating machine attitude disturbances. This effectively solves the problem of ranging errors caused by uneven ground during actual operation, providing a reliable data foundation for subsequent precise control of the cutterhead. Secondly, this invention establishes an accurate kinematic model based on the geometric parameters of the robotic arm-push rod mechanism. The forward kinematics determines the center position of the cutter head in real time, and the inverse kinematics maps the displacement vector to the target length of the push rod, thus realizing the nonlinear accurate calculation of the cutter head position. This overcomes the shortcomings of the traditional linear approximation method in terms of low positioning accuracy and significantly improves the control accuracy of the cutter head position. Thirdly, this invention constructs a closed-loop control structure with the actual swing angle as the feedback quantity. Through the built-in PID control mechanism, the compensation angle is calculated and the target length of the push rod after compensation is generated, realizing the real-time correction of the swing angle of the robotic arm. It effectively resists dynamic disturbances such as changes in the machine's travel speed and tree trunk swaying, ensuring that the constant value of the distance between the cutter head and the trunk is accurately maintained. Fourth, this invention integrates the lidar and attitude sensor on the column, and the angle sensor is coaxially mounted on the hinge point of the robotic arm, forming a compact layout of multi-source information sensing and execution mechanism. This ensures the consistency of coordinate reference between each sensor and execution component, simplifies the system calibration process, and improves the reliability of the overall system's measurement and control.

[0019] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0020] Figure 1 This is a schematic flowchart of the adaptive pruning method for jujube trees based on lidar as described in one technical solution of the present invention; Figure 2 This is a simplified kinematic model of the robotic arm-pussor in another technical solution of the present invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0022] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0023] like Figure 1 , 2 As shown, this invention provides an adaptive pruning method for jujube trees based on lidar, comprising: S1. Obtain the original distance data between the pruning tool and the jujube tree trunk detected by the lidar, and at the same time obtain the pruning tool attitude data detected by the attitude sensor. Calculate the real relative distance between the pruning tool and the trunk after eliminating the attitude disturbance of the pruning tool through the fusion algorithm. S2. Based on the current swing angle of the robotic arm relative to the pruning tool, the forward kinematics model of the robotic arm-push rod is used to determine the current actual position of the center of the cutter head on the robotic arm. S3. Receive the target distance between the tool turret and the main body, input by the user; S4. Based on the target distance, the relative distance between the actual pruning tool and the main trunk, and the current actual position of the cutter head center, calculate the required displacement vector of the cutter head center, and perform inverse kinematics solution through the robotic arm-puss kinematics model to map the displacement vector to the target length of the servo push rod. S5 drives the servo push rod to move according to the target length, so that the cutter head moves to the target distance, and monitors the position of the cutter head in real time to form a closed-loop control, so as to realize the constant value control of the distance between the cutter head and the trunk of the pruning machine under dynamic disturbance; The robotic arm, lidar, and attitude sensor are all mounted on the column of the pruning machine, and the robotic arm is hinged to the column, while the push rod is hinged between the column and the robotic arm.

[0024] The above technical solution discloses an adaptive pruning method for jujube trees based on lidar. First, core sensors are selected and assembled on the pruning machine to form a multi-source information sensing module, including lidar, attitude sensor, and angle sensor. The lidar can be a commercially available single-line or multi-line industrial lidar product, such as a ranging radar suitable for outdoor environments produced by brands like SICK, Pepperl+Fuchs, or domestic companies like SLAMTEC. It is rigidly mounted on the pruning machine's column near the bottom to ensure a stable measurement baseline and acquire distance information from the jujube tree trunk. The attitude sensor can be a high-precision MEMS inertial measurement unit or tilt sensor, such as industrial-grade products from ADI or STMicroelectronics. It should be mounted close to the lidar on the column to ensure that both sensors detect the same machine attitude. The angle sensor can be an absolute encoder or a high-precision potentiometer, such as products from Heidenhain (Germany) or Changchun Yuheng (China). It must be coaxially mounted at the hinge point between the robotic arm and the column, with the rotor fixed to the robotic arm and the stator fixed to the column, to directly measure the real-time swing angle of the robotic arm relative to the column. The aforementioned sensors together form the basis of multi-source information perception, solving the key defect in existing technologies that rely solely on a single lidar and cannot detect the tilt attitude of the implement itself. This is because, in actual field operations, uneven ground causes the implement to roll and pitch, which will cause the laser beam to deviate from the horizontal direction, thus introducing huge ranging errors. Simply relying on the original distance data is simply not enough to achieve precise control.

[0025] The data collected by the aforementioned core sensors is further fused and calculated by the controller (central control module). The controller can be an embedded industrial computer based on ARM or x86 architecture. The embedded data fusion algorithm first constructs a rotation matrix using the roll and pitch angles measured by the attitude sensors, transforming the raw distance data measured by the lidar from a body coordinate system tilted with the vehicle body to a horizontally stable inertial coordinate system, thereby calculating the true relative distance after eliminating the machine's attitude disturbances. Simultaneously, based on the swing angle of the robotic arm collected by the angle sensor, a pre-stored robotic arm-push rod kinematic model is called for forward kinematics calculation. This model, based on the geometric parameters of the robotic arm, can accurately calculate the current actual horizontal and vertical position of the cutter head center. Data fusion and calculation, and the construction and calculation of the robotic arm-push rod kinematic model, solve the technical problem of inaccurate control references caused by ignoring machine attitude disturbances and kinematic nonlinearities in existing technologies. Only by obtaining the true and reliable machine-main distance and the actual position of the cutter head can subsequent precise control be meaningful. For servo actuators, industrial-grade electric actuators from brands such as Thomson or Limtec can be selected. One end of the actuator is hinged to the column, and the other end is hinged to the robotic arm, forming a crank-slider mechanism to drive the robotic arm to swing.

[0026] In actual operation, raw distance data and machine posture data are first collected in real time using LiDAR and attitude sensors. After fusion calculation, the actual relative distance between the pruning machine and the main trunk is obtained. Simultaneously, the current swing angle of the robotic arm is obtained through an angle sensor, and the actual position of the cutter head center is determined based on the forward kinematics model. Subsequently, the controller reads the distance between the cutter head and the target main trunk input by the user through the human-machine interface touch screen, and calculates the target position that the cutter head center needs to reach in the horizontal direction by combining it with the actual relative distance. Then, the corresponding target swing angle and servo push rod target length are obtained through inverse kinematics model. Control commands are sent to the servo driver, which drives the electric push rod to move the cutter head towards the target position. During this process, the angle sensor continuously feeds back the actual angle of the robotic arm, and the closed-loop feedback control module calculates the angle deviation in real time and generates compensation through the controller, continuously correcting the push rod target length until the cutter head accurately reaches the target distance.

[0027] The above technical solution effectively eliminates the impact of machine posture disturbances caused by uneven field ground on distance measurement accuracy by integrating LiDAR and attitude sensor data, solving the technical problem of measurement inaccuracy of a single sensor in dynamic environments. Based on the accurate robotic arm-push rod kinematic model, forward and inverse kinematics calculations are performed to achieve nonlinear and accurate mapping of the cutter head position, overcoming the low positioning accuracy of traditional linear approximation methods. The introduction of a closed-loop control structure with the actual swing angle as feedback can resist dynamic disturbances such as changes in machine travel speed and trunk swaying in real time. Ultimately, it achieves constant and accurate control of the distance between the cutter head and the trunk of the pruning machine in complex working environments, significantly improving the consistency and automation level of pruning operations.

[0028] In one technical solution, step S1, which involves calculating the actual relative distance between the pruning tool and the main trunk after eliminating the posture disturbance of the pruning tool using a fusion algorithm, specifically includes: S11. Establish an inertial coordinate system {O} and a body coordinate system {B}. The origin of the body coordinate system {B} is located at the laser radar measurement center, the X-axis is along the forward direction of the pruning tool, the Y-axis is horizontal pointing to the side where the jujube tree is located, and the Z-axis is vertically downward. At the initial moment, the axes of the body coordinate system {B} and the inertial coordinate system {O} are parallel. S12. Obtain the roll angle of the pruning tool detected by the attitude sensor. And the pitch angle θ, where the roll angle is the angle of rotation of the pruning machine body about the X-axis and the pitch angle is the angle of rotation of the pruning machine body about the Y-axis, construct the rotation matrix from the body coordinate system {B} to the inertial coordinate system {O}: ; S13. Obtain the raw distance data d detected by the lidar, and the unit vector V of the direction of the laser beam in the body coordinate system {B} at the current moment. B =[v x ,v y ,v z ] T Then the position vector P of the measurement point on the main surface in the body coordinate system. B For: P B =d·V B ; S14. The position vector P of the main surface measurement points after transforming from the body coordinate system to the inertial coordinate system. O for: ; S15. Extract the horizontal component D of the measurement points on the main surface in the inertial coordinate system. true The true relative distance D after eliminating attitude disturbances is obtained. true : ; Wherein, the unit vector V of the laser beam direction BDetermined by the internal scanning mechanism of the lidar; , is the position vector of the origin of the body coordinate system in the inertial coordinate system; P O,x P O,y And respectively P O The x and y coordinate components in the inertial coordinate system.

[0029] Since the raw distance data directly output by the lidar is measured in a body coordinate system tilted with the vehicle body, when the pruning tool rolls or pitches due to uneven ground, the laser beam deviates from the horizontal direction, resulting in a systematic error between the measured distance value and the actual horizontal distance. This makes it impossible for subsequent cutter head control to be established on an accurate distance reference. The above technical solution further optimizes the method of realizing the relative distance between the actual pruning tool and the trunk. First, an inertial coordinate system and a body coordinate system are defined in the controller. The inertial coordinate system is usually a northeast-northeast coordinate system, while the body coordinate system is fixed to the pruning tool, with its origin set at the lidar measurement center. The X-axis points in the direction of the pruning tool's movement, the Y-axis points horizontally to the side of the jujube tree to be pruned, and the Z-axis points vertically downward. The roll and pitch angles output in real time by the attitude sensor are used to construct a rotation matrix from the body coordinate system to the inertial coordinate system. Each element of the rotation matrix is ​​composed of angle trigonometric function values, which can accurately describe the rotational transformation relationship between the two coordinate systems. Meanwhile, each measurement by the lidar not only outputs the distance value d, but also provides the direction unit vector of the current laser beam in the machine coordinate system based on its internal scanning mechanism. For a fixed-installation single-point lidar, this direction unit vector is constant along the positive Y-axis, i.e., V. B =[0,1,0] T Multiplying the distance by the direction vector yields the three-dimensional position vector of the tree trunk measurement point in the machine's coordinate system. This vector is then rotated to the inertial coordinate system using a rotation matrix, and superimposed with the position vector of the machine's origin in the inertial coordinate system obtained from GPS or odometer measurements, thus obtaining the absolute coordinates of the tree trunk measurement point in the inertial coordinate system. Finally, the horizontal component of this coordinate system is extracted and its modulus is calculated, yielding the true relative distance after eliminating machine attitude disturbances. By using roll and pitch angles to transform the raw sensor data to a unified and stable reference frame, the obtained distance value accurately reflects the true horizontal distance between the pruning machine and the trunk even during severe jogging.

[0030] The above technical solution establishes a coordinate transformation model to transform the lidar ranging value from the machine coordinate system that tilts with the vehicle body to a horizontally stable inertial coordinate system. This fundamentally eliminates the interference of the machine's roll and pitch attitude changes on the ranging accuracy, solves the long-standing technical problem of inaccurate measurement by traditional single lidar in dynamic field environments, provides an accurate and reliable distance reference for the subsequent precise control of the cutter head position, and significantly improves the adaptability and control accuracy of the adaptive pruning system in complex terrain.

[0031] In one of the technical solutions, the robotic arm-pussor kinematic model described in step S2 is established through the following steps: S21. Define the geometric parameters of the robotic arm-pussor mechanism: Let point A be the hinge point between the pruning tool column and the robotic arm, and point C be the connection point between the robotic arm and the center of the cutter head. The robotic arm is a rigid link, and the length of the line AC is denoted as L. AC ; Let point D be the hinge point between the servo actuator and the column, and point B be the hinge point between the actuator and the robotic arm. Point B lies on the line AC, and the length of the line AB is denoted as L. AB And L AB <L AC ; Establish a Cartesian coordinate system with point A as the origin, the x-axis pointing horizontally towards the side where the jujube tree is located, and the y-axis pointing vertically downwards. Then the coordinates of point D are (0, -H). D ); S22. Establish the positive equation relationship between the swing angle β of the robotic arm and the center position of the cutter head: When the swing angle is β, the coordinates of point B are (x, y). B ,y B ) is represented as: x B =L AB ·sinβ,y B =-L AB ·cosβ; The coordinates of point C (x) C ,y C ) is represented as: x C =L AC ·sinβ,y C =-L AC ·cosβ; S23. Establish the constraint equations between the push rod length S and the robot arm swing angle β: The real-time length S of the push rod is the distance between point B and point D, which is: ; The inverse solution of the constraint equations gives the swing angle β of the robotic arm: ; Among them, H D Let be the perpendicular distance from point D to point A.

[0032] Because the robotic arm and push rod form a crank-slider mechanism, the trajectory of the cutter head center when the robotic arm swings around the hinge point is not a simple straight line, but an arc curve that varies with the swing angle. Simultaneously, there is a non-linear functional relationship between the push rod's extension / retraction and the robotic arm's swing angle. If this non-linearity is ignored and only approximate linearization is used, the actual position of the cutter head will deviate significantly from the desired position when the robotic arm's swing range is large or fine adjustments are required, directly affecting the consistency of the trimming depth. The above technical solution first defines points A, B, C, and D as key hinge points, and the length of the line connecting points A and C (presumably the robotic arm length) L. AC The length L of the hinge point between the push rod and the robotic arm AB And the coordinates of point D (0, -H) D This allows for precise geometric modeling of the robotic arm-pussor mechanism, abstracting the complex physical mechanism into a mathematical model with clear geometric meaning, and transforming the actual physical problem into a computable mathematical problem. It also eliminates the need to simplify the pusher drive and cutter head movement into a linear proportional relationship, thus ensuring control precision.

[0033] Furthermore, the above technical solution establishes a forward kinematic relationship between the swing angle β of the robotic arm and the center position of the cutter head. Through trigonometric functions, the swing angle β of the robotic arm is directly correlated with the coordinates of the cutter head center; that is, the horizontal position of the cutter head is proportional to sinβ, and the vertical position is proportional to cosβ. This forward kinematic relationship accurately reflects the motion law of the cutter head as it swings with the robotic arm. Simultaneously, the relationship between the coordinates of point B and β is also established, providing the conditions for subsequently deriving the constraint equations for the length and angle of the push rod. Through this forward kinematic relationship, the control system can calculate the precise position of the cutter head center in real time based on the known swing angle, providing reliable feedback for closed-loop control.

[0034] Finally, the constraint equations between the push rod length S and the robotic arm swing angle β are established, forming the core of the robotic arm-push rod kinematic model. The relationship between the push rod length S and the swing angle β depends on the spatial distance between points B and D, which exhibits a cosine function relationship as β changes. This application derives the constraint equations between S and β by substituting the coordinates of points B and D into the distance formula between two points. These constraint equations precisely describe the mapping relationship between the drive input (push rod length) and the joint output (swing angle). The derivation process of the real-time push rod length S as the distance between points B and D is as follows: Substituting the coordinates of points B and D, we get... After simplification, we get .

[0035] With the aforementioned constraint equations, the control system can solve for the actual angle of the robotic arm when the push rod length is known (for forward solution), and can also calculate the required push rod length when the target angle is known (for inverse solution), thus achieving precise conversion between drive space and joint space.

[0036] The above technical solution establishes a complete kinematic model including geometric parameters, forward kinematics, and constraint equations, achieving a precise nonlinear mapping between the swing angle of the robotic arm, the extension and retraction of the push rod, and the center position of the cutter head. This fundamentally solves the problem of low cutter head positioning accuracy caused by neglecting the kinematic characteristics of the mechanism in existing technologies, providing accurate mathematical model support for subsequent closed-loop control. This allows the control of the cutter head position to no longer rely on approximate estimation but to be based on strict geometric relationships, significantly improving the accuracy and consistency of pruning operations.

[0037] In one of the technical solutions, step S2, which involves determining the current actual position of the cutter head center on the robotic arm by performing a forward kinematics solution using a pre-established robotic arm-push rod kinematic model, is as follows: Real-time detected swing angle Substituting the coordinates of point C from step S22, we obtain the current actual position of the tool turret center: x C =L AC ·sinβ,y C =-L AC ·cosβ.

[0038] Furthermore, by utilizing the real-time actual swing angle β of the push rod based on the forward kinematics, the current actual position of the cutter head center is accurately calculated. Through the mechanical arm-push rod kinematic model, a precise mapping from the driving space to the working space is achieved, enabling the control system to grasp the real coordinates of the cutter head in real time. This provides accurate and reliable feedback data for subsequent closed-loop control, thereby overcoming the position estimation error caused by mechanical backlash or transmission nonlinearity, and significantly improving the accuracy of cutter head positioning and the system's anti-disturbance capability.

[0039] In one of the technical solutions, the specific steps for mapping the displacement vector to the target length of the servo push rod in step S4 are as follows: S41. Based on the target distance D set and the true relative distance D true Calculate the target position that the center of the cutter head needs to reach in the x-axis direction. : ; S42, based on the length L of the robotic arm AC The target swing angle β is obtained by inverse solving the geometric relationship in step S22. target : ; S43, Based on the target swing angle β target Calculate the target position that the center of the cutter head needs to reach in the y-axis direction. : ; S44. Solve for the target swing angle β target Substituting the constraint equations from step S23, we obtain the target length S of the corresponding push rod. target : .

[0040] The above technical solution combines the user-defined target distance with the actual relative distance to accurately calculate the target position that the cutter head center needs to reach in the horizontal direction. It then uses a kinematic model to solve for the corresponding target swing angle and target push rod length, achieving a precise conversion from agronomic requirements to actuator control commands. These steps solve the problem of fuzzy control targets caused by the lack of precise mapping relationships in existing technologies. The system can accurately calculate the displacement vector that the cutter head should move based on the actual distance deviation, and map the displacement vector to the target extension / retraction of the push rod through inverse kinematic model calculation. This not only ensures that the cutter head can quickly and accurately reach the target distance position, but also provides a clear target reference for subsequent closed-loop control, significantly improving the response speed and positioning accuracy of the control system, and ensuring high consistency in pruning operations at different working distances.

[0041] In one technical solution, step S5 involves real-time monitoring of the tool turret position to form a closed-loop control, specifically including: S51. During each control cycle, the current actual swing angle β of the robotic arm is collected in real time by the angle sensor. actual ; S52. Calculate the swing angle deviation Δβ: Δβ = β target -β actual ; S53. Calculate the compensation angle Δβ for the swing angle deviation using the built-in PID control mechanism. comp : ; S54. Based on the compensation angle and constraint equations, recalculate the target length S of the servo actuator after compensation. comp : ; S55, the compensated target length S comp The new control command is sent to the servo push rod, which drives the robotic arm to swing, so that the cutter head approaches the target position; S56. Repeat steps S51 to S55 until the angle deviation Δβ converges within the preset error range, thereby achieving closed-loop control of the tool head position. K1, K2, and K3 are the proportional, integral, and differential coefficients, respectively, which are tuned using the Ziegler-Nichols method.

[0042] The above technical solution transforms the control of the cutter head position into real-time monitoring of the swing angle, establishing a closed-loop control architecture with the actual swing angle as the feedback quantity. This avoids actual position deviations caused by factors such as tool jolting, trunk swaying, or mechanical transmission clearances during operation. Specifically, by using an angle sensor to collect the current actual swing angle of the robotic arm in real time during each control cycle, and comparing it with the target swing angle calculated in step S42, accurate angle deviation is obtained. The feedback loop allows the control system to perceive the difference between the actual and desired states in real time, providing a basis for subsequent compensation and correction. After obtaining the angle deviation, a PID control mechanism is introduced to process the deviation and generate a compensation angle. As a classic and effective feedback control algorithm, the PID control mechanism comprehensively considers the proportional, integral, and derivative terms of the deviation. The proportional term determines the response strength to the current deviation, the integral term eliminates long-term accumulated static errors, and the derivative term predicts the trend of deviation changes and suppresses overshoot. By inputting the angle deviation into the PID control mechanism, the system can calculate a reasonable compensation angle, which represents the additional swing adjustment required to allow the robotic arm to quickly and smoothly approach the target position. The proportional, integral, and derivative coefficients in the PID control mechanism are tuned using the Ziegler-Nichols method. This method obtains the critical proportional coefficient and oscillation period through critical oscillation experiments, and then calculates the optimal PID parameters to ensure that the control system has good dynamic response performance and stability. After obtaining the compensation angle, it is added to the current actual swing angle to obtain the corrected target swing angle. Then, the constraint equation between the push rod length and the swing angle established in step S23 is used again to calculate the corresponding compensated target push rod length in reverse. This step reflects the deep coupling between closed-loop control and kinematic model: the compensation angle must be converted into the actual extension and retraction of the push rod through precise geometric constraints in order to drive the robotic arm to produce the required swing adjustment. Subsequently, the compensated target length is sent to the servo push rod for execution, causing the cutter head to approach the target position further. The above process is repeated at a very high frequency in each control cycle until the angle deviation converges within the preset allowable error range, thereby achieving continuous and precise control of the cutter head position.

[0043] In one of the technical solutions, the steps described in step S14 are as follows: The odometer is installed on the wheels of the tractor on which the pruning implement is mounted to calculate the distance traveled by the pruning implement. This allows for continuous and autonomous positioning without relying on external signals.

[0044] In one of the technical solutions, the laser beam direction unit vector V mentioned in step S13 B The method for determining it is as follows: A single-point lidar fixedly mounted on the column emits a laser beam horizontally along the positive Y-axis. Then V B =[0,1,0] T .

[0045] For multi-line or scanning lidar, the laser beam direction is determined in real time by the scanning mechanism. The horizontal rotation angle γ1 and vertical elevation angle γ2 of the current laser beam are obtained through the radar communication protocol, and the direction vector in the body coordinate system is obtained after coordinate transformation: v x =cosγ2·sinγ1;v y = cosγ2·cosγ1;v z = sinγ2; where the horizontal rotation angle is positive in the direction of the positive X-axis and the vertical pitch angle is positive in the direction of upward.

[0046] This invention further claims a lidar-based adaptive pruning system for jujube trees, used to implement the lidar-based adaptive pruning method for jujube trees, comprising: A multi-source information sensing module, installed on the upright of the pruning machine, includes: The lidar, installed near the bottom of the column, is used to collect raw distance data between the pruning equipment and the main trunk of the jujube tree; An attitude sensor, mounted on the column near the lidar, is used to collect real-time attitude data of the pruning machine, including roll angle. and pitch angle θ; An angle sensor, installed at the hinge point between the robotic arm and the column, is used to collect the real-time swing angle β of the robotic arm relative to the trimming tool. The data fusion and calculation module is connected to the multi-source information sensing module. It is used to receive the original distance data and attitude data, and calculate the real relative distance Dtrue between the pruning machine and the trunk after eliminating the attitude disturbance of the pruning machine through the fusion algorithm. A kinematic model storage module is used to store a pre-established kinematic model of a robotic arm-push rod, which includes the geometric parameters of the robotic arm-push rod mechanism, the forward kinematics relationship between the center position of the cutter head and the swing angle of the robotic arm, and the constraint equations between the push rod length and the swing angle of the robotic arm. The central control module, connected to the data fusion and calculation module, the kinematic model storage module, and the angle sensor, is used for: Receive the target distance D between the tool turret and the main body, input by the user. set ; Based on the current swing angle β of the robotic arm actual The kinematic model is called to perform a forward kinematics solution to determine the current actual position of the cutter head center. , ); According to the target distance D set True relative distance D true And the current actual position of the cutter head center, calculate the displacement vector required for the cutter head center; The kinematic model is called to perform inverse kinematics solution, and the displacement vector is mapped to the target length S of the servo actuator. target ; The servo drive execution module is installed between the column and the robotic arm of the pruning machine. It includes a servo push rod and a servo driver. One end of the servo push rod is hinged to the column and the other end is hinged to the robotic arm. The servo driver receives the target length command issued by the central control module and drives the servo push rod to move, so that the cutter head moves to the target distance. The closed-loop feedback control module, connected to the angle sensor and servo drive execution module, is used to monitor the actual position of the cutter head in real time and compare it with the target position. It calculates the compensation angle through a built-in PID control mechanism and generates the compensated push rod target length S. comp The data is then sent to the servo drive execution module to achieve closed-loop precise control of the tool head position; The human-computer interaction module, connected to the central control module, includes a touchscreen for receiving user input of the target distance D. set It also displays system operation parameters and working status in real time.

[0047] The lidar-based adaptive pruning system for jujube trees employs a multi-source information sensing module layout. This layout integrates lidar and attitude sensors on the support column, and angle sensors are coaxially mounted at the hinge point of the robotic arm. This ensures consistency of coordinate references between the sensors and actuators, simplifies the system calibration process, and improves the overall reliability of measurement and control. The system uses a data fusion and calculation module to combine raw distance data from the lidar with roll and pitch angles detected by the attitude sensors. This accurately calculates the true relative distance after eliminating machine attitude disturbances, effectively solving the distance measurement error problem caused by uneven ground during field operations. The kinematic model storage module works collaboratively with the central control module, performing forward and inverse kinematic solutions based on a precise robotic arm-push rod kinematic model. This achieves a nonlinear and precise mapping of the cutterhead position, overcoming the low positioning accuracy of traditional linear approximation methods. The closed-loop feedback control module uses the actual swing angle as the feedback quantity. Through the PID control mechanism built into the central control module, it calculates the compensation angle in real time and generates the target length of the push rod after compensation. This enables real-time correction of the robotic arm's swing angle and effectively resists dynamic disturbances such as changes in the machine's travel speed and trunk swaying. The servo drive execution module accurately converts control commands into push rod actions, ensuring the accurate execution of the target distance set by the human-machine interface module. All modules in the entire system have clearly defined functions and work together to achieve constant and precise control of the cutter head-trunk distance of the pruning machine in complex field environments, significantly improving the consistency and automation level of pruning operations.

[0048] A field trial was conducted in a jujube orchard in November 2025. Fifty 8-year-old, single-trunk jujube trees were selected, with an average tree height of 2.5m and a crown width of approximately 99cm. The experiment included a control group and an experimental group: the control group used a conventional open-loop control method, directly controlling the cutter head position based on the original distance data from the lidar; the experimental group used the adaptive pruning method described in this invention, including attitude fusion calculation, forward and inverse kinematics model solutions, and PID closed-loop control.

[0049] Test equipment: The pruning machine was tractor-tethered; the lidar was a SICK LMS111 industrial lidar (mounted at the bottom of the column); the attitude sensor was an ADIS16470 tactical-grade inertial measurement unit (mounted above the lidar); the angle sensor was a Heidenhain absolute encoder (coaxially mounted at the hinge point of the robotic arm); the controller was a Siemens S7-1500 PLC; and the servo actuators were Thomson Electrak HD industrial electric actuators. The distance between the cutter head and the main target was set to 300mm.

[0050] Experimental Data: During continuous operation, data on distance control error, cutter head positioning accuracy, and trimming depth consistency were collected for both the control and experimental groups. Under actual working conditions with a tractor traveling at 0.5 m / s and ground undulations of 10-15 cm, the control group showed a maximum distance deviation of 45 mm, an average deviation of 18.5 mm, and a standard deviation of 12.3 mm. Using the method of this invention, the correlation coefficient between the actual relative distance calculated from the fused attitude data and the manually measured value reached over 0.91, the maximum distance deviation was reduced to 12 mm, the average deviation was 4.2 mm, and the standard deviation was 3.1 mm. Regarding the positioning accuracy of the cutter head center position, the control group had a maximum error of 22 mm and an average error of 9.5 mm in the X direction (horizontal direction); the method of this invention had a maximum error of 6 mm and an average error of 2.8 mm in the X direction, which is better than the average error of 4.4 mm for existing robotic arm end-effector positioning.

[0051] Comparison of pruning depth consistency: Thirty pruning points were randomly selected to measure the actual pruning depth. The average pruning depth of the control group was 285 mm (15 mm deviation from the target value of 300 mm), with a coefficient of variation (CV) of 14.2%. The average pruning depth of the method of this invention was 298 mm, with a coefficient of variation (CV) of 4.8%. In cutting branches with a diameter of 12 mm or more, the method of this invention maintained a stable cutter head posture, avoiding cutter head offset caused by changes in cutting torque.

[0052] Data Analysis: Experimental results show that this invention significantly improves ranging accuracy by eliminating machine disturbance errors through the fusion of attitude sensor data; based on the forward and inverse kinematics model, it achieves nonlinear precise control of the cutterhead position, reducing positioning error by approximately 70% compared to traditional methods; the PID closed-loop control mechanism, using the actual swing angle as feedback, effectively resists dynamic disturbances in the field, reducing the cutterhead-trunk distance control deviation by 76.2% and improving pruning depth consistency by nearly three times. The pruning time per plant is less than 10 minutes, significantly higher than manual pruning (approximately 25 minutes), but the pruning consistency is significantly better than manual pruning.

[0053] To verify the robustness of the adaptive pruning method provided by this invention under different operating environments, three sets of comparative experiments were conducted on sunny days (light intensity > 80,000 lux), cloudy days (light intensity approximately 20,000 lux), and nighttime (artificial lighting). Each set included 15 jujube trees, with the same target distance of 300 mm. The results showed that the measurement deviations of the true relative distance calculated by attitude fusion under the three lighting conditions were: 4.5 mm on sunny days, 3.8 mm on cloudy days, and 4.1 mm at night, with coefficients of variation all less than 5%. This verifies that the lidar ranging is minimally affected by lighting conditions, and that the attitude fusion algorithm exhibits good stability. The PID closed-loop control mechanism showed a settling time of less than 0.3 s and an overshoot of less than 5% under all three operating conditions, indicating that the controller parameters tuned using the Ziegler-Nichols method have good dynamic response characteristics and adaptability.

[0054] The number of devices and processing scale described herein are for simplification of the present invention. Applications, modifications, and variations of the lidar-based adaptive pruning method and system for jujube trees of this invention will be readily apparent to those skilled in the art.

[0055] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A lidar-based adaptive pruning method for jujube trees, characterized in that, include: S1. Obtain the original distance data between the pruning tool and the jujube tree trunk detected by the lidar, and at the same time obtain the pruning tool attitude data detected by the attitude sensor. Calculate the real relative distance between the pruning tool and the trunk after eliminating the attitude disturbance of the pruning tool through the fusion algorithm. S2. Based on the current swing angle of the robotic arm relative to the pruning tool, the forward kinematics model of the robotic arm-push rod is used to determine the current actual position of the center of the cutter head on the robotic arm. S3. Receive the target distance between the tool turret and the main body, input by the user; S4. Based on the target distance, the relative distance between the actual pruning tool and the main trunk, and the current actual position of the cutter head center, calculate the required displacement vector of the cutter head center, and perform inverse kinematics solution through the robotic arm-puss kinematics model to map the displacement vector to the target length of the servo push rod. S5 drives the servo push rod to move according to the target length, so that the cutter head moves to the target distance, and monitors the position of the cutter head in real time to form a closed-loop control, so as to realize the constant value control of the distance between the cutter head and the trunk of the pruning machine under dynamic disturbance; The robotic arm, lidar, and attitude sensor are all mounted on the column of the pruning machine, and the robotic arm is hinged to the column, while the push rod is hinged between the column and the robotic arm.

2. The adaptive pruning method for jujube trees based on lidar as described in claim 1, characterized in that, Step S1, which involves calculating the actual relative distance between the pruning tool and the trunk after eliminating the pruning tool's posture disturbance using a fusion algorithm, specifically includes: S11. Establish an inertial coordinate system {O} and a body coordinate system {B}. The origin of the body coordinate system {B} is located at the laser radar measurement center, the X-axis is along the forward direction of the pruning tool, the Y-axis is horizontal pointing to the side where the jujube tree is located, and the Z-axis is vertically downward. At the initial moment, the axes of the body coordinate system {B} and the inertial coordinate system {O} are parallel. S12. Obtain the roll angle of the pruning tool detected by the attitude sensor. And the pitch angle θ, where the roll angle is the angle of rotation of the pruning machine body around the X-axis and the pitch angle is the angle of rotation of the pruning machine body around the Y-axis, construct the rotation matrix from the body coordinate system {B} to the inertial coordinate system {O}: ； S13. Obtain the raw distance data d detected by the lidar, and the unit vector V of the direction of the laser beam in the body coordinate system {B} at the current moment. B =[v x ,v y ,v z ] T Then the position vector P of the measurement point on the main surface in the body coordinate system. B For: P B =d·V B ; S14. The position vector P of the main surface measurement points after transforming from the body coordinate system to the inertial coordinate system. O for: ; S15. Extract the horizontal component D of the measurement points on the main surface in the inertial coordinate system. true The true relative distance D after eliminating attitude disturbances is obtained. true : ; Wherein, the unit vector V of the laser beam direction B Determined by the internal scanning mechanism of the lidar; , is the position vector of the origin of the body coordinate system in the inertial coordinate system; P O,x P O,y And respectively P O The x and y coordinate components in the inertial coordinate system.

3. The adaptive pruning method for jujube trees based on lidar as described in claim 2, characterized in that, The robotic arm-pussor kinematic model described in step S2 is established through the following steps: S21. Define the geometric parameters of the robotic arm-pussor mechanism: Let point A be the hinge point between the pruning tool column and the robotic arm, and point C be the connection point between the robotic arm and the center of the cutter head. The robotic arm is a rigid link, and the length of the line AC is denoted as L. AC ; Let point D be the hinge point between the servo actuator and the column, and point B be the hinge point between the actuator and the robotic arm. Point B lies on the line AC, and the length of the line AB is denoted as L. AB And L AB <L AC ; Establish a Cartesian coordinate system with point A as the origin, the x-axis pointing horizontally towards the side where the jujube tree is located, and the y-axis pointing vertically downwards. Then the coordinates of point D are (0, -H). D ); S22. Establish the positive equation relationship between the swing angle β of the robotic arm and the center position of the cutter head: When the swing angle is β, the coordinates of point B are (x, y). B ,y B ) is represented as: x B =L AB ·sinβ,y B =-L AB ·cosβ; The coordinates of point C (x) C ,y C ) is represented as: x C =L AC ·sinβ,y C =-L AC ·cosβ; S23. Establish the constraint equations between the push rod length S and the robot arm swing angle β: The real-time length S of the push rod is the distance between point B and point D, which is: ； The inverse solution of the constraint equations gives the swing angle β of the robotic arm: ； Among them, H D Let be the perpendicular distance from point D to point A.

4. The adaptive pruning method for jujube trees based on lidar as described in claim 3, characterized in that, Step S2 involves determining the current actual position of the cutter head center on the robotic arm using a pre-established robotic arm-push rod kinematic model through forward kinematics analysis. Real-time detected swing angle Substituting the coordinates of point C from step S22, we obtain the current actual position of the tool turret center: x C =L AC ·sinβ,y C =-L AC ·cosβ。 5. The adaptive pruning method for jujube trees based on lidar as described in claim 4, characterized in that, The specific steps for mapping the displacement vector to the target length of the servo actuator in step S4 are as follows: S41. Based on the target distance D set and the true relative distance D true Calculate the target position that the center of the cutter head needs to reach in the x-axis direction. : ； S42, based on the length L of the robotic arm AC The target swing angle β is obtained by inverse solving the geometric relationship in step S22. target : ； S43, Based on the target swing angle β target Calculate the target position that the center of the cutter head needs to reach in the y-axis direction. : ； S44. Solve for the target swing angle β target Substituting the constraint equations from step S23, we obtain the target length S of the corresponding push rod. target : 。 6. The adaptive pruning method for jujube trees based on lidar as described in claim 5, characterized in that, Step S5 involves real-time monitoring of the tool turret position to form a closed-loop control, specifically including: S51. During each control cycle, the current actual swing angle β of the robotic arm is collected in real time by the angle sensor. actual ; S52. Calculate the swing angle deviation Δβ: Δβ = β target -β actual ; S53. Calculate the compensation angle Δβ for the swing angle deviation using the built-in PID control mechanism. comp : ； S54. Based on the compensation angle and constraint equations, recalculate the target length S of the servo actuator after compensation. comp : ； S55, the compensated target length S comp The new control command is sent to the servo push rod, which drives the robotic arm to swing, so that the cutter head approaches the target position; S56. Repeat steps S51 to S55 until the angle deviation Δβ converges within the preset error range, thereby achieving closed-loop control of the tool head position. K1, K2, and K3 are the proportional, integral, and differential coefficients, respectively, which are tuned using the Ziegler-Nichols method.

7. The adaptive pruning method for jujube trees based on lidar as described in claim 2, characterized in that, The steps described in step S14 The distance traveled by the pruning machine is calculated by integrating the odometer and updated in real time based on the initial position.

8. The adaptive pruning method for jujube trees based on lidar as described in claim 2, characterized in that, The laser beam direction unit vector V mentioned in step S13 B The method for determining it is as follows: A single-point lidar fixedly mounted on the column emits a laser beam horizontally along the positive Y-axis. Then V B =[0,1,0] T .

9. A lidar-based adaptive pruning system for jujube trees, used to implement the lidar-based adaptive pruning method for jujube trees as described in any one of claims 1 to 8, characterized in that, include: A multi-source information sensing module, installed on the upright of the pruning machine, includes: The lidar, installed near the bottom of the column, is used to collect raw distance data between the pruning equipment and the main trunk of the jujube tree; An attitude sensor, mounted on the column near the lidar, is used to collect real-time attitude data of the pruning machine, including roll angle. and pitch angle θ; An angle sensor, installed at the hinge point between the robotic arm and the column, is used to collect the real-time swing angle β of the robotic arm relative to the trimming tool. The data fusion and calculation module is connected to the multi-source information sensing module. It is used to receive the original distance data and attitude data, and calculate the real relative distance Dtrue between the pruning machine and the trunk after eliminating the attitude disturbance of the pruning machine through the fusion algorithm. A kinematic model storage module is used to store a pre-established kinematic model of a robotic arm-push rod, which includes the geometric parameters of the robotic arm-push rod mechanism, the forward kinematics relationship between the center position of the cutter head and the swing angle of the robotic arm, and the constraint equations between the push rod length and the swing angle of the robotic arm. The central control module, connected to the data fusion and calculation module, the kinematic model storage module, and the angle sensor, is used for: Receive the target distance D between the tool turret and the main body, input by the user. set ; Based on the current swing angle β of the robotic arm actual The kinematic model is called to perform a forward kinematics solution to determine the current actual position of the cutter head center. , ); According to the target distance D set True relative distance D true And the current actual position of the cutter head center, calculate the displacement vector required for the cutter head center; The kinematic model is called to perform inverse kinematics solution, and the displacement vector is mapped to the target length S of the servo actuator. target ; The servo drive execution module is installed between the column and the robotic arm of the pruning machine. It includes a servo push rod and a servo driver. One end of the servo push rod is hinged to the column and the other end is hinged to the robotic arm. The servo driver receives the target length command issued by the central control module and drives the servo push rod to move, so that the cutter head moves to the target distance. The closed-loop feedback control module, connected to the angle sensor and servo drive execution module, is used to monitor the actual position of the cutter head in real time and compare it with the target position. It calculates the compensation angle through a built-in PID control mechanism and generates the compensated push rod target length S. comp The data is then sent to the servo drive execution module to achieve closed-loop precise control of the tool head position; The human-computer interaction module, connected to the central control module, includes a touchscreen for receiving user input of the target distance D. set It also displays system operation parameters and working status in real time.

Images