Pear tree branch real-time identification and collaborative pruning control system

Abstract

The invention discloses a pear tree branch real-time identification and collaborative pruning control system. The system comprises hardware equipment and a software system, wherein the hardware equipment comprises a visual acquisition module; a motion control module; a navigation positioning module; an environment sensing module; the software system includes; an image preprocessing module; a branch identification module; a three-dimensional reconstruction module; a trimming path planning module; and a cooperative control module. The invention relates to the technical field of artificial intelligence and cross-culture transmission, in particular to a pear tree branch real-time identification and collaborative pruning control system, which has the following advantages: 1, real-time accurate identification of pear tree branches is realized through a binocular camera and a deep learning algorithm; 2, the safety and efficiency of pruning are improved; 3, the adaptability and the intelligent level of the system are improved; and 4, cooperative work of hardware equipment and a software system is realized, the labor intensity is reduced, and the efficiency and the quality of pear tree pruning are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and cross-cultural communication technology, and in particular to a real-time recognition and collaborative pruning control system for pear tree branches. Background Art

[0002] Pear tree pruning is a crucial step in pear orchard management. Appropriate pruning can improve tree structure, fruit quality, and yield. Traditional pear tree pruning relies primarily on manual labor, which is labor-intensive, inefficient, and significantly impacts pruning quality. In recent years, with the advancement of intelligent agriculture, automated pruning equipment has emerged. However, these devices often suffer from inaccurate branch recognition and illogical pruning path planning, making them difficult to adapt to the complex pear orchard environment. Therefore, a system is urgently needed that can accurately identify pear tree branches in real time and implement coordinated pruning control. Summary of the Invention

[0003] In view of this, the present invention aims to provide a real-time identification and collaborative pruning control system for pear tree branches, to solve the problems of inaccurate identification of pear tree branches and unreasonable pruning path planning by existing automated pruning equipment, and to improve the efficiency and quality of pear tree pruning.

[0004] The technical solution of the embodiment of the present invention is achieved as follows:

[0005] A pear tree branch real-time identification and coordinated pruning control system, comprising hardware equipment and software systems: the hardware equipment includes:

[0006] A visual acquisition module, equipped with a binocular camera to collect three-dimensional image information of the pear tree;

[0007] A motion control module, which includes a servo motor, a reducer, and a transmission mechanism to control the movement of the trimming robot arm;

[0008] Navigation and positioning module, integrating GPS positioning system and inertial navigation system to achieve device positioning and navigation;

[0009] Mobile platform, carrying pruning robot arms and other equipment;

[0010] Environmental sensing module, including soil moisture sensors and meteorological sensors to collect soil moisture, temperature and meteorological data;

[0011] The software system includes:

[0012] Image preprocessing module, which performs denoising, enhancement and correction on binocular images;

[0013] The branch recognition module uses the improved YOLOv8 algorithm combined with the attention mechanism to identify branch features;

[0014] 3D reconstruction module, which builds 3D models of branches and trunks based on binocular vision stereo matching algorithm;

[0015] The pruning path planning module uses the improved A* algorithm to generate collision-free pruning paths;

[0016] Collaborative control module coordinates the collaborative work of hardware devices and software modules.

[0017] Preferably, the improved YOLOv8 algorithm of the branch recognition module is implemented by the following steps:

[0018] Extract features from the input image I to obtain a feature map F;

[0019] The attention weight A is calculated through the attention mechanism module, and the output feature map F′=F×A, where A is generated by the global average pooling and fully connected layer;

[0020] The recognition results were optimized by combining the pear tree growth characteristics database.

[0021] Preferably, the three-dimensional reconstruction module is based on binocular vision principle and calculates the depth information of branches and trunks through parallax:

[0022] Assume the focal length of the binocular camera is f, the baseline distance is b, and the parallax is d, then the depth Z = fb / d;

[0023] Multiple sets of parallax data are used to fit the three-dimensional coordinates of branches and trunks and construct a three-dimensional model.

[0024] Preferably, the improved A* algorithm of the pruning path planning module includes:

[0025] Input the 3D model of branches and trunks, pruning requirements, and the pear orchard environment map (including obstacle distribution);

[0026] Taking the shortest path without collision as the objective function, the optimal pruning path is generated through heuristic search.

[0027] Preferably, the collaborative mechanism of the collaborative control module includes:

[0028] Receive branch recognition results from the visual acquisition module, location data from the navigation and positioning module, and sensor environment data;

[0029] Sending a motion control instruction containing a trimming path to the motion control module;

[0030] Monitor the equipment status in real time and dynamically adjust the robot arm movements based on the pruning effect.

[0031] Preferably, the output data of the environment perception module is used to drive the adjustment of the pruning strategy:

[0032] When the soil moisture is lower than the threshold or the wind speed is greater than the safety value, the collaborative control module automatically reduces the movement speed of the pruning robot arm;

[0033] Optimize the exposure parameters of the binocular camera based on light intensity data.

[0034] Preferably, the workflow includes a real-time feedback mechanism:

[0035] During the pruning process, the binocular camera continuously collects images of the pruning area;

[0036] The branch recognition module determines the thickness and position of the remaining branches in real time;

[0037] The collaborative control module triggers the robotic arm's complementary shearing action based on the judgment results.

[0038] Preferably, the mechanical structure of the hardware device satisfies:

[0039] The mobile platform adopts wheeled or crawler structure, and the chassis ground clearance is ≥20cm to adapt to the terrain of the pear orchard;

[0040] The pruning robot arm is a 6-degree-of-freedom robot arm, and the end effector integrates electric pruning shears and force sensors.

[0041] Preferably, the module connection relationship of the software system is:

[0042] The visual acquisition module, image preprocessing module, branch recognition module, 3D reconstruction module, pruning path planning module, collaborative control module, and motion control module are sequentially connected;

[0043] The navigation and positioning module and the environmental perception module directly output data to the collaborative control module.

[0044] Preferably, the pear tree trunk pruning method of the system comprises the following steps: S1: determining the position of the device by a navigation and positioning module, and moving the mobile platform to the target pear tree;

[0045] S2: The binocular camera collects images, and after preprocessing, the branch recognition module extracts branch features;

[0046] S3: The 3D reconstruction module builds a 3D model of the branches, and the path planning module generates a pruning path;

[0047] S4: The collaborative control module drives the robotic arm to trim along the path and adjusts its movements in real time based on environmental data and trimming effects.

[0048] The embodiment of the present invention adopts the above technical solution, which has the following advantages:

[0049] 1. The present invention realizes real-time and accurate recognition of pear tree branches through binocular cameras and deep learning algorithms, improves the recognition accuracy of branches of different thicknesses and shapes, and solves the problem of inaccurate recognition of existing equipment.

[0050] Second, a binocular vision stereo matching algorithm is used to construct a three-dimensional model of the branches and trunks, and an improved A* algorithm is used to plan the pruning path, making the pruning path more reasonable, avoiding collisions between the robotic arm and other objects, and improving the safety and efficiency of pruning.

[0051] 3. The system integrates soil moisture sensors and meteorological sensors, which can adjust pruning strategies according to environmental data, adapt to different pear orchard environments and climatic conditions, and improve the adaptability and intelligence level of the system.

[0052] Fourth, the hardware equipment and software system work together, and the entire system has a high degree of automation, which reduces manual intervention, reduces labor intensity, and improves the efficiency and quality of pear tree pruning.

[0053] The above summary is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present invention will be readily apparent by reference to the accompanying drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 This is a diagram of the system hardware architecture of the present invention;

[0056] Figure 2 is a flow chart of the software modules of the present invention;

[0057] Figure 3 This is the collaborative control flow chart of the present invention. DETAILED DESCRIPTION

[0058] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and description are to be considered as illustrative in nature and not restrictive.

[0059] It should be noted that the terms "first," "second," "symmetrical," "array," etc. are used only to distinguish descriptions from positional descriptions and should not be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, limitations on features such as "first" and "symmetrical" may explicitly or implicitly include one or more of these features; similarly, when the number of certain features is not limited in the form of words such as "two" or "three," it should be noted that these features also explicitly or implicitly include one or more of the number of features.

[0060] In the present invention, unless otherwise expressly specified or limited, terms such as "installation," "connection," and "fixation" should be understood broadly; for example, they may refer to fixed connection, detachable connection, or integral molding; they may refer to mechanical connection, direct connection, welding, or indirect connection through an intermediate medium; they may refer to internal communication between two components or interaction between two components. Those skilled in the art will be able to understand the specific meanings of the above terms in the present invention based on the specification and drawings in conjunction with specific circumstances.

[0061] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0062] like Figure 1-3 The present invention provides a real-time identification and coordinated pruning control system for pear tree branches, including hardware equipment and software systems: the hardware equipment includes:

[0063] A visual acquisition module, equipped with a binocular camera to collect three-dimensional image information of the pear tree;

[0064] A motion control module, which includes a servo motor, a reducer, and a transmission mechanism to control the movement of the trimming robot arm;

[0065] Navigation and positioning module, integrating GPS positioning system and inertial navigation system to achieve device positioning and navigation;

[0066] Mobile platform, carrying pruning robot arms and other equipment;

[0067] Environmental sensing module, including soil moisture sensors and meteorological sensors to collect soil moisture, temperature and meteorological data;

[0068] The software system includes:

[0069] Image preprocessing module, which performs denoising, enhancement and correction on binocular images;

[0070] The branch recognition module uses the improved YOLOv8 algorithm combined with the attention mechanism to identify branch features;

[0071] 3D reconstruction module, which builds 3D models of branches and trunks based on binocular vision stereo matching algorithm;

[0072] The pruning path planning module uses the improved A* algorithm to generate collision-free pruning paths;

[0073] Collaborative control module coordinates the collaborative work of hardware devices and software modules.

[0074] like Figure 1-3 As shown in the figure, the improved YOLOv8 algorithm of the branch recognition module is implemented by the following steps:

[0075] Extract features from the input image I to obtain a feature map F;

[0076] The attention weight A is calculated through the attention mechanism module, and the output feature map F′=F×A, where A is generated by the global average pooling and fully connected layer;

[0077] Optimize the recognition results by combining the pear tree growth characteristics database;

[0078] The 3D reconstruction module is based on binocular vision and calculates the depth information of branches and trunks through parallax:

[0079] Assume the focal length of the binocular camera is f, the baseline distance is b, and the parallax is d, then the depth Z = fb / d;

[0080] Multiple sets of parallax data are used to fit the three-dimensional coordinates of branches and trunks and construct a three-dimensional model.

[0081] like Figure 1-3 As shown, the improved A* algorithm of the pruning path planning module includes:

[0082] Input the 3D model of branches and trunks, pruning requirements, and the pear orchard environment map (including obstacle distribution);

[0083] Taking the shortest path without collision as the objective function, the optimal pruning path is generated through heuristic search;

[0084] The collaborative mechanism of the collaborative control module includes:

[0085] Receive branch recognition results from the visual acquisition module, location data from the navigation and positioning module, and sensor environment data;

[0086] Sending a motion control instruction containing a trimming path to the motion control module;

[0087] Monitor the equipment status in real time and dynamically adjust the robot arm movements based on the pruning effect.

[0088] like Figure 1-3 As shown in Figure 2, the output data of the environment perception module is used to drive the pruning strategy adjustment:

[0089] When the soil moisture is lower than the threshold or the wind speed is greater than the safety value, the collaborative control module automatically reduces the movement speed of the pruning robot arm;

[0090] Optimize the exposure parameters of the binocular camera based on light intensity data;

[0091] The workflow includes real-time feedback mechanisms:

[0092] During the pruning process, the binocular camera continuously collects images of the pruning area;

[0093] The branch recognition module determines the thickness and position of the remaining branches in real time;

[0094] The collaborative control module triggers the robotic arm's complementary shearing action based on the judgment results.

[0095] like Figure 1-3 As shown, the mechanical structure of the hardware equipment meets the following requirements:

[0096] The mobile platform adopts wheeled or crawler structure, and the chassis ground clearance is ≥20cm to adapt to the terrain of the pear orchard;

[0097] The pruning robot arm is a 6-DOF robot arm, and the end effector integrates electric pruning shears and force sensors;

[0098] The module connection relationship of the software system is:

[0099] The visual acquisition module, image preprocessing module, branch recognition module, 3D reconstruction module, pruning path planning module, collaborative control module, and motion control module are sequentially connected;

[0100] The navigation and positioning module and the environmental perception module directly output data to the collaborative control module.

[0101] like Figure 1-3 As shown, the pear tree trunk pruning method includes the following steps: S1: determining the device position through a navigation and positioning module, and moving the mobile platform to the target pear tree;

[0102] S2: The binocular camera collects images, and after preprocessing, the branch recognition module extracts branch features;

[0103] S3: The 3D reconstruction module builds a 3D model of the branches, and the path planning module generates a pruning path;

[0104] S4: The collaborative control module drives the robotic arm to trim along the path and adjusts its movements in real time based on environmental data and trimming effects.

[0105] In this embodiment, the present invention is specifically designed to work as follows:

[0106] 1. Implementation of hardware equipment

[0107] 1. Visual acquisition module

[0108] A binocular camera (e.g., Basler acA2040-90um, 2048×2048 resolution, 90fps) was installed at the front end of the pruning robot (15–20cm from the end effector), with an optical axis angle of 12°, covering the robot's working range (radius 0.8–1.2m). Zhang's calibration method was used to obtain the camera's intrinsic parameters (focal length f = 5 mm, baseline distance b = 10 cm), providing a basis for parallax calculation.

[0109] 2. Motion control module

[0110] A servo motor (Panasonic A6 series, torque 0.5-2 N·m) connected to a reducer (reduction ratio 1:50) and a transmission mechanism drives a 6-DOF robotic arm (joint angle ±180°, repeatability ±0.1mm). The end effector integrates electric pruning shears (shearing force ≥ 50N) and a force sensor (range 0-100N, accuracy 0.1N) to provide real-time feedback on pruning resistance.

[0111] 3. Navigation and positioning module

[0112] GPS (Trimble R10, accuracy ±2cm) and inertial navigation (Xsens MTi-670) are installed on the top of the platform. Through Kalman filtering and data fusion, positioning (error ≤ 5cm) and attitude estimation (heading angle error ≤ 0.5°) are achieved, and the A* algorithm is supported to generate obstacle avoidance paths (obstacle avoidance distance ≥ 0.3m).

[0113] 4. Mobile Platform

[0114] Tracked chassis (ground pressure ≤ 50kPa, ground clearance 25cm), equipped with lithium battery (200Ah, endurance ≥ 8 hours), adapted to the pear garden terrain (softness ≤ 15kPa).

[0115] 5. Environmental Perception Module

[0116] A soil moisture sensor (Decagon5TE, accuracy ±2% volumetric water content) was inserted 10 cm into the soil, and a meteorological sensor (VaisalaWXT536, accuracy ±0.2°C, ±2% RH, ±0.1 m / s) was installed on the top of the platform with a refresh rate of 1 Hz.

[0117] 2. Implementation of the Software System

[0118] 1. Image preprocessing module

[0119] Based on Python+OpenCV, median filtering (3×3 kernel) is used for denoising, histogram equalization is used to enhance contrast (brightness mean ±10%), and epipolar correction is used to improve disparity calculation efficiency (30%).

[0120] 2. Branch Recognition Module

[0121] Improve the YOLOv8 algorithm (add CBAM attention), train 5000 annotated images (data enhancement: rotation ±15°, scaling 0.8-1.2 times), inference frame rate ≥ 20fps, mAP@0.5 ≥ 95% (GPU: RTX3090).

[0122] 3.3D reconstruction module

[0123] Semi-global matching (SGM) calculates the disparity d and depth \(Z = \frac{fb}{d}\), and Poisson reconstruction generates a triangular mesh model (accuracy ≤ 1mm, point cloud resolution 0.5mm / pixel) and stores it in .obj format.

[0124] 4. Pruning path planning module

[0125] An improved A* algorithm (dynamic weights prioritize retaining main branches) was used to input the branch model, pruning rules (main branch 1.5-2m, side branch spacing ≥30cm) and environment map (obstacle spacing ≥1m), and the robot arm trajectory (joint coordinates, interpolation step size 0.1s, speed ≤0.2m / s) was output. The GJK algorithm was used for real-time collision detection (100% collision-free).

[0126] 5. Collaborative control module

[0127] Built on C++ / ROS, the data flow is: vision → preprocessing → recognition → reconstruction → planning → control → motion, with a control frequency of 100Hz. Dynamic adjustments are made based on force sensors (angle adjustment when resistance > 30N) and environmental data (50% speed reduction when wind speed > 5m / s).

[0128] 3. Implementation of system workflow

[0129] 1. Initialization and positioning

[0130] The navigation module was initialized within 30 s, and the mobile platform drove along the GIS grid path (spacing 3 m) to reach the target pear tree (distance ≤ 0.5 m).

[0131] 2. Image Acquisition and Processing

[0132] The binocular camera collects data at 20fps, recognizes branches (bounding box) in 0.05s after preprocessing, and generates a 3D model in 0.1s (point cloud ≥ 10 5 indivual).

[0133] 3. Path planning and pruning

[0134] The pruning path (3-5 points) is generated in 0.2s, the inverse kinematics solution is 0.01s, and the robotic arm completes single-branch pruning (shearing force ≤ 40N) within 2s.

[0135] 4. Real-time feedback and adjustments

[0136] 20fps monitoring, residual branches (diameter > 0.5cm, length > 10cm) trigger re-pruning (path adjustment within 0.5s), soil moisture < 15% is recorded in the log to prompt irrigation.

[0137] 4. Performance Verification

[0138] Recognition accuracy: mAP@0.5 for branches ≥ 95%, recognition rate for diseased branches ≥ 90%;

[0139] Pruning efficiency: ≤3 minutes for a single pear tree (10-15 minutes for manual labor);

[0140] Collision rate: 0 times (path planning + real-time monitoring);

[0141] Environmental adaptability: Wind speed ≤ 8m / s, soil moisture 5%-30% stable operation, main branch retention rate ≥ 95%, diseased branch removal rate ≥ 98%.

[0142] The following are several other specific embodiments of the present invention:

[0143] Example 1: Intelligent pruning of a pear orchard

[0144] Scenario: In a solar greenhouse pear orchard, space is limited (row spacing 2.5m, plant spacing 1.5m), and meticulous pruning is required.

[0145] 1. Hardware adaptation

[0146] The mobile platform adopts a narrow body design (width ≤ 1.2m), and the width of the robotic arm is ≤ 0.8m after folding. It avoids obstacles (spacing ≥ 0.3m) through lidar (SLAM mapping, accuracy ± 5cm).

[0147] The binocular camera is equipped with infrared fill light (wavelength 850nm, illumination ≥500lux) to solve the image acquisition problem in the low-light environment of the greenhouse (light ≤2000lux).

[0148] 2. Software optimization

[0149] The branch recognition model adds a "low-light enhancement" branch (Retinex algorithm preprocessing), and the mAP@0.5 is increased to 94% in low light (normally 88%).

[0150] The path planning module introduces "greenhouse space constraints" (the range of movement of the robotic arm is ≤1.5m 3), and the RRT* algorithm is used to generate a compact path (turning radius ≤ 0.5m).

[0151] 3. Pruning strategy

[0152] The density of branches of the facility pear tree is high, so cross branches (angle <30°) and drooping branches (length >0.5m) should be removed first, and the middle and upper fruiting branches (length 0.3-0.6m) should be retained.

[0153] The collaborative control module combines the CO2 concentration in the greenhouse (accelerates pruning when ≥800ppm to reduce personnel exposure) to improve operational safety.

[0154] Example 2: Intelligent Identification and Pruning of Pear Tree Trunk Diseases

[0155] Scenario: Identify diseases such as black spot and ring rot on pear tree branches and remove diseased branches in a targeted manner.

[0156] 1. Disease identification module

[0157] A multi-task learning model (YOLOv8+U-Net) was trained to simultaneously detect branch positions (mAP@0.5≥95%) and diseased areas (segmentation accuracy≥90%, Dice coefficient≥0.85).

[0158] Construct a disease feature library (lesion color, texture, and distribution pattern) and input meteorological data (enhance disease detection weight when humidity is >80%).

[0159] 2. Pruning Decisions

[0160] Diseased branch removal strategy: Cut off branches with black spot disease from the base (5 cm away from the main trunk), and cut off branches with ring rot disease to a length ≥ 3 cm above the lesion to avoid recurrence.

[0161] The collaborative control module records the location of diseased branches and generates a disease heat map (GIS annotation) to provide data support for subsequent spraying operations (lesion density>5 / m 2 Focus on spraying when necessary).

[0162] 3. Effect verification

[0163] The diseased branch identification rate is ≥92%, and the disease recurrence rate after pruning is reduced by 18%, improving the health management level of the pear orchard.

[0164] Example 3: Annual pruning management of pear trees

[0165] Scenario: Differentiated pruning needs in spring (before budding), summer (growing period), and winter (dormant period).

[0166] 1. Seasonal parameter library

[0167] Spring: retain flower buds (diameter > 0.8 cm, spacing ≥ 10 cm), remove weak flower buds (diameter < 0.5 cm), and prioritize smooth cuts (to reduce sap flow) during path planning.

[0168] Summer: Control vigorous growth of branches (length > 1.2m, angle < 45°), adopt "top-cutting" pruning (retain 2-3 leaves at the end), and switch the end effector of the robot arm to tenderloin shears (shearing force ≤ 10N);

[0169] Winter: heavy pruning and renewal (remove old branches and retain 1-2 year old branches), apply healing agent to the cut ends (mechanical arm integrated application device, application accuracy ±2mm).

[0170] 2. System linkage

[0171] The calendar module automatically switches season modes, and sensors (such as bud recognition sensors with an accuracy of ±0.2mm) assist in determining the state of flower buds, improving targeted pruning.

[0172] After winter pruning, the navigation module guides the equipment to synchronously complete the whitewashing of tree trunks (integrated whitewashing device, whitewashing height 1.2-1.5m) to enhance cold resistance.

[0173] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various modifications and substitutions within the technical scope disclosed in the present invention, and such modifications and substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A real-time identification and coordinated pruning control system for pear tree branches, characterized in that: Including hardware equipment and software system: the hardware equipment includes: A visual acquisition module, equipped with a binocular camera to collect three-dimensional image information of the pear tree; A motion control module, which includes a servo motor, a reducer, and a transmission mechanism to control the movement of the trimming robot arm; Navigation and positioning module, integrating GPS positioning system and inertial navigation system to achieve device positioning and navigation; Mobile platform, carrying pruning robot arms and other equipment; Environmental sensing module, including soil moisture sensors and meteorological sensors to collect soil moisture, temperature and meteorological data; The software system includes: Image preprocessing module, which performs denoising, enhancement and correction on binocular images; The branch recognition module uses the improved YOLOv8 algorithm combined with the attention mechanism to identify branch features; 3D reconstruction module, which builds 3D models of branches and trunks based on binocular vision stereo matching algorithm; The pruning path planning module uses the improved A* algorithm to generate collision-free pruning paths; Collaborative control module coordinates the collaborative work of hardware devices and software modules.

2. A pear tree branch real-time identification and coordinated pruning control system according to claim 1, characterized in that: The improved YOLOv8 algorithm of the branch recognition module is implemented by the following steps: Extract features from the input image I to obtain a feature map F; The attention weight A is calculated through the attention mechanism module, and the output feature map F′=F×A, where A is generated by the global average pooling and fully connected layer; The recognition results were optimized by combining the pear tree growth characteristics database.

3. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The 3D reconstruction module is based on binocular vision principle and calculates the depth information of branches through parallax: Assume the focal length of the binocular camera is f, the baseline distance is b, and the parallax is d, then the depth Z = fb / d; Multiple sets of parallax data are used to fit the three-dimensional coordinates of branches and trunks and construct a three-dimensional model.

4. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The improved A* algorithm of the pruning path planning module includes: Input the 3D model of branches and trunks, pruning requirements and the pear orchard environment map; Taking the shortest path without collision as the objective function, the optimal pruning path is generated through heuristic search.

5. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The collaborative mechanism of the collaborative control module includes: Receive branch recognition results from the visual acquisition module, location data from the navigation and positioning module, and sensor environment data; Sending a motion control instruction containing a trimming path to the motion control module; Monitor the equipment status in real time and dynamically adjust the robot arm movements based on the pruning effect.

6. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The output data of the environment perception module is used to drive the adjustment of the pruning strategy: When the soil moisture is lower than the threshold or the wind speed is greater than the safety value, the collaborative control module automatically reduces the movement speed of the pruning robot arm; Optimize the exposure parameters of the binocular camera based on light intensity data.

7. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The described workflow includes a real-time feedback mechanism: During the pruning process, the binocular camera continuously collects images of the pruning area; The branch recognition module determines the thickness and position of the remaining branches in real time; The collaborative control module triggers the robotic arm's complementary shearing action based on the judgment results.

8. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The mechanical structure of the hardware device meets the following requirements: The mobile platform adopts wheeled or crawler structure, and the chassis ground clearance is ≥20cm to adapt to the terrain of the pear orchard; The pruning robot arm is a 6-degree-of-freedom robot arm, and the end effector integrates electric pruning shears and force sensors.

9. The real-time identification and coordinated pruning control system for pear tree branches according to claim 1, characterized in that: The module connection relationship of the software system is: The visual acquisition module, image preprocessing module, branch recognition module, 3D reconstruction module, pruning path planning module, collaborative control module, and motion control module are sequentially connected; The navigation and positioning module and the environmental perception module directly output data to the collaborative control module.

10. A method for pruning pear tree trunks based on the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1: The device location is determined by the navigation and positioning module, and the mobile platform drives to the target pear tree; S2: The binocular camera collects images, and after preprocessing, the branch recognition module extracts branch features; S3: The 3D reconstruction module builds a 3D model of the branches, and the path planning module generates a pruning path; S4: The collaborative control module drives the robotic arm to trim along the path and adjusts its movements in real time based on environmental data and trimming effects.

Images