Automatic fruit tree pruning system based on machine vision

Abstract

The present invention relates to the technical field of fruit tree pruning, and in particular to an automatic fruit tree pruning system based on machine vision, comprising a camera for capturing images of fruit trees and determining the specific positions of fruits and branches and leaves. An image processing module is used to identify the structure of branches and trunks through image processing technology, and identify branches and leaves that need to be pruned. A decision-making module is used to calculate the cutting path and driving path of a pruner and a driving vehicle from a current position to a target position based on the results of the image processing module. An intelligent control module is used to control the pruner and the driving vehicle to move to a target position to prune branches and leaves based on information provided by the decision-making module, while avoiding fruits and branches and leaves that do not need to be pruned during the pruning process. The present invention accurately identifies the branch and trunk structure of the fruit tree, adjusts the angle and force of the pruner in real time, avoids fruits and branches and leaves that do not need to be pruned, makes pruning more precise, and helps maintain the healthy growth of the fruit tree.

Description

Technical Field

[0001] The present invention relates to the technical field of fruit tree pruning, and in particular to an automatic fruit tree pruning system based on machine vision. Background Art

[0002] Pruning is an essential part of fruit tree growth. Pruning involves cutting away branches that are detrimental to the tree's growth. This process is crucial for improving light distribution and gas exchange at the canopy and orchard scales, balancing the tree's nutritional and reproductive growth, and promoting the implementation of cultivation and harvesting measures. However, fruit tree pruning requires specialized agricultural technicians, making it an expensive and labor-intensive task. Consequently, various types of fruit tree pruning machines have emerged.

[0003] In existing fruit tree pruning technologies, a rotary cutting mechanism is a common design. This mechanism uses high-speed rotating blades to complete pruning tasks. Its advantage is that it can quickly prune large areas and is suitable for large-scale operations in orchards. However, this design also has some obvious limitations and problems. For example, it is difficult to achieve precise control when pruning with a rotating blade, and it is easy to "cut across the board", that is, all branches are cut off regardless of their size or position. This means that during the operation, not only will the parts that need to be pruned be removed, but some healthy branches and leaves that do not need to be pruned may also be cut off, which may also cause damage to the growing fruit.

[0004] In summary, how to solve the problem of damage to fruit trees caused by pruning fruit trees through a rotary cutting mechanism in the existing technology has become a difficult problem that needs to be solved urgently in this field. Therefore, it is necessary to propose an automatic fruit tree pruning system based on machine vision. Summary of the Invention

[0005] To solve the above problems, the present invention provides an automatic fruit tree pruning system based on machine vision. By accurately identifying the branch structure of the fruit tree, the angle and force of the pruner are adjusted in real time to avoid the fruits and branches and leaves that do not need to be pruned, making pruning more precise and helping to maintain the healthy growth of the fruit trees.

[0006] In order to achieve the above-mentioned purpose, the technical solution of the present invention is as follows: A machine vision-based automatic fruit tree pruning system includes an image processing module, a decision-making module, an intelligent control module, a camera, a driving vehicle and a pruner.

[0007] The camera is used to capture images of fruit trees and determine the specific locations of fruits and branches and leaves.

[0008] The image processing module is used to identify the branch structure through image processing technology and identify the branches and leaves that need to be pruned.

[0009] The decision-making module is used to calculate the cutting path and driving path of the trimmer and driving vehicle from the current position to the target position based on the results of the image processing module.

[0010] The intelligent control module is used to control the pruning machine and the driving vehicle to move to the target position to prune branches and leaves according to the cutting path and driving path information provided by the decision-making module, while avoiding fruits and branches and leaves that do not need to be pruned during the pruning process.

[0011] The technical principles of the above solution are as follows:

[0012] The camera, mounted on the drive vehicle, moves with the vehicle, capturing real-time images of the fruit trees. These images provide crucial information about the tree's trunk structure and fruit location. The image processing module receives image data from the camera and processes it using computer vision technology. This allows the system to identify branches and leaves that require pruning and those that should be retained. The decision-making module calculates the optimal pruning path based on the information provided by the image processing module and incorporating principles and standards for fruit tree pruning. After receiving the path information output by the decision-making module, the intelligent control module coordinates and controls the movements of the pruners and the drive vehicle, ensuring that the pruners accurately reach their target locations for pruning and avoiding unnecessary damage to the trees in the process.

[0013] The above scheme has the following beneficial effects:

[0014] 1. This invention accurately identifies the branch structure of fruit trees, determines the specific location of fruits and branches and leaves, and uses image processing technology to accurately distinguish branches and leaves that need to be pruned from those that need to be retained, avoiding the mis-pruning that may occur with traditional pruning methods. Dynamically adjust the pruning strategy according to the actual situation of the fruit tree, making pruning more precise and helping to maintain the healthy growth of the fruit tree.

[0015] 2. During the fruit tree pruning process, the present invention ensures pruning accuracy and safety by adjusting the angle and force of the pruner in real time. At the same time, based on the feedback from the camera, the pruner can promptly avoid the fruit and branches that do not need to be pruned, thus avoiding unnecessary damage.

[0016] 3. The present invention realizes a high degree of automation in the entire process from image acquisition to pruning execution, which reduces the influence of human factors and makes pruning operations more efficient; it also reduces dependence on manual pruning and reduces the labor cost of orchard management.

[0017] Furthermore, the branches and leaves that need to be pruned in the image processing module include but are not limited to crossing branches, overlapping branches, weak branches, dry branches, diseased and insect-infested branches, overgrown branches, inner branches, competing branches, drooping branches and overcrowded branches.

[0018] Beneficial effects: By identifying diseased and weak branches of fruit trees, pruning operations can be more targeted, which can optimize the tree structure, improve ventilation and light conditions, and be conducive to the healthy growth of fruit trees, thereby improving the quality of fruit trees.

[0019] Furthermore, the top of the driving vehicle is fixedly connected to a shell, and the bottom wall of the shell is rotatably engaged with a ball joint; the top of the ball joint is fixedly connected to a hydraulic rod, and the end of the hydraulic rod away from the ball joint extends from the inside of the shell to the outside of the shell; an opening for the movement of the hydraulic rod is opened on the top of the shell, and a movable rubber ring is fixedly connected to the opening; the trimmer and the camera are both fixedly connected to the output shaft of the hydraulic rod; the inside of the shell is provided with a lateral adjustment component and a longitudinal adjustment component for driving the hydraulic rod to move laterally and longitudinally, respectively.

[0020] Beneficial effect: The design of the spherical joint and the inner bottom wall of the shell is used to rotate together, so that the hydraulic rod can move flexibly in multiple directions, thereby meeting the operating requirements of the trimmer and camera at different angles.

[0021] Furthermore, the lateral adjustment assembly includes a controller and a first drive member, the controller is used to control the operation of the hydraulic rod and the first drive member; the first drive member is fixedly connected to the inner bottom wall of the outer shell, and the output shaft of the first drive member is fixedly connected to a semicircular adjustment plate; the adjustment plate has an adjustment groove for the movement of the hydraulic rod along its length, and the inner wall of the adjustment plate is slidably engaged with the spherical joint; the end of the adjustment plate away from the first drive member is rotatably engaged with the inner wall of the outer shell.

[0022] Beneficial effect: The first driving member drives the adjustment plate to rotate back and forth. Since the adjustment plate is semicircular and slides with the spherical joint, when the adjustment plate rotates back and forth laterally, it can drive the hydraulic rod to rotate accordingly. After moving to the specified position, the hydraulic rod is extended to make the trimmer and camera reach the preset position, thereby realizing the adjustment of the position of the trimmer and camera and improving working efficiency.

[0023] Furthermore, the longitudinal adjustment assembly includes a second driving member, and the controller is used to control the operation of the second driving member; the output shaft of the second driving member is fixedly connected to a semicircular moving plate, and the moving plate is installed vertically with the adjustment plate; the inner side wall of the moving plate is slidably engaged with the adjustment plate, and the moving plate is provided with a sliding groove along its length for the movement of the hydraulic rod; the second driving member is fixedly connected to the inner side wall of the outer shell, and the end of the moving plate away from the second driving member is rotatably engaged with the inner side wall of the outer shell.

[0024] Beneficial effects: The moving plate is driven to rotate back and forth by the second driving member. Since the moving plate is semicircular and slides with the adjusting plate, the moving plate and the adjusting plate are installed vertically; therefore, the hydraulic rod can be made to perform longitudinal reciprocating motion through the longitudinal reciprocating motion of the moving plate; by combining with the adjusting plate, the hydraulic rod can be rotated in any direction, thereby improving the adaptability of the equipment; ensuring that the trimmer and camera can accurately reach the required position, improving the operation accuracy and efficiency.

[0025] Furthermore, the trimmer includes scissors, an arc-shaped plate and a straight rod, the straight rod is fixedly connected to the top of the hydraulic rod output shaft; the arc-shaped plate is fixedly connected to the top of the straight rod, and the scissors are hinged to the top of the straight rod; a driving assembly for driving the scissors to rotate is provided on the straight rod.

[0026] Beneficial effects: By driving the scissors to rotate, the scissors can rotate and close during the pruning process, thereby pruning branches and leaves; the automation efficiency of the device is increased, thereby reducing labor input.

[0027] Furthermore, the drive assembly includes a third drive member, and the controller is used to control the operation of the third drive member; the third drive member is fixedly connected to the outer wall of the straight rod, and the output shaft of the third drive member is coaxially fixedly connected to the first gear, and the first gear is engaged with a fan-shaped gear; the side of the fan-shaped gear away from the first gear is fixedly connected to the scissors.

[0028] Beneficial Effect: The third driving member drives the first gear to rotate, which in turn drives the meshing sector gear to rotate. Since the sector gear is fixedly connected to the scissors, the scissors can be closed when the sector gear rotates. This design allows the scissors to be closed accurately when needed, thereby improving trimming efficiency.

[0029] Furthermore, the outer wall of the straight rod away from the third driving member is fixedly connected to a pump assembly for sucking branches and leaves, and the controller is used to control the operation of the pump assembly; the input end of the pump assembly is fixedly connected to an air intake port, and the inner wall of the air intake port is fixedly connected to a buffer layer.

[0030] Beneficial Effects: The pump assembly generates negative pressure to suck the branches and leaves to be pruned; it can quickly secure lighter branches and leaves, making pruning more stable. This effectively prevents branches and leaves from shaking and moving during pruning, thereby improving pruning accuracy and efficiency.

[0031] Furthermore, a movable baffle is hingedly connected to the outer wall of the straight rod away from the pump assembly, and a rotating assembly for rotating the movable baffle is provided on the straight rod.

[0032] Beneficial effect: For branches and leaves that do not need to be pruned, they can be avoided through the movable baffle to avoid accidental pruned or damaged, making the pruning work more flexible and accurate.

[0033] Furthermore, the rotating assembly includes a fourth driving member, and the controller is used to control the operation of the fourth driving member; the fourth driving member is embedded and installed on the straight rod, and the output shaft of the fourth driving member is fixedly connected to the movable baffle.

[0034] Beneficial effects: The movable baffle is driven to rotate by the fourth driving member, thereby realizing adjustment of the angle of the movable baffle; the automated operation mode is utilized to make the trimming work smoother and optimize the overall workflow.

[0035] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is an axonometric diagram of the machine vision-based automatic fruit tree pruning system in an embodiment of the present invention.

[0037] Figure 2 It is a front cross-sectional view of the housing in an embodiment of the present invention.

[0038] Figure 3 2 is a side sectional view of the housing in an embodiment of the present invention.

[0039] Figure 4 It is a front cross-sectional view of a trimmer in an embodiment of the present invention.

[0040] Figure 5 2 is a side view of the movable baffle in an embodiment of the present invention.

[0041] Figure 6 For the embodiment of the present invention Figure 1 Enlarged view of part A.

[0042] The figure marks in the drawings of the specification include: 1. camera; 2. drive vehicle; 3. housing; 4. ball joint; 5. hydraulic rod; 6. first motor; 7. adjustment plate; 8. second motor; 9. moving plate; 10. scissors; 11. curved plate; 12. straight rod; 13. third motor; 14. first gear; 15. sector gear; 16. air pump; 17. air inlet; 18. movable baffle; 19. fourth motor. DETAILED DESCRIPTION

[0043] The following is further described in detail through specific implementation methods:

[0044] Example 1:

[0045] As attached Figures 1-6 As shown: A fruit tree automatic pruning system based on machine vision, including an image processing module, a decision-making module, an intelligent control module, a camera 1, a driving vehicle 2 and a pruner.

[0046] Camera 1 is primarily used to capture images of fruit trees and determine the specific locations of fruit and branches. The image processing module uses image processing technology to identify branch structure and branches that require pruning. The decision-making module calculates the cutting and travel paths for the pruner and drive vehicle 2 from their current position to the target location based on the image processing module's results. The intelligent control module uses the cutting and travel path information provided by the decision-making module to control the pruner and drive vehicle 2 to move to the target location for pruning, avoiding fruit and branches that do not require pruning.

[0047] The following is a detailed explanation of the functions of each module and component:

[0048] The camera 1 is mounted on a driving vehicle 2 and can move with the vehicle to capture images of the fruit trees in real time; these images provide important information such as the structure of the branches and trunks of the fruit trees and the location of the fruits.

[0049] The image processing module receives image data from camera 1 and processes the image using computer vision techniques, such as image segmentation, feature extraction, and object recognition. Through these processes, the system can identify branches and leaves that need to be pruned and those that need to be retained, thus avoiding the potential for mis-pruning with traditional pruning methods. Dynamically adjusting pruning strategies based on the actual condition of the fruit tree ensures more precise pruning and helps maintain healthy growth.

[0050] Specifically, computer vision technology uses the SIFT feature extraction method to extract characteristics of branches and leaves, including edge, texture, and color information, and uses a convolutional neural network (CNN) for feature learning and classification. The model is trained with a large amount of labeled data, enabling it to accurately identify different types of branches and leaves.

[0051] U-Net semantic segmentation technology is used to segment the image into different parts, generating pixel-level label maps. Mask R-CNN instance segmentation technology is then used to individually identify and segment each branch, ensuring clear boundaries. Multi-scale feature extraction and the output of the deep learning model are combined to improve recognition accuracy and robustness.

[0052] The branches and leaves that need to be pruned include, but are not limited to, crossing branches, overlapping branches, weak branches, dead branches, diseased and insect-infested branches, overgrown branches, inner branches, competing branches, drooping branches, and overcrowded branches. In this embodiment, diseased and insect-infested branches are selected. By identifying diseased and weak branches in fruit trees, pruning operations can be more targeted, optimizing tree structure, improving ventilation and light conditions, and promoting healthy growth of fruit trees, thereby improving the quality of the fruit trees.

[0053] Specifically, using the CNN algorithm for comparative analysis, by detecting the intersection points and angles between branches, crossing branches are identified. By detecting the overlapping areas between branches, overlapping branches are identified. By detecting the thickness and color of branches, weak branches are identified. By detecting the color and texture of branches, dead branches are identified. By detecting abnormal spots and textures on branches, diseased and insect-infested branches are identified. By detecting the growth direction and length of branches, overgrown branches are identified. By detecting the position and lighting conditions of branches, ingrown branches are identified. By detecting the competitive relationship between branches (such as distance and angle), competing branches are identified. By detecting the degree of bending and direction of branches, drooping branches are identified. By detecting the density and distribution of branches, overcrowded branches are identified.

[0054] The decision-making module calculates the optimal pruning path based on the information provided by the image processing module and in combination with the principles and standards of fruit tree pruning; including determining the location that the pruner needs to reach and the route that the driving vehicle 2 needs to travel.

[0055] Specifically, a series of pruning rules were developed based on professional knowledge and experience in horticultural pruning. For example, crossing and overlapping branches, which may affect the tree's growth and ventilation and light transmission, should be pruned, while strong and healthy branches should be retained.

[0056] The decision-making process also considers factors such as the tree's growing environment, soil type, and climatic conditions to ensure scientific and rational pruning decisions. Each branch is assigned a priority score based on its type and location, with branches with high scores being pruned first. For example, diseased and insect-infested branches and dead branches are given the highest priority, followed by crossing and overlapping branches, and finally weak and overgrown branches.

[0057] After receiving the path information output by the decision-making module, the intelligent control module controls the movements of the pruner and drive vehicle 2. This ensures that the pruner accurately reaches the target location for pruning, adjusting the pruner's angle and force in real time to ensure precision and safety. Based on feedback from camera 1, the module promptly avoids fruit and branches that do not require pruning, preventing unnecessary damage to the trees. From image acquisition to pruning execution, the entire process is highly automated, reducing the impact of human intervention and making pruning more efficient. It also reduces reliance on manual pruning, lowering the labor costs of orchard management.

[0058] Specifically, the Dijkstra algorithm is used for path planning, generating the shortest path based on the current and target positions of the pruners. Taking into account the motion constraints of the mechanical shears and the complexity of the orchard environment, the RRT (Rapidly-exploring Random Trees) algorithm is used for path planning to ensure the feasibility and safety of the path.

[0059] During path planning, obstacles in the orchard (such as other branches, fruit, and supports) are considered, and the RVO dynamic obstacle avoidance algorithm is used to ensure that the pruner does not collide with obstacles during movement. The multi-objective optimization algorithm NSGA-II is used to simultaneously consider the length, time, and safety of the pruning path to generate the optimal pruning path.

[0060] Finally, during the pruning process, the intelligent control module monitors the position and status of the mechanical shears in real time and makes adaptive adjustments based on the actual situation. For example, when encountering thick branches, the intelligent control module adjusts the cutting force and speed of the pruner to ensure the pruning effect and quality.

[0061] Example 2:

[0062] As attached Figure 1-Figure 3 As shown, the difference from the above embodiment is that the top of the driving vehicle 2 is fixedly connected to the shell 3 with bolts, and the inner bottom wall of the shell 3 is rotatably engaged with a ball joint 4; the top of the ball joint 4 is fixedly connected to the hydraulic rod 5 with bolts, and the top of the hydraulic rod 5 extends from the inside of the shell 3 to the outside of the shell 3; an opening for the movement of the hydraulic rod 5 is opened at the top of the shell 3, and a movable rubber ring is fixedly bonded to the opening; the trimmer and the camera 1 are both fixedly connected to the output shaft of the hydraulic rod 5 with bolts; the inside of the shell 3 is respectively provided with a lateral adjustment component and a longitudinal adjustment component for driving the hydraulic rod 5 to move laterally and longitudinally.

[0063] like Figure 2 As shown, the lateral adjustment assembly includes a controller and a first driving member. In this embodiment, the first driving member is a first motor 6; the controller is used to control the operation of the hydraulic rod 5 and the first motor 6; the first motor 6 is fixedly connected to the inner bottom wall of the shell 3 by bolts, and the output shaft of the first motor 6 is fixedly connected to a semicircular adjustment plate 7 by bolts; the adjustment plate 7 is provided with an adjustment groove for the movement of the hydraulic rod 5 along its length, and the inner wall of the adjustment plate 7 is in sliding cooperation with the spherical joint 4; the left end of the adjustment plate 7 is in rotational cooperation with the inner wall of the shell 3.

[0064] like Figure 3 As shown, the longitudinal adjustment assembly includes a second driving member. In this embodiment, the second driving member is a second motor 8, and the controller is used to control the operation of the second motor 8; the output shaft of the second motor 8 is bolted and fixedly connected with a semicircular moving plate 9, and the moving plate 9 is installed vertically with the adjustment plate 7; the inner wall of the moving plate 9 is slidably matched with the adjustment plate 7, and the moving plate 9 is provided with a sliding groove along its length for the movement of the hydraulic rod 5; the second motor 8 is bolted and fixedly connected to the inner wall of the outer shell 3, and the left end of the moving plate 9 is rotatably matched with the inner wall of the outer shell 3.

[0065] The specific implementation process is as follows: Figure 2As shown, the first motor 6 drives the adjustment plate 7 to rotate back and forth. Since the adjustment plate 7 is semicircular and slides with the ball joint 4, the moving plate 9 has a sliding groove along its length for the hydraulic rod 5 to move. Therefore, when the adjustment plate 7 rotates back and forth horizontally, it can drive the hydraulic rod 5 to rotate accordingly. After moving to the specified position, the hydraulic rod 5 is extended to make the trimmer and camera 1 reach the preset position, thereby realizing the adjustment of the position of the trimmer and camera 1 and improving the working efficiency. Figure 1 and Figure 3 For example, when the first motor 6 drives the adjusting rod to move to the right, the hydraulic rod 5 can move to the right; otherwise, the hydraulic rod 5 can move to the left.

[0066] like Figure 3 As shown, the moving plate 9 is driven by the second motor 8 to rotate back and forth. Since the moving plate 9 is semicircular and slides with the adjusting plate 7, the moving plate 9 is installed vertically to the adjusting plate 7, and the adjusting plate 7 has an adjusting groove for the movement of the hydraulic rod 5 along its length. Therefore, the longitudinal reciprocating motion of the moving plate 9 can make the hydraulic rod 5 perform longitudinal reciprocating motion; by combining with the adjusting plate 7, the hydraulic rod 5 can be rotated in any direction, thereby improving the adaptability of the equipment; ensuring that the trimmer and camera 1 can accurately reach the required position, improving the operation accuracy and efficiency.

[0067] by Figure 1 and Figure 2 For example, when the second motor 8 drives the motion rod to the right, the hydraulic rod 5 can be moved to the right; conversely, the hydraulic rod 5 can be moved to the left. The design of the spherical joint 4 and the inner bottom wall of the housing 3 rotate together, allowing the hydraulic rod 5 to flexibly move in multiple directions, thereby meeting the operating requirements of the trimmer and camera 1 at different angles.

[0068] Example 3:

[0069] As attached Figure 1 、 Figure 4 and Figure 6 As shown, the difference from the above embodiment is that the trimmer includes scissors 10, an arc-shaped plate 11 and a straight rod 12, the straight rod 12 is bolted to the top of the output shaft of the hydraulic rod 5; the arc-shaped plate 11 is bolted to the top of the straight rod 12, and the scissors 10 are hinged to the top of the straight rod 12; the straight rod 12 is provided with a driving assembly for driving the scissors 10 to rotate.

[0070] like Figure 4As shown, the drive assembly includes a third drive member. In this embodiment, the third drive member is a third motor 13, and the controller is used to control the operation of the third motor 13; the third motor 13 is bolted fixedly connected to the outer wall of the straight rod 12, and the output shaft of the third motor 13 is coaxially bolted fixedly connected to the first gear 14, and the first gear 14 is engaged with a fan-shaped gear 15; the right side of the fan-shaped gear 15 is bolted fixedly connected to the scissors 10.

[0071] The specific implementation process is as follows: the third motor 13 drives the first gear 14 to rotate, and the first gear 14 drives the fan gear 15 meshing with it to rotate; since the fan gear 15 is fixedly connected to the scissors 10 with bolts, when the fan gear 15 rotates, the scissors 10 can be closed. Figure 4 For example, when the first gear 14 rotates clockwise, it drives the sector gear 15 to rotate counterclockwise. Counterclockwise rotation drives the scissors 10 downward, thereby closing the scissors 10 downward; conversely, it opens the scissors 10 upward. This design allows the scissors 10 to close accurately when needed, thereby improving pruning efficiency. By driving the scissors 10 to rotate, the scissors 10 rotate and close during the pruning process, thereby achieving the desired pruning of branches and leaves, increasing the automation efficiency of the device and reducing labor input.

[0072] Example 4:

[0073] As attached Figure 1 and Figure 6 As shown, the difference from the above embodiment is that the outer wall on the left side of the straight rod 12 is fixedly connected with a pump assembly for sucking branches and leaves by bolts. In this embodiment, the pump assembly is an air pump 16, and the controller is used to control the operation of the air pump 16; the input end of the air pump 16 is fixedly connected with an air intake port 17 by bolts, and a buffer layer is fixedly bonded to the inner wall of the air intake port 17.

[0074] The specific implementation process is as follows: the air pump 16 generates negative pressure to suck the branches and leaves to be pruned; for lighter branches and leaves, it can quickly fix them, making the pruning more stable. It effectively prevents the branches and leaves from shaking and moving during the pruning process, thereby improving the pruning accuracy and efficiency.

[0075] Example 5:

[0076] As attached Figure 5 and Figure 6 As shown, the difference from the above embodiment is that a movable baffle 18 is hingedly connected to the outer wall on the right side of the straight rod 12, and a rotating assembly for rotating the movable baffle 18 is provided on the straight rod 12.

[0077] The rotating assembly includes a fourth driving member. In this embodiment, the fourth driving member is a fourth motor 19. The controller is used to control the operation of the fourth motor 19. The fourth motor 19 is bolted and embedded in the straight rod 12. The output shaft of the fourth motor 19 is bolted and fixedly connected to the movable baffle 18.

[0078] The specific implementation process is as follows: the fourth motor 19 drives the movable baffle 18 to rotate, thereby adjusting the angle of the movable baffle 18. The use of an automated operation method makes pruning more smooth and optimizes the overall workflow. For branches and leaves that do not need to be pruned, the movable baffle 18 can avoid them, avoiding accidental cutting or damage, making pruning more flexible and accurate.

[0079] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A machine vision-based automatic pruning system for fruit trees, comprising an image processing module, a decision-making module, an intelligent control module, a camera (1), a driving vehicle (2) and a pruner, characterized in that: Camera (1), used to capture images of fruit trees and determine the specific locations of fruits and branches and leaves; An image processing module, for receiving image data transmitted from the camera (1) and processing the image using computer vision technology; Computer vision technology uses the SIFT feature extraction method to extract the features of branches and leaves, including edge, texture, and color information, and uses a convolutional neural network (CNN) for feature learning and classification; The image is segmented into different parts using U-Net semantic segmentation technology to generate pixel-level label maps. Mask R-CNN instance segmentation technology is then used to individually identify and segment each branch, ensuring clear boundaries for each branch and identifying branches and leaves that need to be pruned. a decision-making module for calculating a cutting path and a driving path of the trimmer and the driving vehicle (2) from a current position to a target position based on the result of the image processing module; The intelligent control module is used to control the trimmer and the driving vehicle (2) to move to the target position to trim branches and leaves according to the cutting path and driving path information provided by the decision-making module, while avoiding fruits and branches and leaves that do not need to be trimmed during the trimming process.

2. The machine vision-based automatic pruning system for fruit trees according to claim 1, characterized in that: The branches and leaves that need to be pruned in the image processing module include crossing branches, overlapping branches, weak branches, dry branches, diseased and insect-infested branches, overgrown branches, inner branches, competing branches, drooping branches and overcrowded branches.

3. The machine vision-based automatic pruning system for fruit trees according to claim 2, characterized in that: The top of the driving vehicle (2) is fixedly connected to a housing (3), and the inner bottom wall of the housing (3) is rotatably matched with a spherical joint (4); the top of the spherical joint (4) is fixedly connected to a hydraulic rod (5), and one end of the hydraulic rod (5) away from the spherical joint (4) extends from the inside of the housing (3) to the outside of the housing (3); the top of the housing (3) is provided with an opening for the hydraulic rod (5) to move, and a movable rubber ring is fixedly connected to the opening; the trimmer and the camera (1) are both fixedly connected to the output shaft of the hydraulic rod (5); and the inside of the housing (3) is provided with a lateral adjustment component and a longitudinal adjustment component for driving the hydraulic rod (5) to move laterally and longitudinally, respectively.

4. The machine vision-based automatic pruning system for fruit trees according to claim 3, characterized in that: The lateral adjustment component includes a controller and a first drive member, the controller is used to control the operation of the hydraulic rod (5) and the first drive member; the first drive member is fixedly connected to the inner bottom wall of the housing (3), and the output shaft of the first drive member is fixedly connected to a semicircular adjustment plate (7); the adjustment plate (7) is provided with an adjustment groove for the movement of the hydraulic rod (5) along its length direction, and the inner side wall of the adjustment plate (7) is in sliding engagement with the spherical joint (4); and the end of the adjustment plate (7) away from the first drive member is in rotational engagement with the inner side wall of the housing (3).

5. The machine vision-based automatic pruning system for fruit trees according to claim 4, characterized in that: The longitudinal adjustment component includes a second driving member, and the controller is used to control the operation of the second driving member; the output shaft of the second driving member is fixedly connected to a semicircular moving plate (9), and the moving plate (9) is vertically installed with the adjustment plate (7); the inner wall of the moving plate (9) is slidably matched with the adjustment plate (7), and the moving plate (9) is provided with a sliding groove for the hydraulic rod (5) to move along its length direction; the second driving member is fixedly connected to the inner wall of the housing (3), and the end of the moving plate (9) away from the second driving member is rotatably matched with the inner wall of the housing (3).

6. The machine vision-based automatic pruning system for fruit trees according to claim 5, characterized in that: The trimmer comprises scissors (10), an arc-shaped plate (11) and a straight rod (12); the straight rod (12) is fixedly connected to the top end of the output shaft of the hydraulic rod (5); the arc-shaped plate (11) is fixedly connected to the top end of the straight rod (12); the scissors (10) are hinged to the top end of the straight rod (12); and a driving assembly for driving the scissors (10) to rotate is provided on the straight rod (12).

7. The machine vision-based automatic pruning system for fruit trees according to claim 6, characterized in that: The driving assembly includes a third driving member, and the controller is used to control the operation of the third driving member; the third driving member is fixedly connected to the outer wall of the straight rod (12), the output shaft of the third driving member is coaxially fixedly connected to the first gear (14), and the first gear (14) is meshed with a fan-shaped gear (15); the fan-shaped gear (15) is fixedly connected to the scissors (10) on a side away from the first gear (14).

8. The machine vision-based automatic pruning system for fruit trees according to claim 7, characterized in that: A pump assembly for sucking branches and leaves is fixedly connected to the outer wall of the straight rod (12) away from the third driving member, and a controller is used to control the operation of the pump assembly; an input end of the pump assembly is fixedly connected to an air intake port (17), and an inner wall of the air intake port (17) is fixedly connected to a buffer layer.

9. The machine vision-based automatic pruning system for fruit trees according to claim 8, characterized in that: A movable baffle (18) is hingedly connected to the outer wall of the straight rod (12) away from the pump assembly, and a rotating assembly for rotating the movable baffle (18) is provided on the straight rod (12).

10. The machine vision-based automatic pruning system for fruit trees according to claim 9, characterized in that: The rotating assembly includes a fourth driving member, and the controller is used to control the operation of the fourth driving member; the fourth driving member is embedded and installed on the straight rod (12), and the output shaft of the fourth driving member is fixedly connected to the movable baffle (18).

Images