Intelligent operation system for fruit tree girdling

Abstract

The present application relates to the technical field of fruit tree ring peeling, and particularly relates to an intelligent operation system for fruit tree ring peeling, comprising: a state acquisition unit for determining whether to perform image depth analysis according to a pre-ring peeling state; a reference screening unit for determining a preset number of reference task combinations based on dimensional difference representation values; an image analysis unit for determining a ring peeling execution mode based on regional difference degrees when performing image depth analysis; and a ring peeling analysis unit for determining whether to perform fixed ring peeling or stable analysis based on a traveling tendency value. The present application effectively improves the ring peeling effect of intelligent ring peeling equipment.

Description

Technical Field

[0001] This invention relates to the field of fruit tree girdling technology, and in particular to an intelligent operating system for fruit tree girdling. Background Technology

[0002] In smart agriculture systems, the stable and reliable operation of fruit tree girdling robots is the core foundation for ensuring efficient girdling operations and reducing the risk of excessive bark damage or incomplete girdling. However, as the operating environment of girdling robots becomes increasingly complex, their working modes exhibit poor scenario adaptability. Existing working modes often employ a single operation mode, failing to fully consider the actual operating state of the girdling robot and the dynamic impact of environmental disturbances. This leads to problems such as substandard girdling quality, damage to the xylem, or equipment jamming. Therefore, how to achieve adaptive state perception for girdling robots is a problem that urgently needs to be solved by those skilled in the art.

[0003] Chinese Patent Publication No. CN114793667A discloses a device for cutting the lateral branches of seedlings in the median strip of a highway. The device includes a mobile vehicle and a support. A multi-stage electric push rod is installed in the middle of the support. The output shaft of the multi-stage electric push rod is connected to a support box. A dual-head motor is installed on the inner wall of the support box. Both the front and rear output shafts of the dual-head motor are connected to drive wheels. Transmission shafts are rotatably connected to the four corners of the support box. Saw blades are connected to the front and rear ends of each transmission shaft. Transmission wheels are connected to the front drive wheels via a first transmission belt, and transmission wheels are connected to the rear drive wheels via a second transmission belt. However, the above technical solution does not consider the actual operating state of the girdling robot and the dynamic influence of environmental disturbances on the girdling task execution, thus reducing girdling efficiency and consequently reducing the girdling quality of the girdling robot. Summary of the Invention

[0004] Therefore, the present invention provides an intelligent operation system for girdling fruit trees, which overcomes the problem in the prior art that does not take into account the actual working state of the girdling robot and the dynamic influence of environmental disturbances on the execution of the girdling task, thereby reducing girdling efficiency and thus reducing the girdling quality of the girdling robot.

[0005] To achieve the above objectives, the present invention provides an intelligent operation system for girdling fruit trees, comprising: The state acquisition unit is used to determine the pre-ringing state based on the speed fluctuation and gripper variability, and to determine whether to perform image depth analysis based on the pre-ringing state. The reference filtering unit is used to determine the feature dimension based on the feature vectors of the current task combination and each historical task combination, and to determine a preset number of reference task combinations based on the dimension difference representation value of the feature dimension. The image analysis unit is used to determine the regional difference of the target region based on the regional characterization value of each sub-girdling region when performing image depth analysis. Based on the comparison result of the regional difference and the preset regional difference, it determines the girdling execution method as either the girdling execution method based on the travel tendency value or the baseline girdling method. The girdling analysis unit is used to determine the travel tendency value based on the complexity of the girdling cutter trajectory and the complexity of the girdling robot's movement trajectory, and to determine whether to perform fixed girdling or stable analysis based on the comparison result between the travel tendency value and the preset travel tendency value.

[0006] Furthermore, the status acquisition unit includes: The speed fluctuation determination submodule is used to determine the speed fluctuation based on the maximum and minimum moving speeds during the evaluation period. The gripper variability determination submodule is used to determine the gripper variability based on the girdling force measurement value at each analysis time. The pre-ringing state submodule is used to determine the pre-ringing state of the ringing robot based on the speed fluctuation and gripper variability. The pre-ringing state includes a first state and a second state. The image depth analysis submodule is used to determine and perform image depth analysis for the pre-ringing state in the first state.

[0007] Furthermore, the pre-ringing state submodule determines the pre-ringing state as the first state when the speed fluctuation is greater than the preset speed fluctuation or the gripper variability is greater than the preset gripper variability. The pre-ringing state submodule determines the pre-ringing state as the second state when the speed fluctuation is less than or equal to the preset speed fluctuation and the gripper variability is less than or equal to the preset gripper variability.

[0008] Furthermore, the reference filtering unit includes: The combination construction submodule is used to combine tasks based on the current cyclopetting task and neighboring cyclopetting tasks to determine the current task combination and the historical task combination; The feature vector construction submodule is used to construct feature vectors that include spatial location features and process parameter features; The feature similarity determination submodule is used to determine a preset number of reference task combinations based on the dimensional difference representation values ​​of the current task combination and each historical task combination in each feature dimension.

[0009] Furthermore, the reference filtering unit determines a preset number based on the number of historical task combinations in the historical database, and the reference task combinations simultaneously include records of both fixed girdling methods and mobile girdling methods.

[0010] Furthermore, the image analysis unit includes: The regional dissimilarity determination submodule is used to determine the regional dissimilarity based on the regional characterization value of each sub-girdling region; The regional characterization value determination submodule is used to determine the regional characterization value based on branch density and branch diameter; The image acquisition and determination submodule is used to determine the girdling method based on regional differences.

[0011] Furthermore, the image acquisition and determination submodule is used to determine the girdling execution method based on the travel tendency value when the regional difference is greater than the preset regional difference. When the regional difference is less than or equal to the preset regional difference, the benchmark girdling method is adopted.

[0012] Furthermore, the girdling analysis unit includes: The travel tendency value determination submodule is used to determine the travel tendency value based on the complexity of the girder's trajectory and the complexity of the girder robot's movement trajectory. The ring-stripping execution mode analysis submodule is used to determine whether to perform fixed ring-stripping for the travel state in the first tendency state, and to determine whether to perform stability analysis for the travel state in the second tendency state.

[0013] Furthermore, the girdling execution mode analysis submodule determines the movement state where the movement tendency value is greater than the preset movement tendency value as the first tendency state; The girdling execution mode analysis submodule determines the movement state where the movement tendency value is less than the preset movement tendency value as the second tendency state.

[0014] Furthermore, when the girdling execution mode analysis submodule performs stability analysis, it determines to perform fixed girdling when the pre-girdling state of the reference data is in the first state; When the pre-ringing state of historical data is in the second state, it is determined to perform moving ringing.

[0015] Compared with existing technologies, the beneficial effects of this invention are that it uses speed fluctuation to reflect the degree of jitter of the intelligent girdling equipment during the task, and gripper variability to reflect the smoothness of the robotic arm's trajectory during movement. When the degree of jitter is large or the trajectory smoothness of the gripper during movement is poor, it means that there is a large uncertainty in the current girdling action; that is, high-speed movement may lead to positioning deviation, and instability may lead to image blurring or target displacement. At this time, the system determines that image depth analysis needs to be performed to compensate for the information loss caused by motion uncertainty, avoiding the indiscriminate performance of time-consuming depth image analysis on the acquired images, which would lead to a waste of computing resources, thereby improving the girdling effect of the intelligent girdling equipment.

[0016] In this invention, when performing image depth analysis, the region characterization value is determined based on branch density and branch diameter, and the region difference degree is determined based on the region characterization value of each sub-girdling region. In actual girdling scenarios, there is significant spatial heterogeneity in branch distribution density and branch thickness among different sub-girdling regions of the same fruit tree. In dense areas, the girdling knife is prone to interference with adjacent branches, while in thicker branches, a larger girdling force is required. The combination of these two factors results in significant differences in the difficulty of operation in each region. Therefore, in performing image depth analysis, this invention determines the region characterization value based on both branch density and branch diameter. Preferably, after normalization, the region characterization value is weighted to represent the difficulty of operation. Furthermore, the region difference degree is calculated based on the region characterization value of each sub-girdling region to determine whether to enable the directional decision. This invention effectively realizes the quantitative perception of the spatial heterogeneity of the girdling region, avoiding the difficulty of comprehensively quantifying the actual impact of branch structure on girdling execution by relying solely on a single visual feature or fixed rules to divide the operation area in the prior art. This results in high-difficulty regions being assigned the same execution parameters as low-difficulty regions, thereby improving the girdling effect.

[0017] In this invention, when the regional difference exceeds a preset regional difference, the girdling execution method is determined based on a travel tendency value. This travel tendency value is determined based on the complexity of the girdling blade trajectory and the complexity of the girdling robot's movement trajectory. The complexity of the girdling blade trajectory represents the cost of the robotic arm's movement, while the complexity of the robot's movement trajectory represents the cost of movement. The travel tendency value is compared with a preset threshold: when in a first tendency state, fixed girdling is executed, as the robotic arm path is more complex than the chassis path, and fixed girdling avoids redundant chassis movement; when in a second tendency state, reference backtracking analysis is performed to further evaluate whether mobile girdling is suitable, avoiding blindly activating the chassis when the chassis movement cost is higher than the robotic arm's detour cost, thereby improving the girdling effect. Attached Figure Description

[0018] Figure 1 This is a unit connection diagram of an intelligent operation system for girdling fruit trees according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating how the pre-ringing state is determined based on speed fluctuation and gripper variability, according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating the processing strategy based on the comparison result between regional difference degree and preset regional difference degree in an embodiment of the present invention. Figure 4 This is a flowchart illustrating how the travel state is determined based on a comparison between a travel tendency value and a preset travel tendency value, according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0020] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0021] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0022] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0023] Please see Figure 1 The diagram shown is a unit connection diagram of an intelligent operation system for girdling fruit trees according to an embodiment of the present invention. The present invention provides an intelligent operation system for girdling fruit trees, the system comprising: The state acquisition unit is used to determine the pre-ringing state based on the speed fluctuation and gripper variability, and to determine whether to perform image depth analysis based on the pre-ringing state. The reference filtering unit, which is connected to the status acquisition unit, is used to determine the feature dimension based on the feature vectors of the current task combination and each historical task combination, and to determine a preset number of reference task combinations based on the dimension difference representation value of the feature dimension. The image analysis unit is connected to the state acquisition unit and the reference filtering unit respectively. When performing image depth analysis, it determines the regional difference degree of the target region based on the regional characterization value of each sub-girdling region, and determines the girdling execution method based on the comparison result of the regional difference degree and the preset regional difference degree, whether it is a girdling execution method based on the travel tendency value or a baseline girdling method. The girdling analysis unit is connected to the state acquisition unit, the reference screening unit, and the image analysis unit, respectively. It is used to determine the travel tendency value based on the complexity of the girdling cutter trajectory and the complexity of the girdling robot's movement trajectory, and to determine whether to perform fixed girdling or stable analysis based on the comparison result between the travel tendency value and the preset travel tendency value.

[0024] In this embodiment, the girdling robot includes a mobile chassis and a girdling actuator. The girdling actuator includes a girdling blade and a multi-degree-of-freedom robotic arm, with a depth camera mounted at the end of the robotic arm. The system uses a 5-minute evaluation cycle, collecting data from several analysis points within each cycle and storing historical data in a local or cloud database. Each historical data entry includes: a girdling image sequence, the girdling robot's operating speed record, gripping force record, joint trajectory, girdling blade trajectory, the corresponding pre-girdling state, and the final girdling execution method. The target area is a two-dimensional image of the tree to be girdled, and there are multiple girdling points within the target area.

[0025] Several sampling points are set within a single evaluation cycle. Specifically, the preset sampling cycle is 5 seconds, meaning that a sampling point is set every 5 seconds, and this sampling point is recorded as the analysis time.

[0026] Specifically, the status acquisition unit includes: The speed fluctuation determination submodule is used to determine the speed fluctuation based on the maximum and minimum moving speeds during the evaluation period. The gripper variability determination submodule is used to determine the gripper variability based on the girdling force measurement value at each analysis time. The pre-ringing state submodule is used to determine the pre-ringing state of the ringing robot based on the speed fluctuation and gripper variability. The pre-ringing state includes a first state and a second state. The image depth analysis submodule is used to determine whether to perform image depth analysis for the pre-ringing state in the first state.

[0027] In this embodiment of the invention, speed fluctuation is used to determine the degree of fluctuation in the moving speed of the girdling robot. A larger speed fluctuation indicates that the girdling robot frequently accelerates and decelerates during its movement, which causes positional jitter at the end of the robotic arm, i.e., poor continuity of the girdling action. Preferably, the speed fluctuation is determined based on the maximum and minimum moving speeds of the most recent evaluation period. Specifically, the speed fluctuation is the maximum and minimum moving speeds of the girdling robot at each detection time within the most recent evaluation period, with the maximum moving speed recorded as the maximum moving speed and the minimum moving speed recorded as the minimum moving speed. The speed fluctuation is the absolute value of the difference between the maximum and minimum moving speeds. The moving speed is obtained by a speed sensor installed on the chassis of the girdling robot.

[0028] The evaluation period is the time period from the moment the ring-stripping robot completes the work at the previous ring-stripping point and retracts the ring cutter to the moment it moves to the predetermined work position of the current ring-stripping point and stops moving. The preceding girdling point is the girdling point that is closest to the current girdling point in the task execution order and whose position of the girdling robot has changed. It is easy to understand that if there is no preceding girdling point, the speed fluctuation of the current girdling point is 0.

[0029] The gripper variability is used to determine the degree of fluctuation in the girdling force output by the girdling robot during girdling operations. The greater the gripper variability, the more the robotic arm repeatedly adjusts the clamping force during the girdling process, resulting in unstable contact pressure between the girdling blade and the branch surface. This can easily lead to inconsistent girdling cut widths or phloem residue, resulting in poor girdling effects.

[0030] Preferably, the gripper variability is determined based on the girdling force measurement values ​​at each analysis time within the most recent work cycle. Specifically, the gripper variability is determined based on the ratio of the standard deviation of the girdling force measurement values ​​at each analysis time within the most recent work cycle to the average girdling force. The girdling force is measured by a pressure sensor installed inside the robotic arm of the girdling robot. The work cycle is the time period from when the robotic arm contacts and clamps the branch at the girdling point until the girdling blade completes the girdling operation at that point and detaches from the branch surface.

[0031] In this embodiment of the invention, the sampling frequency of the pressure sensor is 200Hz, the sampling frequency of the speed sensor is 100Hz, and the frame rate of the depth camera is 30 frames / second. The sampling frequencies of each sensor in this embodiment are preferred values. In practical applications, they can be adaptively adjusted according to the working speed of the girdling robot, the dynamic response characteristics of the robotic arm, and the computing power of the control system to ensure that the data collected by each sensor can fully reflect the dynamic changes in the girdling operation process, providing a reliable data foundation for subsequent pre-girdling state determination, image depth analysis, and girdling execution mode decision-making.

[0032] Please see Figure 2 As shown, it is a flowchart of the process for determining the pre-ringing state based on speed fluctuation and gripper variability in an embodiment of the present invention.

[0033] Specifically, the pre-ringing state submodule determines a pre-ringing state where the speed fluctuation is greater than a preset speed fluctuation or the gripper variability is greater than a preset gripper variability as the first state. The pre-ringing state submodule determines the pre-ringing state as the second state when the speed fluctuation is less than or equal to the preset speed fluctuation and the gripper variability is less than or equal to the preset gripper variability.

[0034] In this embodiment of the invention, the continuity of the girdling action of the girdling robot is determined by the speed fluctuation, and the fluctuation of the output girdling force of the girdling robot is determined by the gripper variability. If the speed fluctuation or gripper variability is large, it indicates that the girdling process of the girdling robot is jittery or unstable, which is prone to causing defects in the quality of the girdling opening. Therefore, it is determined to be in the first state. In this state, image depth analysis is immediately triggered to further determine the cause of the jitter abnormality, thereby improving the rationality of the girdling method. Therefore, the higher the user's requirements for the girdling effect of the girdling equipment, the smaller the preset speed fluctuation value and the preset gripper variability value should be. Specifically, girdling data of the girdling robot containing different speed fluctuation degrees and gripper variability are acquired. The speed fluctuation range is 0-0.5, and the gripper variability range is 0.03-0.35. The accuracy of triggering image depth analysis for each parameter combination is evaluated by random sampling, and an accuracy contour map is plotted. The preset threshold combination that maximizes the accuracy is selected. Simulation results show that image depth analysis is triggered when the preset speed fluctuation is 0.08 and the preset gripper variability is 0.18. Therefore, in this embodiment, the preset speed fluctuation is set to 0.08 m / s, and the preset gripper variability is set to 0.18. It is worth noting that the preset values ​​can be adaptively adjusted according to the girdling requirements of the fruit tree variety and the mechanical precision of the girdling robot.

[0035] Specifically, the reference filtering unit includes: The combination construction submodule is used to combine tasks based on the current cyclopetting task and neighboring cyclopetting tasks to determine the current task combination and the historical task combination; The feature vector construction submodule is used to construct feature vectors that include spatial location features and process parameter features; The feature similarity determination submodule is used to determine a preset number of reference task combinations based on the dimensional difference representation values ​​of the current task combination and each historical task combination in each feature dimension.

[0036] In this embodiment of the invention, a historical database is provided, which includes historical data of the historical girdling process. The historical data includes a complete record of the girdling task. The girdling task is a complete girdling operation process. The girdling task corresponding to each girdling point is recorded as a set of working condition characteristic data when it is executed. Among them, the neighborhood girdling task is used to determine the next girdling task that needs to be executed after the current girdling task in the movement sequence; The current task combination is determined based on the ordered pairs formed by the current girdling task and its neighboring girdling tasks in the order of movement. The combination method is to arrange the feature vector of the current girdling task and the feature vector of the neighboring tasks to form a composite vector containing twice the number of features.

[0037] The feature vectors of the current task combination and the historical task combination are obtained respectively. The feature vectors include, but are not limited to, spatial location features and process parameter features. The spatial location features are the three-dimensional coordinates of the starting point and the ending point of the task within the task combination. The process parameter features include the operating parameters of the girdling robot, including but not limited to girdling depth, feed speed and movement speed.

[0038] Get the distance between the current task combination and all historical task combinations, sort the historical task combinations in ascending order of distance, and select the first preset number of historical combinations as reference task combinations. The distance is determined based on the sum of squared dimensional difference representation values. The feature dimensions of the current task combination and each historical task combination are normalized. For a single feature dimension, the difference between the feature dimension in the current task combination and the feature dimension in a single historical task combination is recorded as the dimensional difference representation value. Each feature dimension corresponds to a dimensional difference representation value. The dimensional difference representation value of a single feature dimension is used to determine the similarity between the current task combination and the historical task combination on that feature dimension. It is easy to understand that the larger the dimensional difference representation value, the smaller the similarity between the current task combination and the historical task combination on that feature dimension.

[0039] Specifically, the reference filtering unit determines a preset number based on the number of historical task combinations in the historical database, and the reference task combinations simultaneously include records of both fixed girdling methods and mobile girdling methods.

[0040] The preset quantity is determined based on the number of historical task combinations. It is easy to understand that the more historical task combinations there are, the larger the value of the preset quantity will be. In this embodiment of the invention, a method for determining the preset quantity of reference task combinations is provided. Preferably, historical working conditions in which the girdling effect meets the usage requirements are obtained, and the average value of the reference task combinations corresponding to the historical working conditions is 5, which is recorded as the preset quantity. It is worth noting that the girdling methods in the obtained reference task combinations cover both fixed girdling and mobile girdling. The historical operating conditions under which the girdling effect meets the usage requirements are the operating periods that have been manually verified and confirmed to be consistent with the system's judgment results.

[0041] Specifically, the image analysis unit includes: The regional dissimilarity determination submodule is used to determine the regional dissimilarity based on the regional characterization value of each sub-girdling region; The regional characterization value determination submodule is used to determine the regional characterization value based on branch density and branch diameter; The image acquisition and determination submodule is used to determine the girdling method based on regional differences.

[0042] The regional variability degree is used to determine the degree of balance between the regional characterization values ​​of each sub-girdling area. The greater the regional variability degree, the more significant the difference between each sub-girdling area. Specifically, the target girdling area is evenly divided into several sub-girdling areas. The regional variability degree is determined based on the regional characterization value of each sub-girdling area. The girdling method is determined by comparing the regional variability degree with the preset regional variability degree to improve the regional adaptability of the girdling operation. Therefore, the higher the user's requirements for the environmental adaptability of the girdling robot in regional operation, the smaller the preset threshold value corresponding to the regional variability degree. Preferably, historical data on the girdling effect meeting the usage requirements are obtained, and the average value of the regional variability degree corresponding to the historical data is 1.5, which is recorded as the preset regional variability degree.

[0043] The regional variability of a single sub-girdled region is determined based on the ratio of kurtosis difference to trough difference; the kurtosis difference is determined based on the maximum regional characterization value and the average regional characterization value, and the trough difference is determined based on the minimum regional characterization value and the average regional characterization value. In this embodiment of the invention, the kurtosis difference is the absolute value of the difference between the maximum regional characterization value and the average regional characterization value, and the trough difference is the absolute value of the difference between the minimum regional characterization value and the average regional characterization value. The region characterization value is used to determine the girdling complexity of the sub-girdling region. The larger the region characterization value, the higher the girdling complexity in the sub-girdling region, indicating that the girdling robot needs to deal with more complex branch structures when operating in the region, that is, the girdling complexity is higher. Therefore, the girdling execution method is determined based on the travel tendency value to improve the rationality of the girdling execution method. The smaller the region characterization value, the easier the girdling operation is, and the baseline girdling method is adopted.

[0044] Preferably, the region characterization value is determined based on branch density and branch diameter. Specifically, the region characterization value is determined after normalization of branch density and branch diameter. In this invention, branch density is used to determine the structural complexity of the sub-girdling area. Higher branch density makes it easier for the girdling cutter to be interfered with by adjacent branches during the cutting process, increasing the difficulty of girdling. Branch diameter is used to determine the girdling resistance; a larger branch diameter requires a greater girdling force. Both branch density and branch diameter jointly characterize the operational difficulty of the sub-girdling area. In this embodiment, branch density is determined based on the ratio of the branch area to the area of ​​the sub-girdling area. The branch area is obtained through LiDAR point cloud analysis, and the branch diameter is the average diameter of the branches where the girdling point is located within the sub-girdling area. The diameter of a single branch is determined through visual measurement. Obtaining the branch area through LiDAR point cloud analysis and the branch diameter through visual measurement are readily understood by those skilled in the art and will not be elaborated upon here.

[0045] Specifically, the image acquisition and determination submodule is used to determine the girdling execution method based on the travel tendency value when the regional difference is greater than the preset regional difference. When the regional difference is less than or equal to the preset regional difference, the benchmark girdling method is adopted.

[0046] In this embodiment of the invention, the regional difference degree is used to characterize the balance between the regional characterization values ​​of each sub-girdling region. When the regional difference degree is greater than its corresponding preset threshold, it indicates that the balance of regional characterization values ​​of different sub-girdling regions within the target area is poor, that is, the girdling difficulty of each sub-girdling region varies greatly. Therefore, if the same girdling execution method is used for different sub-girdling regions, the girdling execution method is difficult to effectively adapt to the actual scenario, thereby reducing the girdling effect. Therefore, in the case of large differences in the girdling difficulty of each sub-girdling region, it is determined that the girdling execution method is determined based on the travel tendency value, thereby improving the rationality of the girdling execution method and further improving the girdling effect. When the regional difference degree is less than or equal to the preset regional difference degree, it indicates that the girdling difficulty of each sub-girdling region is relatively similar. Therefore, it is determined that the baseline girdling method is used for each sub-girdling region, and there is no need to differentiate the girdling execution method for each sub-girdling region, thereby saving system energy consumption. In this embodiment of the invention, the baseline girdling method is mobile girdling, that is, the girdling task of all girdling points is completed by moving the base corresponding to the girdling robot. The movement order is determined based on the effective distance. The girdling point closest to the ground and closest to the trunk of the tree in the target area is taken as the starting point. Starting from the starting point, the girdling point that is closest to the current girdling point and has not yet been worked on is selected as the next work target until all girdling points have been girdled. The effective distance is the minimum distance between the current girdling point and other girdling points that have not yet been worked on.

[0047] Specifically, the girdling analysis unit includes: The travel tendency value determination submodule is used to determine the travel tendency value based on the complexity of the girder's trajectory and the complexity of the girder robot's movement trajectory. The ring-stripping execution mode analysis submodule is used to determine whether to perform fixed ring-stripping for the travel state in the first tendency state, and to determine whether to perform reference backtracking analysis for the travel state in the second tendency state.

[0048] The travel tendency value is used to determine the behavioral tendency of the girdling robot when performing neighborhood girdling tasks. Specifically, the travel tendency value is determined based on the ratio of the complexity of the girdling blade trajectory to the complexity of the device movement trajectory. Preferably, historical working conditions that meet the requirements for girdling effect are obtained, including cases of successful fixed girdling and cases of switching to mobile mode due to failure of fixed girdling. For each historical working condition, its travel tendency value is calculated, and the decision accuracy rate is statistically analyzed when different travel tendency values ​​are used as decision thresholds. When the preset travel tendency value is 1, the decision accuracy rate reaches its maximum value. If the threshold is reduced to 0.9, there will be many false alarms that misjudge working conditions that should use fixed girdling as requiring backtracking analysis, resulting in unnecessary backtracking calculations. If the preset travel tendency value is increased to 1.1, there will be missed reports that misjudge working conditions that should use backtracking analysis as suitable for fixed girdling, resulting in interference or failure in actual operation. Therefore, in this embodiment of the invention, the preset travel tendency value is set to 1.

[0049] Among them, the complexity of the girdling blade trajectory is used to characterize the task execution of the girdling robot at the neighboring girdling points when the girdling equipment performs girdling on the neighboring girdling points without moving, that is, only by moving the robotic arm; specifically, the complexity of the girdling blade trajectory is determined based on the average curvature of the girdling blade trajectory; the complexity of the equipment movement trajectory is used to determine the task execution of the neighboring girdling points by the movement of the girdling robot when performing girdling on the neighboring girdling points; specifically, the complexity of the equipment movement trajectory is determined based on the average curvature of the robot movement trajectory.

[0050] In this embodiment of the invention, the target trajectory includes the girder trajectory and the equipment movement trajectory. By setting a reference number of points on the target trajectory and recording each point as a trajectory discrete point, for a single trajectory discrete point, a sub-trajectory segment with the trajectory discrete point as the center and a fixed length is extracted on the target trajectory, and the curvature of the trajectory discrete point is obtained by parametric fitting. Curvature is a measure of the degree of bending of a curve near a certain point. An excessively long fixed length introduces the influence of distant points, causing the curvature to reflect a broader bending trend rather than local features, thus losing its ability to distinguish local complexity. Conversely, a too-small fixed length includes only a few points, making the fitted curve susceptible to measurement noise or discretization errors, resulting in drastic fluctuations in the curvature value. Preferably, the fixed length is 15 mm for the girder trajectory and 30 mm for the device movement trajectory. These values ​​are obtained through statistical optimization of numerous historical working conditions. Historical working conditions meeting the requirements for girdling effect are collected, and curvature is calculated using different fixed length values. The correlation between the calculated curvature and the actual curvature of the trajectory is evaluated. Experimental results show that for the girder trajectory, the mean square error of curvature estimation reaches its minimum when the fixed length is 15 mm; for the girdling robot movement trajectory, the mean square error of curvature estimation reaches its minimum when the fixed length is 30 mm.

[0051] The value of the baseline quantity is easy to understand. The larger the value of the baseline quantity, the higher the accuracy of the trajectory complexity corresponding to the target trajectory. Preferably, historical working conditions that meet the requirements of girdling effect are obtained, and the average number of trajectory discrete points corresponding to the historical working conditions is 10, which is denoted as the baseline quantity.

[0052] Specifically, the ring-stripping execution mode analysis submodule determines the travel state where the travel tendency value is greater than the preset travel tendency value as the first tendency state; The girdling execution mode analysis submodule determines the movement state where the movement tendency value is less than the preset movement tendency value as the second tendency state.

[0053] The travel tendency value characterizes the tendency of the girdling robot to prioritize moving girdling or fixed girdling for neighboring girdling points after completing the current girdling point. When the travel tendency value is greater than its corresponding preset threshold, it indicates that the complexity of the girdling blade trajectory is higher than the complexity of the girdling robot's movement trajectory. That is, reaching the neighboring girdling point by moving the robotic arm alone requires a longer detour, while the robot chassis movement path is relatively direct. In this case, fixed girdling is determined to be performed. Conversely, when the travel tendency value is less than its corresponding preset threshold, it indicates that the complexity of the girdling blade trajectory is lower than the complexity of the girdling robot's movement trajectory. In this case, reference backtracking analysis is performed to further determine whether moving girdling can be performed, so as to avoid positioning errors and vibrations introduced by frequent chassis movements, thereby improving the girdling effect.

[0054] When the travel tendency value is equal to the preset travel tendency value, it means that the complexity of the girdling blade trajectory and the complexity of the girdling robot's movement trajectory are equal. At this time, either the fixed girdling or the moving girdling can be selected to perform the girdling of the neighboring girdling points.

[0055] Specifically, when the ring-stripping execution mode analysis submodule performs reference backtracking analysis, it determines to perform fixed ring-stripping in the first state of the pre-ring-stripping state of historical data. If the pre-ringing state of the historical data is in the second state, it is determined to perform moving ringing.

[0056] When performing reference backtracking analysis, the pre-ringing state of the reference task combination is used to determine the actual dynamic stability of the reference task combination. If the stability of the reference task combination is strong, it is determined to perform moving ringing; if the stability of the reference task combination is poor, it is determined to perform fixed ringing.

[0057] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An intelligent operating system for girdling fruit trees, characterized in that, include: The state acquisition unit is used to determine the pre-ringing state based on the speed fluctuation and gripper variability, and to determine whether to perform image depth analysis based on the pre-ringing state. The reference filtering unit is used to determine the feature dimension based on the feature vectors of the current task combination and each historical task combination, and to determine a preset number of reference task combinations based on the dimension difference representation value of the feature dimension. The image analysis unit is used to determine the regional difference of the target region based on the regional characterization value of each sub-girdling region when performing image depth analysis. Based on the comparison result of the regional difference and the preset regional difference, it determines the girdling execution method as either the girdling execution method based on the travel tendency value or the baseline girdling method. The girdling analysis unit is used to determine the travel tendency value based on the complexity of the girdling cutter trajectory and the complexity of the girdling robot's movement trajectory, and to determine whether to perform fixed girdling or stable analysis based on the comparison result between the travel tendency value and the preset travel tendency value.

2. The intelligent operation system for girdling fruit trees according to claim 1, characterized in that, The status acquisition unit includes: The speed fluctuation determination submodule is used to determine the speed fluctuation based on the maximum and minimum moving speeds during the evaluation period. The gripper variability determination submodule is used to determine the gripper variability based on the girdling force measurement value at each analysis time. The pre-ringing state submodule is used to determine the pre-ringing state of the ringing robot based on the speed fluctuation and gripper variability. The pre-ringing state includes a first state and a second state. The image depth analysis submodule is used to determine and perform image depth analysis for the pre-ringing state in the first state.

3. The intelligent operation system for girdling fruit trees according to claim 2, characterized in that, The pre-ringing state submodule determines the pre-ringing state as the first state when the speed fluctuation is greater than the preset speed fluctuation or the gripper variability is greater than the preset gripper variability. The pre-ringing state submodule determines the pre-ringing state as the second state when the speed fluctuation is less than or equal to the preset speed fluctuation and the gripper variability is less than or equal to the preset gripper variability.

4. The intelligent operation system for girdling fruit trees according to claim 3, characterized in that, The reference filtering unit includes: The combination construction submodule is used to combine tasks based on the current cyclopetting task and neighboring cyclopetting tasks to determine the current task combination and the historical task combination; The feature vector construction submodule is used to construct feature vectors that include spatial location features and process parameter features; The feature similarity determination submodule is used to determine a preset number of reference task combinations based on the dimensional difference representation values ​​of the current task combination and each historical task combination in each feature dimension.

5. The intelligent operation system for girdling fruit trees according to claim 4, characterized in that, The reference filtering unit determines a preset number based on the number of historical task combinations in the historical database, and the reference task combinations simultaneously include records of both fixed girdling methods and mobile girdling methods.

6. The intelligent operation system for girdling fruit trees according to claim 5, characterized in that, The image analysis unit includes: The regional dissimilarity determination submodule is used to determine the regional dissimilarity based on the regional characterization value of each sub-girdling region; The regional characterization value determination submodule is used to determine the regional characterization value based on the branch density and branch diameter; The image acquisition and determination submodule is used to determine the girdling execution method based on regional differences.

7. The intelligent operation system for girdling fruit trees according to claim 6, characterized in that, The image acquisition and determination submodule is used to determine the girdling execution method based on the travel tendency value when the regional difference is greater than the preset regional difference. When the regional difference is less than or equal to the preset regional difference, the benchmark girdling method is adopted.

8. The intelligent operation system for girdling fruit trees according to claim 7, characterized in that, The girdling analysis unit includes: The travel tendency value determination submodule is used to determine the travel tendency value based on the complexity of the girder's trajectory and the complexity of the girder robot's movement trajectory. The ring-stripping execution mode analysis submodule is used to determine whether to perform fixed ring-stripping for the travel state in the first tendency state, and to determine whether to perform stability analysis for the travel state in the second tendency state.

9. The intelligent operation system for girdling fruit trees according to claim 8, characterized in that, The girdling execution mode analysis submodule determines the first tendency state as the movement state where the movement tendency value is greater than the preset movement tendency value; The girdling execution mode analysis submodule determines the movement state where the movement tendency value is less than the preset movement tendency value as the second tendency state.

10. The intelligent operation system for girdling fruit trees according to claim 8, characterized in that, When the ring-stripping execution mode analysis submodule performs stability analysis, it determines to perform fixed ring-stripping when the pre-ring-stripping state of the reference data is in the first state. When the pre-ringing state of historical data is in the second state, it is determined to perform moving ringing.

Images