Self-adaptive fruit picking method and system

Abstract

The invention belongs to the technical field of agricultural picking, and particularly relates to a fruit self-adaptive picking method and system.The method comprises the following steps that a bionic cobweb grabbing mechanism formed by interweaving flexible thin lines makes contact with the surface of a fruit, and intersection point grabbing force at each intersection point is obtained; acquiring touch information between the fruit and the latticed grabbing surface in real time, wherein the touch information comprises spatial force distribution, a contact area and a micro-slip index; on the basis of the intersection point grabbing force, the space force distribution and the contact area, the real-time total grabbing force and the wrapping force of the latticed grabbing face to the fruit are obtained; the real-time total grabbing force, the wrapping force and the micro-slip index are input into a preset maturity judgment module, and the maturity grade of the fruits is obtained in real time; according to the maturity grade and preset target grabbing force, the grabbing force is dynamically adjusted; and executing a picking action. By accurately sensing touch information and maturity, lossless picking of fruits can be achieved, and fruit damage caused by excessive clamping or insufficient force is avoided.

Description

Technical Field

[0001] This invention belongs to the field of agricultural harvesting technology, specifically relating to an adaptive fruit harvesting method and system. Background Technology

[0002] Fruit harvesting is a simple yet crucial link in the agricultural production chain. It requires the rapid collection of fruits in the field while ensuring that the fruit's appearance and internal quality are not damaged during the harvesting process. Traditional manual harvesting relies on the operator's tactile experience to judge the grip strength and timing of picking. This method has an inherent advantage in flexibility, but it encounters bottlenecks in terms of manpower shortages and large-scale, standardized operations. Therefore, the introduction of robots and automated harvesting equipment to improve efficiency and reduce labor intensity has become an inevitable trend.

[0003] Currently, the main types of automated harvesting terminals are rigid or semi-rigid grippers, negative pressure suction cups, and flexible soft grippers. Rigid or semi-rigid grippers are relatively reliable in positioning and load-bearing, but the rigidity of the contact surface leads to high local contact pressure, easily causing indentations or internal tissue damage to soft fruits such as strawberries, tomatoes, and kiwis. Negative pressure suction cups are sensitive to surface smoothness and shape; their adsorption effect decreases and they risk detachment when encountering slippery surfaces or uneven fruit peels. Soft / flexible grippers disperse contact pressure through material or structural flexibility. Although they have made progress in reducing damage compared to rigid or semi-rigid grippers and negative pressure suction cups, simple passive flexibility makes it difficult to precisely adjust the gripping force according to the different ripeness levels of the fruit. Fruit damage caused by differences in gripping force remains a major drawback in the current harvesting process. For example, patent number CN110432000A provides a fruit and vegetable harvesting actuator and harvesting method based on flexible gripping and clamping / shearing integration, proposing a structural improvement that combines flexible gripping and clamping / shearing functions to reduce harvesting damage while maintaining efficiency. Another example is patent number CN108738702B, which provides a peeling apple harvester, focusing on the design of the end mechanism of the harvesting actuator, including fruit stem positioning and elastic buffering mechanisms. While these two solutions alleviate harvesting damage from different angles, most remain at the structural or static detection level, making it difficult to adaptively adjust the gripping strategy according to changes in contact force to adapt to harvesting fruits of different ripeness levels, resulting in a persistently high damage rate during harvesting. Summary of the Invention

[0004] In order to solve the problems existing in the prior art, the purpose of this invention is to provide an adaptive fruit picking method that can accurately sense tactile information and maturity to adaptively adjust the grasping force, thereby achieving fruit picking without damage and avoiding fruit damage caused by excessive gripping or insufficient force.

[0005] The technical solution of this invention is: An adaptive fruit harvesting system, comprising: A biomimetic spider web grasping mechanism includes an actuator module and two opposing grid-like grasping surfaces. The grid-like grasping surfaces are mesh structures woven from multiple flexible fine threads. At each intersection of the fine threads within the grid-like grasping surfaces, there are contact blocks made of magnetoelastic material. These contact blocks are used to generate magnetic field changes when subjected to force and deformation. The actuator module includes a driver and two picking arms. The driver has two moving ends, each of which is fixed to a corresponding grid-like grasping surface via one of the picking arms. The driver is used to drive the two picking arms to move towards each other through the two moving ends, so that the two grid-like grasping surfaces can grasp and pick the fruit. The acquisition module includes sensors corresponding to multiple contact blocks, and the multiple sensors are respectively fixed on the corresponding contact blocks. The sensors are used to monitor the magnetic field changes of the contact blocks in real time when deformation occurs. The first processor is electrically connected to the multiple sensors and is used to obtain the gripping force at the intersection of each contact block based on the real-time magnetic field changes fed back by each sensor, and to obtain the effective contact area and micro-slip index between the grid-like gripping surface and the fruit through the real-time changes of the gripping force at each intersection. The second processor, electrically connected to the first processor, is used to obtain the real-time total grasping force and wrapping force of the grid-shaped grasping surface on the fruit through the effective contact area and the grasping force at each intersection. The maturity determination module is electrically connected to the first processor and the second processor respectively, and is used to classify the maturity level of the fruit by means of the real-time total grasping force, wrapping force and micro-slip index. The third processor is electrically connected to the second processor, the maturity discrimination module, and the driver, respectively. It is used to match the maturity level with the preset target grasping force, and dynamically adjust the real-time total grasping force of the two grid-shaped grasping faces on the fruit based on the matching result through the driver. At the same time, it acquires the real-time execution grasping force until the real-time execution grasping force reaches the preset target grasping force, so as to complete the non-destructive harvesting of the fruit.

[0006] Preferably, the contact block of the magnetoelastic material comprises a ferromagnetic alloy, magnetic rubber, or magnetic polymer.

[0007] Preferably, the gripping force at the intersection corresponding to the contact block is determined according to the following formula: , in, Intersection of thin lines P i The gripping force at the intersection of the contact blocks; Intersection Pi The stiffness coefficient of the contact block; Intersection of thin lines P i The contact block shape at that location.

[0008] Preferably, the real-time total grasping force of the grid-like grasping surface on the fruit is determined according to the following formula: , in, The total real-time grasping force for the fruit when the grid-like grasping surface is engaged. V The volume of the grid-like gripping surface; For the grid-shaped gripping surface in x , y , z Spatial force distribution in direction , Intersection of thin lines P i The component of the gripping force at the intersection point in the x-direction; Intersection of thin lines P i The component of the grasping force at the intersection point in the y-direction; Intersection of thin lines P i The component of the gripping force at the intersection point in the z-direction; i =1, 2, 3...

[0009] Preferably, the real-time wrapping force of the grid-like grasping surface on the fruit is determined according to the following formula: , in, For real-time wrapping force of the grid-like grasping surface facing the fruit; S The contact area between the fruit and the grid-like grasping surface. , The effective number of contact blocks within the grid-like grasping surface of the fruit. The surface area of ​​each grid within the grid-like grasping surface; For the grid-shaped gripping surface in x , y , z Spatial force distribution in direction , Intersection of thin lines P i The component of the gripping force at the intersection point in the x-direction; Intersection of thin lines P i The component of the grasping force at the intersection point in the y-direction; Intersection of thin lines P iThe component of the gripping force at the intersection point in the z-direction; i =1, 2, 3...

[0010] Preferably, the preset maturity determination module is determined according to the following formula: , In the formula: Rate the ripeness of the fruit; For real-time total crawling power; For real-time wrapping force; For microslip index; A Effective contact area; The benchmark gripping force is the standard hardness of similar fruits obtained during the trial harvesting phase. The benchmark wrapping force is the standard hardness of similar fruits obtained during the trial harvesting phase. This represents the baseline contact area under the standard hardness of similar fruits obtained during the trial harvesting phase. , , and All are weighted coefficients.

[0011] Preferably, the third processor further includes classifying fruit maturity levels based on the fruit's maturity score and trial harvesting threshold. The fruit maturity levels include Class I (unripe), Class II (suitable for harvesting), and Class III (overripe), determined according to the following formula: Class I - Immature: M > M 1; Class II - Suitable for use: M 2≤ M ≤ M 1; Class III - Overripe: M < M 2; In the formula, M To test the firmness of the fruit during trial harvesting; M 1 represents the upper limit of the fruit firmness threshold during trial harvesting. M 2 represents the lower limit of the threshold for fruit hardness during trial harvesting.

[0012] Preferably, the grasping force is determined according to the following formula: , in, In the formula, To execute the crawling force in real time; For the preset target grasping force; The total real-time grasping force for the fruit when the grid-like grasping surface is engaged. This is the feedback adjustment coefficient, used to adjust the response of the feedback intensity; This is the baseline proportional coefficient under trial mining conditions; Rate the ripeness of the fruit; This is the center score value for the trial production level; This is the sensitivity adjustment factor.

[0013] Preferably, an adaptive fruit harvesting method, implemented according to any of the above-described adaptive fruit harvesting systems, includes the following steps: The actuator module drives two grid-like gripping surfaces to contact the fruit surface. Multiple sensors are used to collect the changes in the magnetic field of the contact blocks at each intersection point on the grid-like gripping surface after force is applied. Based on the change in magnetic field after each contact block is subjected to force, the grasping force at each intersection point is obtained, and the effective contact area and micro-slippage index between the grid-like grasping surface and the fruit are obtained by using the real-time change of the grasping force at each intersection point. The real-time total gripping force and wrapping force of the grid-shaped gripping surface on the fruit are obtained based on the gripping force at each intersection and the effective contact area. Based on the real-time total grasping force, wrapping force, and micro-slippage index, the fruit is input into a preset maturity discrimination module to classify the maturity level. The system matches the maturity level with the preset target grasping force, and dynamically adjusts the real-time total grasping force of the two grid-shaped grasping faces on the fruit based on the matching result through the actuator module. At the same time, it acquires the real-time execution grasping force until the real-time execution grasping force reaches the preset target grasping force, so as to complete the fruit harvesting without damage.

[0014] Compared with the prior art, the fruit adaptive harvesting method and system of the present invention has the following beneficial effects: This device uses a biomimetic spiderweb grasping mechanism to contact the fruit surface, acquiring multi-dimensional tactile information about the grid-like grasping surface and the fruit in real time. Based on the spatial force distribution of the grasping force at the intersection of the fruit within the grid-like grasping surface, as well as parameters such as the effective contact area and micro-slippage index, it dynamically calculates the grasping force and wrapping force. By inputting the real-time total grasping force, wrapping force, and micro-slippage index into the maturity discrimination module, it accurately identifies the fruit's maturity level. Then, an adaptive force control algorithm dynamically adjusts the grasping force according to the maturity level, and finally, harvesting is performed after the force stabilizes. This method effectively solves the problem of grasping force imbalance caused by the inability to perceive maturity in existing technologies, realizing adaptive and non-destructive harvesting of fruits at different maturity levels, and significantly reducing fruit damage caused by excessive gripping or insufficient force. Attached Figure Description

[0015] Figure 1This is a flowchart of the adaptive fruit harvesting method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the adaptive fruit harvesting system in an embodiment of the present invention.

[0016] Explanation of reference numerals in the attached figures: 1. Contact block; 2. Grid-like gripping surface; 3. Actuator module; 4. Connecting bracket; 5. Flexible thin wire. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0019] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0020] See Figure 1 and Figure 2 As shown, in order to achieve adaptive adjustment of grasping force by accurately sensing tactile information and ripeness, thus realizing damage-free fruit harvesting and avoiding fruit damage caused by excessive gripping or insufficient force, this embodiment provides a fruit adaptive harvesting method and system. The fruit adaptive harvesting system mainly consists of a biomimetic spiderweb grasping mechanism, a collection module, a processing unit, and a ripeness determination module, and also includes a drive mechanism for the overall movement of the biomimetic spiderweb grasping mechanism.

[0021] The biomimetic spiderweb grasping mechanism includes an actuator module 3 and two opposing grid-like grasping surfaces 2. Each grid-like grasping surface 2 is a mesh structure woven from flexible fine threads 5. At each intersection of the fine threads within the grid-like grasping surface 2, a contact block 1 made of magnetoelastic material is provided. These contact blocks generate magnetic field changes when subjected to force and deformation. The acquisition module includes sensors corresponding one-to-one with the multiple contact blocks 1. Each sensor is mounted on its corresponding contact block 1 to monitor the magnetic field changes of the contact block 1 in real time, thereby obtaining the grasping force at the corresponding intersection. The acquisition unit also collects tactile information between the fruit and the grid-like grasping surface 2 in real time. This tactile information includes the spatial force distribution of the grasping force at the intersection, the contact area, and micro-slippage indicators. The processing unit includes a first processor, a second processor, a maturity discrimination module, and a third processor. The first processor is electrically connected to the multiple sensors and obtains the grasping force at the intersection of each contact block based on the real-time magnetic field changes fed back by the sensors. It also obtains the effective contact area and micro-slippage indicators between the grid-like grasping surface 2 and the fruit through the real-time changes in the grasping force at each intersection. The second processor is electrically connected to the first processor and is used to obtain the real-time total grasping force and wrapping force of the grid-like grasping surface 2 on the fruit through the effective contact area and the grasping force at each intersection. The maturity discrimination module is electrically connected to both the first and second processors and is used to score the maturity of the fruit through the real-time total grasping force, wrapping force, and micro-slip index, and to classify the fruit maturity level accordingly. Preferably, the maturity discrimination module is calibrated using training data, its loss function is Focal Loss or cross-entropy, the validation strategy is hierarchical K-fold cross-validation, and the maturity discrimination AUC is ≥ 0.90. The third processor is electrically connected to both the maturity assessment module and the actuator module. It matches the maturity level with a preset target grasping force and, based on the matching result, dynamically adjusts the real-time total grasping force of the two grid-like grasping surfaces 2 on the fruit via the actuator module. An adaptive force control algorithm is used to obtain the real-time grasping force until it reaches the preset target grasping force, at which point the actuator module stops operating. At this point, a drive mechanism moves the actuator module as a whole to complete the non-destructive harvesting of the fruit. The drive mechanism preferably uses a linear motor. After the harvested fruit is moved to the designated position by the drive mechanism, the actuator module releases the fruit, which falls to an external receiving system.

[0022] The actuator module 3 connects two grid-shaped gripping surfaces 2 via a connecting bracket 4, and adjusts the gripping force of the two grid-shaped gripping surfaces 2 on the fruit according to the adjustment instructions fed back by the third processor, thus completing the fruit harvesting without damage. Specifically, the actuator module 3 includes a driver and two harvesting arms. The driver uses a servo motor or a stepper motor and has two moving ends. Each of the two moving ends is fixed to a corresponding grid-shaped gripping surface 2 via a harvesting arm. The driver is used to drive the two harvesting arms to move towards each other through the two moving ends, so that the two grid-shaped gripping surfaces 2 can grasp and harvest the fruit. The actuator module 3 can also be composed of a pneumatic soft hand, a servo motor, or a stepper motor. One grid-shaped gripping surface 2 is fixed to the outer shell of the actuator module 3, and the other grid-shaped gripping surface 2 is fixed to the pneumatic soft hand. The pneumatic soft hand is installed at the output end of the motor. The motor drives the pneumatic soft hand to move the grid-shaped gripping surface 2 connected to it, thereby completing the grasping of the fruit by the two grid-shaped gripping surfaces 2. This facilitates the adjustment of the gripping force of the two grid-shaped gripping surfaces 2.

[0023] It should be noted that the first processor, the second processor, and the third processor can be different processing units integrated in the same chip, or they can be discrete processing chips whose functions are implemented by software modules or hardware circuits.

[0024] See Figure 1 As shown, based on the above system design, the adaptive fruit harvesting method is implemented through the following steps: S1: The actuator module 3 drives the two grid-shaped gripping surfaces 2 to contact the fruit surface; S2: Use multiple sensors to collect the changes in the magnetic field of the contact block 1 at each intersection point on the grid-like gripping surface 2 after being subjected to force; S3: Based on the change in magnetic field after each contact block 1 is subjected to force, the grasping force at each intersection point is obtained, and the effective contact area and micro-slip index between the grid-shaped grasping surface 2 and the fruit are obtained by using the real-time change of the grasping force at each intersection point. S4: Based on the gripping force and effective contact area at each intersection, obtain the real-time total gripping force and wrapping force of the two grid-shaped gripping surfaces on the fruit; S5: Based on real-time total grasping force, wrapping force and micro-slip index, input the preset maturity discrimination module to score the maturity of the fruit and classify the maturity level of the fruit accordingly. S6: Matches the fruit based on maturity level and preset target grasping force, and dynamically adjusts the real-time total grasping force of the two grid-like grasping surfaces 2 on the fruit based on the matching result through the actuator module. At the same time, it acquires the real-time execution grasping force until the real-time execution grasping force reaches the preset target grasping force to complete the non-destructive harvesting of the fruit.

[0025] The following is a further detailed explanation based on the above system and method: The fruit adaptive harvesting system proposed in this invention is a non-destructive harvesting mode based on biomimetic principles. Combining sensors, magnetoelastic materials, and adaptive force control technology, it aims to solve the problems of insufficient mechanical sensing, inaccurate maturity identification, and harvesting damage in existing automated harvesting technologies. By accurately sensing the surface characteristics and maturity of the fruit, this invention can intelligently adjust the gripping force to achieve non-destructive harvesting. This technology can improve the efficiency and quality of automated harvesting in large-scale agricultural production, especially in the harvesting of high-value fruits (such as strawberries, grapes, and tomatoes), significantly reducing the damage rate.

[0026] 1. The core technical solution of this invention 1.1 Bionic Spider Web Grasping Mechanism: This structure imitates the structure of spider webs in nature. Through the interlacing of flexible fine threads 5, a grid-like grasping surface 2 with strong adaptability and good grasping force distribution is formed, which can effectively reduce the damage to the fruit surface caused by local pressure.

[0027] Magnetoelastic Material and Mechanical Feedback Mechanism: Contact blocks 1 of magnetoelastic material are embedded at the intersections of the fine lines within the grid-like gripping surface 2. Preferably, the magnetoelastic material can be a ferromagnetic alloy, magnetic rubber, or magnetic polymer. This material deforms under stress, and its deformation is linearly related to the applied force. The system calculates the gripping force of the grid-like gripping surface 2 in real time by detecting changes in the magnetic field, forming real-time tactile feedback. Specifically, the system uses sensors placed at multiple intersections to detect changes in the magnetic field. Preferably, the sensors are flexible three-dimensional tactile sensors, such as Hall effect sensors, magnetoresistive sensors, or fluxgate sensors. These sensors can monitor the changes in magnetic field strength caused by the deformation of the magnetoelastic material in real time, thereby calculating the gripping force applied to the fruit by the grid-like gripping surface 2. Combined with tactile information such as the spatial force distribution, contact area, and micro-slippage index of the fruit on the grid-like gripping surface 2, which is fed back in real time by the acquisition unit, the gripping force is dynamically adjusted to ensure the stability and safety of the fruit during harvesting.

[0028] Grab surface design: The gripping surface of this structure is composed of multiple interwoven fine lines, forming a geometric grid similar to a spider web. Each intersection has a distinct node, exhibiting a hexagonal or triangular pattern, creating a dense and uniform grid.

[0029] Through the fine design of the grid, this structure can maximize the dispersion of contact forces and reduce pressure on the object being grasped, making it especially suitable for non-destructive grasping of soft objects (such as solids).

[0030] Structural form: The gripping structure should ideally consist of two circular, mesh-like gripping surfaces 2, resembling two hemispheres arranged opposite each other. The outer edge of each circular mesh-like gripping surface 2 is connected to the central fixed structure via a frame or support component. The support component can be made of a relatively rigid material to maintain the stability of the entire structure.

[0031] The fine lines along the edges of the mesh-like gripping surfaces 2 are fixedly connected to the circular frame. The two mesh-like gripping surfaces 2 should be symmetrically arranged. When they are in contact, the overall shape presents two symmetrical meshes, similar to a pair of wheels.

[0032] Connection and support: The central structure, serving as a support platform, is located at the center of the grid-like gripping surface 2. It is preferably a sphere with a diameter less than 2 cm. One side of the sphere, connecting to the grid-like gripping surface 2, is made of elastic rubber, while the other side is made of rigid material for connecting to the output end of the actuator module 3. Preferably, the actuator module 3 can also be a multi-degree-of-freedom robotic arm or similar actuator structure, capable of supporting and driving the movement of the grid-like gripping surface 2. This connection method gives the grid-like gripping surface 2 sufficient flexibility while maintaining high structural strength, enabling it to withstand mechanical changes during the gripping process.

[0033] 1.2 Fruit Maturity Recognition and Force Control Closed-Loop System: This invention detects the fruit's surface hardness, elasticity, and other physical properties by generating contact between the fruit and contact blocks 1 at the intersections of multiple fine lines within the grid-like grasping surface 2. By combining the maturity discrimination module with real-time tactile information from the acquisition unit, the system intelligently identifies the fruit's maturity. Based on the maturity data, the system can dynamically adjust the motion parameters of the actuator module 3 to adjust the grasping force of the grid-like grasping surface 2, ensuring damage-free harvesting of fruits at different maturity levels. The maturity discrimination module is calibrated using training data, employing a BiLSTM-based deep learning model with Focal Loss or cross-entropy as the loss function and hierarchical K-fold cross-validation as the validation strategy.

[0034] The maturity level classification is based on real-time total grasping force, wrapping force, and micro-slippage indicators, which are obtained in real time by inputting into the preset maturity judgment module.

[0035] In this invention, the micro-slip index is used to monitor minute slippage between the fruit and the grid-like grasping surface in real time. During fruit harvesting, a certain degree of relative slippage occurs between the grid-like grasping surface 2 and the fruit surface. This slippage usually reflects the unevenness of the grasping force, changes in the pressure distribution of the contact surface, and may even indicate the ripeness of the fruit.

[0036] Specific indications of the microslip index: Microslip refers to a tiny displacement between the gripping surface and the object's surface. This displacement does not immediately lead to obvious gripping failure, but it provides crucial information about the contact surfaces and mechanical state.

[0037] When there is slight slippage between the grid-shaped gripping surface 2 and the fruit surface, it indicates that the force distribution is uneven or the contact surface is not completely in contact with the fruit surface during the gripping process. This is usually related to factors such as the ripeness of the fruit, surface elasticity, and fruit shape.

[0038] Microslip can be obtained in the following ways: Magnetoelastic materials and deformation: A magnetoelastic material is embedded at the intersection of the thin wires. This material generates a changing magnetic field as the wires deform. By monitoring the changes in the magnetic field, the system can calculate the deformation and further estimate the amount of microslip.

[0039] Feedback is processed by the unit: The micro-slip index is transmitted to the processing unit in real time. The processing unit dynamically adjusts the gripping strategy based on the monitored slip amount to ensure that the gripping force is always within an appropriate range.

[0040] 1.3 Adaptive force control strategy and real-time closed-loop feedback control: This invention adopts an adaptive force control algorithm to adjust the gripping force according to the real-time ripeness of the fruit, surface elasticity and sensor feedback data, thereby ensuring that the most suitable gripping force is maintained during the fruit picking process and avoiding damage to the fruit due to over-gripping or under-gripping.

[0041] 2. Design and Implementation of a Bionic Spiderweb Grasping Mechanism The biomimetic spider web grasping mechanism is a key innovation of this invention, inspired by the structural characteristics of spider webs. Spider webs possess extremely high adaptability and flexibility, providing sufficient grasping force without damaging prey. To mimic this characteristic, this invention designs a grasping mesh based on the interweaving of flexible fine threads 5.

[0042] 2.1 Design and Material Selection of Flexible Fine Wire 5 The design of the thin string must ensure that the applied force is evenly distributed during the grasping process, avoiding localized pressure concentration that could damage the fruit. The string material should possess good flexibility, elasticity, and durability; highly elastic synthetic fibers, such as ultra-high molecular weight polyethylene (UHMWPE) fibers and polyurethane elastomers, are commonly used. These materials offer the following advantages:

[0043] Flexibility and elasticity: It can effectively disperse the applied contact force and avoid damage to the fruit surface due to force concentration.

[0044] High toughness and wear resistance: Ensures that the fine wires are not easily worn during the gripping process, thus improving the service life of the gripping structure.

[0045] Adjustability: The flexibility and adjustability of the fine threads allow the gripping structure to adapt to fruits of different shapes and sizes.

[0046] 2.2 Mesh Structure and Mechanical Analysis Fine threads are interwoven to form a grid structure, creating a flexible gripping surface. The grid design needs to be optimized according to the size, shape, and surface characteristics of the fruit to maximize contact with the fruit surface during gripping, ensuring that the fruit does not slip or fail to be picked due to insufficient grip.

[0047] The design of a grid needs to consider the following aspects: Line spacing: The spacing between the lines in the grid needs to be adjusted according to the size, shape, and harvesting requirements of the target fruit. For example, for smaller fruits (such as strawberries and grapes), a denser grid can be used, with the spacing between the lines in the grid less than 5mm; for larger fruits (such as apples and oranges), a grid design with a larger spacing is required.

[0048] Mechanical distribution: The grid structure design should ensure that the contact force is evenly distributed on the fruit surface to avoid excessive local pressure. The spacing between the fine lines in the grid should be between 5mm and 15mm.

[0049] Adaptability: The gripping structure has sufficient flexibility to allow the gripping process to adapt to the shape and surface characteristics of different fruits, avoiding gripping failures caused by irregular or smooth fruit surfaces.

[0050] 2.3 Spatial Force Recognition and Enveloping Force In biomimetic spiderweb grasping mechanisms, the ability to identify and encapsulate spatial forces is crucial. Each thread and its intersection experiences a certain force upon contact with the object. These forces are distributed across the grasping surface, and the force at each intersection is related to the deformation of adjacent threads. By identifying the distribution and changes in these forces, the system can effectively monitor the force state throughout the entire grasping process.

[0051] • Distribution and identification of spatial forces: Each intersection of the thin lines P i The resulting intersection gripping force F i It can be expressed by the following formula: , in, Intersection of thin lines P i The gripping force at the intersection of the contact blocks; Intersection P iThe stiffness coefficient of contact block 1, stiffness coefficient The specific values ​​will vary depending on the materials of the contact block and the wire, the diameter and length of the wire, and the specific requirements of the application. Common value ranges are: polyurethane elastomer: 100N / m to 1000N / m; ultra-high molecular weight polyethylene: 300N / m to 2000N / m; metal wire: 5000N / m to 20000N / m. Intersection of thin lines P i The deformation of the contact block at the point of contact is obtained by the elongation or shortening of the thin wire. This refers to the displacement of the intersection of thin wires caused by external forces. The system measures the deformation in real time through changes in the magnetic field of the magnetoelastic material, thereby obtaining the deformation of each intersection and making force-controlled adjustments.

[0052] The spatial distribution of force, 𝐹(𝑥,𝑦,𝑧), is the combined result of the forces at all the intersections of the fine lines. The force on the entire grid, i.e., the real-time total grasping force of the grid-like grasping surface on the fruit, can be calculated using the following integral formula: , in, The total real-time grasping force for the fruit when the grid-like grasping surface is engaged. V The volume of the grid-like gripping surface; For the grid-shaped gripping surface in x , y , z Spatial force distribution in direction , Intersection of thin lines P i The component of the gripping force at the intersection point in the x-direction; Intersection of thin lines P i The component of the grasping force at the intersection point in the y-direction; Intersection of thin lines P i The component of the gripping force at the intersection point in the z-direction; i =1, 2, 3... The integral represents the sum of the gripping forces at all intersections on the grid-like gripping surface.

[0053] • Wrapping force and surface adaptability of the object: The adaptability of the fine mesh lines allows the mesh-like gripping surface 2 to adjust to irregular shapes and variations in the object's surface. Each fine mesh node dynamically adjusts according to the deformation of the object's surface, forming a uniform mechanical wrapping. The wrapping force is [value missing]. wrap This can be expressed by the following formula:

[0054] , in, For real-time wrapping force of the grid-like grasping surface facing the fruit; S The effective contact area between the fruit and the grid-like grasping surface; For the grid-shaped gripping surface in x , y , z Spatial force distribution in direction , Intersection of thin lines P i The component of the gripping force at the intersection point in the x-direction; Intersection of thin lines P i The component of the grasping force at the intersection point in the y-direction; Intersection of thin lines P i The component of the gripping force at the intersection point in the z-direction; i =1, 2, 3... ds It is the area of ​​the contact surface element.

[0055] The method for obtaining the effective contact area S includes: the system pre-stores the surface area of ​​a single grid cell. (i.e., the area of ​​a single grid within the grid-like gripping surface) and the effective contact force threshold The acquisition unit calculates each intersection point in real time. P i Magnitude of the resultant force at the point and with threshold Compare; count all that satisfy Number of valid intersections Finally, through the formula The effective contact area between the fruit and the grid-like grasping surface is calculated. This method utilizes the structural characteristics of the grid itself to achieve highly efficient estimation of the contact area without additional sensors.

[0056] The gripping force of a grasping structure depends not only on the fine filament structure of the grasping surface but also on the geometry of the contact area on the object's surface. The adaptability of the gripping force can be described by the normal and tangential forces on the contact surfaces. The normal force on the object's surface is then defined. F normal and tangential force F tangential The real-time wrapping force of the grid-like grasping surface on the fruit can also be expressed by the following equation:

[0057] , in, The real-time wrapping force of the grid-like gripping surface 2 on the fruit; The normal force on the surface of the fruit. The tangential force on the fruit surface.

[0058] This formula allows the system to adjust the gripping force in real time, ensuring that the grid-shaped gripping surface 2 maintains good fit with the fruit surface and avoiding indentations or damage to the fruit caused by uneven force distribution.

[0059] 2.4 Dynamic Adaptation and Real-time Adjustment of Spatial Forces The biomimetic spiderweb grasping mechanism can not only sense the magnitude of spatial forces, but also adapt to changes in objects by dynamically adjusting the force distribution of the mesh-like grasping surface 2. During the grasping process, each fine line of the mesh-like grasping surface 2 adjusts according to the irregular shape of the object's surface, thereby maintaining a stable grasping force.

[0060] • Mechanical feedback closed loop To ensure the accuracy of the gripping force, the biomimetic spiderweb gripping mechanism is adjusted in real time through a mechanical feedback closed-loop system. The force on each fine thread node is transmitted to the control system, which adjusts the gripping force in real time based on this feedback information. This feedback mechanism can be described by the following closed-loop feedback formula:

[0061] The closed-loop feedback formula uses an adaptive force control algorithm to dynamically adjust the gripping force of the grid-like gripping surface, and is determined according to the following formula: , In the formula, To execute the crawling force in real time; For the preset target grasping force; The total real-time grasping force for the fruit when the grid-like grasping surface is engaged. This is the feedback adjustment coefficient, used to adjust the response of the feedback intensity; This is the baseline proportional coefficient under trial mining conditions; Rate the ripeness of the fruit; This is the center score value for the trial production level; This is the sensitivity adjustment factor.

[0062] • Adaptive force control strategy Through the above feedback control, the biomimetic spiderweb grasping mechanism can adjust its grasping force in real time according to changes in force during the grasping process, ensuring that the force at each contact point adapts to changes in the object's surface. This adaptive force control strategy enables the grasping structure to effectively avoid fruit damage caused by unstable grasping.

[0063] 2.5 Advantages of Bionic Structures The biomimetic spider web grasping mechanism has the following unique advantages: Uniform force distribution: The interwoven structure of fine threads can effectively distribute the gripping force to the fruit surface, reducing damage to the fruit surface caused by excessive pressure.

[0064] Highly adaptable: This structure can automatically adjust the gripping method according to the different shapes, surface characteristics and ripeness of the fruit, minimizing fruit damage.

[0065] Efficient gripping: This structure provides a stable gripping force during the gripping process, preventing fruit from falling off or failing to be picked due to insufficient gripping.

[0066] 3. Magnetoelastic Materials and Mechanical Feedback Mechanisms As a key innovation of this invention, magnetoelastic materials play a crucial role in sensing mechanical changes during the grasping process in real time and feeding this information back to the control system.

[0067] 3.1 Properties and Applications of Magnetoelastic Materials Magnetoelastic materials are materials that can deform under external forces and exhibit a linear relationship with the applied force. The application of magnetoelastic materials can convert mechanical changes into magnetic field changes, thereby providing real-time feedback signals for control systems.

[0068] The relationship between deformation and force: There is a direct linear relationship between the deformation of a magnetoelastic material and the applied force. Let the stiffness coefficient of the magnetoelastic material be... Its shape is Then the grasping force It can be calculated using the following formula:

[0069] , in, For the applied gripping force, This is the stiffness coefficient of the magnetoelastic material. This represents the deformation of the magnetoelastic material.

[0070] Magnetic field changes and force feedback: When a magnetoelastic material deforms, the distribution of its internal magnetic field changes. The system monitors the change in magnetic field strength using a magnetic field sensor (such as a Hall effect sensor) to obtain the applied gripping force.

[0071] 3.2 Force Feedback Mechanism and Control System By monitoring the deformation of the magnetoelastic material in real time, the system can calculate the currently applied gripping force and feed it back to the control system. The control system uses this information to adjust the gripping force to ensure that the fruit is not damaged during harvesting.

[0072] Real-time monitoring and feedback: The system monitors changes in the magnetic field through sensors, combines this with the known material stiffness coefficient, calculates the applied gripping force in real time, and adjusts the harvesting strategy based on feedback.

[0073] Mechanical feedback closed loop: The control system dynamically adjusts the gripping force based on real-time feedback data to ensure that the mechanical control during the harvesting process is in the best state.

[0074] 4. Maturity Identification and Force Control Closed-Loop Control System This invention achieves precise mechanical control during automated harvesting by integrating maturity recognition technology with a force control closed-loop control system. The system can determine the ripeness of the fruit in real time and dynamically adjust the gripping force accordingly to ensure damage-free harvesting.

[0075] 4.1 Maturity Identification Technology Fruits at different stages of maturity exhibit significant differences in physical properties (such as surface hardness and elastic modulus). To dynamically adjust the gripping force, this invention incorporates maturity recognition technology to sense the maturity of the fruit in real time.

[0076] Real-time maturity assessment: By using real-time sensor data (such as measurements of fruit surface pressure and deformation) and combining this data with the fruit's physical characteristics, the system can calculate the fruit's maturity. During harvesting, the system correlates the fruit's maturity with the grasping force and dynamically adjusts the harvesting intensity accordingly.

[0077] Maturity assessment mechanism The maturity assessment module determines the fruit's maturity level by acquiring real-time tactile information between the fruit and the grid-like grasping surface, particularly the real-time total grasping force, wrapping force, and micro-slippage indices. The system inputs this mechanical information into a pre-defined maturity assessment module, which calculates and outputs the fruit's maturity level in real time. The model uses deep learning algorithms (such as BiLSTM) to process this time-series data and combines it with mechanical feedback from multiple sensors, such as normal force, tangential force, micro-slippage, and contact area, to determine the fruit's maturity.

[0078] Maturity level The maturity assessment module typically classifies fruit maturity into three main levels: Immature: The fruit surface is hard, with strong gripping force and little micro-slippage.

[0079] Suitable for harvesting (early ripening): The surface of the fruit begins to soften, the gripping force is moderate, and slight slippage increases.

[0080] Overripe: The fruit surface is softer, the gripping force is weaker, and microslippage is significantly increased.

[0081] Judgment based on real-time total gripping force, wrapping force, and micro-slippage indicators Real-time total gripping force, wrapping force, and microslippage indices are key factors in assessing fruit firmness and surface elasticity. Immature fruit requires a larger gripping force, while mature fruit requires a smaller gripping force.

[0082] The microslip index reflects the firmness of the fruit surface; softer fruits tend to exhibit more microslip. Fruit maturity is further confirmed by observing changes in microslip.

[0083] By collecting these parameters in real time, a fruit ripeness score can be calculated, and adaptive force control adjustments can be made based on this score to ensure damage-free harvesting. This process combines tactile perception and feedback from deep learning models to ensure that the fruit is not damaged throughout the harvesting process.

[0084] Specifically, the system uses BiLSTM to process time series data, and we define a comprehensive tactile maturity index. M (Maturity Index). This index serves as a score for fruit maturity and is calculated using normalized parameters obtained during the trial harvesting phase.

[0085] First, define the temporal feature vector input to the BiLSTM. X ( t ): , in: for t Real-time total capture capability at any given moment.

[0086] for t Real-time wrapping force.

[0087] for t The micro-slippage index at any given moment.

[0088] A for t Effective contact area at any given moment.

[0089] The maturity score is obtained by weighted mapping through the output layer of BiLSTM. M : , in, For activation function, For the sake of patent description, we can express it as an equivalent mechanical weighting function:

[0090] , in: Rate the ripeness of the fruit.

[0091] This is the benchmark gripping force obtained during the trial harvesting phase under the standard hardness of similar fruits.

[0092] The benchmark encapsulation force is the standard hardness of similar fruits obtained during the trial harvesting phase.

[0093] This represents the baseline contact area under the standard hardness of similar fruits obtained during the trial harvesting phase.

[0094] , , and All are weighted coefficients.

[0095] The trial mining threshold determined based on the trial mining data ( M 2, M 1), that is M 1 represents the upper limit of the fruit firmness threshold during trial harvesting. M 2 represents the lower limit of the threshold for fruit firmness during trial harvesting. M To determine the firmness of the fruit during trial harvesting, maturity was divided into three grades as shown in Table 1 below: Table 1. Maturity Level Classification Table To achieve adaptive crawling, we need to... Dynamically adjust feedback dynamic coefficient This allows for the calculation of the final real-time execution crawling power. .

[0096] It is a coefficient that decreases with increasing maturity. We define it using a piecewise linear function or an exponentially decaying function:

[0097] , This is the feedback adjustment coefficient, used to adjust the response of the feedback intensity; This is the baseline proportional coefficient under trial mining conditions; Rate the ripeness of the fruit; This is the center score value for the trial production level; This is the sensitivity adjustment factor.

[0098] 4.2 Force-controlled closed-loop control system The force-controlled closed-loop control system is an important component of this invention. Its function is to adjust the gripping force based on real-time feedback of tactile information and maturity data.

[0099] Mechanical feedback and adjustment: The system acquires the fruit's gripping force data through real-time monitoring of the magnetoelastic material and, combined with ripeness information, automatically adjusts the gripping force. This force adjustment is achieved using the following formula:

[0100] , in To execute the grabbing force in real time, For the purpose of grasping, This refers to the change in gripping force adjusted based on real-time feedback.

[0101] Closed-loop control: The control system adjusts the gripping force based on tactile feedback and ripeness information to ensure that the fruit is not subjected to excessive pressure or insufficient gripping throughout the harvesting process.

[0102] 5. System Hardware Configuration and Integration The system hardware of this invention consists of multiple modules, including sensors, a control system, actuators, and a data processing unit. These modules exchange data through a communication interface, enabling real-time control and optimization during the harvesting process.

[0103] 5.1 Sensors The sensor consists of mechanical sensors used to collect the mechanical information of the fruit in real time. It mainly includes:

[0104] Hall effect sensor: Used to monitor the deformation of magnetoelastic materials and calculate the applied gripping force by measuring changes in the magnetic field.

[0105] 5.2 Control System The control system processes sensor data to calculate the required gripping force and adjusts the gripping force via actuators. The control system includes an embedded microprocessor, a signal processing unit, and a feedback adjustment module.

[0106] 5.3 Actuator Module The actuator adjusts the gripping force according to the instructions of the control system. Common actuators include pneumatic soft grippers, servo motors, or stepper motors. The actuator ensures stable gripping of the fruit through precise force adjustment.

[0107] 5.4 Data Processing and Feedback The system performs real-time data analysis through the data processing unit and transmits tactile feedback to the control system to ensure that the gripping force is adjusted in real time and that the fruit is not damaged during the harvesting process.

[0108] 6. System Operation Flow Initialization: When the system starts up, it performs a self-test to ensure that all sensors and actuators are working properly.

[0109] Fruit contact: When the gripping mechanism comes into contact with the fruit, the sensor begins to collect contact force data.

[0110] Maturity Analysis: The system analyzes the maturity of the fruit based on sensor data and calculates the target grasping force.

[0111] Force control adjustment: The system adjusts the gripping force in real time to ensure stable fruit gripping.

[0112] Non-destructive harvesting: Throughout the harvesting process, the system maintains the optimal gripping force through feedback adjustment to avoid damage to the fruit.

[0113] 7. Deep Learning-Driven Maturity Assessment Module 7.1 Objectives and Overall Plan To address the inconsistency in mechanical response caused by differences in variety, size, and physiological maturity among individual fruits, this invention introduces a lightweight BiLSTM-based deep learning ripeness assessment module, building upon tactile perception and force control closed-loop systems. This module takes multi-channel time-series data (including normal / tangential force, torque, deformation, contact area estimation, and microslip indexes) as input, outputs fruit maturity levels, and maps the results to upper limits of grasping force, loading slope, and release strategies, thus forming an integrated "ripness assessment-force control" closed loop.

[0114] 7.2 Data Acquisition Protocol and Signal Definition To improve discrimination stability, a standardized detection sequence was established and data was collected in real time: Detection sequence: linear ascending load (0→Fmax, slope SN / s) → load holding → linear unloading (F_max→0).

[0115] Sampling frequency: ≥200 Hz; Window length: 1.5–2.0 s.

[0116] Input signal and output result: Triaxial force / torque: Fx,Fy,Fz,Mx,My,Mz F_x,F_y,F_z,M_x,My,M_z Fx,Fy,Fz,Mx,My,Mz; Displacement / deformation: δ\deltaδ (estimated by magnetoelastic displacement or tactile array); Label format: Three categories (unripe / suitable for harvesting / overripe).

[0117] 7.3 Training and Calibration Loss function: Focal Loss / Cross-entropy is used for classification.

[0118] Regularization: Dropout (0.3), random undersampling alignment.

[0119] Verification strategy: stratified K-fold by "quality / batch".

[0120] 7.4 Performance and Technical Effects If the AUC of the product is ≥ 0.90, the recall rate for suitable products is ≥ 0.92. Average inference latency at the edge is ≤20 ms.

[0121] Experimental design: Randomly group the same batch into A / B groups and compare whether the ripeness assessment module is enabled; stratify the damage rate and fruit drop rate by variety / size.

[0122] 6. The technical effects achieved by this invention, and its advantages and disadvantages. Technical effects: 1. Non-destructive harvesting: By accurately sensing tactile information and ripeness, non-destructive harvesting of fruits can be achieved, avoiding fruit damage caused by excessive clamping or insufficient force.

[0123] 2. High-sensitivity tactile perception: Through a three-dimensional tactile sensor network, the contact state between the fruit and the gripper can be captured in real time, including the distribution of contact force and micro-slippage information, which improves the system's response speed and accuracy.

[0124] 3. Maturity Recognition and Adaptive Force Control: The system can automatically adjust the gripping force according to the fruit's maturity and changes in contact force, ensuring optimal results when harvesting fruits at different maturity levels.

[0125] advantage: High flexibility and adaptability: The gripping mechanism of this invention can cope with the changes in the shape, surface characteristics and ripeness of different kinds of fruits, and is highly adaptable.

[0126] Avoid damage: Through real-time tactile sensing and adaptive adjustments, ensure that the fruit is not subjected to excessive pressure or damage during the harvesting process.

[0127] High efficiency: This technology can improve the efficiency of automated harvesting systems, especially in large-scale operations, by reducing human intervention.

[0128] shortcoming: High complexity: The system integrates tactile perception, maturity recognition and adaptive force control functions, which is relatively complex and may require fine debugging and optimization during implementation.

[0129] High cost: Due to the use of magnetoelastic materials and high-sensitivity sensors, the initial manufacturing cost may be high, but the cost is expected to gradually decrease with technological advancements and mass production.

[0130] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A fruit adaptive harvesting system, characterized in that, include: A biomimetic spider web grasping mechanism includes an actuator module and two opposing grid-like grasping surfaces. The grid-like grasping surfaces are mesh structures woven from multiple flexible fine threads. At each intersection of the fine threads within the grid-like grasping surfaces, there are contact blocks made of magnetoelastic material. These contact blocks are used to generate magnetic field changes when subjected to force and deformation. The actuator module includes a driver and two picking arms. The driver has two moving ends, each of which is fixed to a corresponding grid-like grasping surface via one of the picking arms. The driver is used to drive the two picking arms to move towards or away from each other through the two moving ends, so that the two grid-like grasping surfaces can grasp and pick the fruit. The acquisition module includes sensors corresponding to multiple contact blocks, and the multiple sensors are respectively fixed on the corresponding contact blocks. The sensors are used to monitor the magnetic field changes of the contact blocks in real time when deformation occurs. The first processor is electrically connected to the multiple sensors and is used to obtain the gripping force at the intersection of each contact block based on the real-time magnetic field changes fed back by each sensor, and to obtain the effective contact area and micro-slip index between the grid-like gripping surface and the fruit through the real-time changes of the gripping force at each intersection. The second processor, electrically connected to the first processor, is used to obtain the real-time total grasping force and wrapping force of the grid-shaped grasping surface on the fruit through the effective contact area and the grasping force at each intersection. The maturity determination module is electrically connected to the first processor and the second processor respectively, and is used to classify the maturity level of the fruit by means of the real-time total grasping force, wrapping force and micro-slip index; The third processor is electrically connected to the second processor, the maturity discrimination module, and the driver, respectively. It is used to match the maturity level with the preset target grasping force, and dynamically adjust the real-time total grasping force of the two grid-shaped grasping faces on the fruit based on the matching result through the driver. At the same time, it acquires the real-time execution grasping force until the real-time execution grasping force reaches the preset target grasping force, so as to complete the non-destructive harvesting of the fruit.

2. The fruit adaptive harvesting system according to claim 1, characterized in that, The contact blocks of the magnetoelastic material include ferromagnetic alloys, magnetic rubbers, or magnetic polymers.

3. The fruit adaptive harvesting system according to claim 1, characterized in that, The gripping force at the intersection point corresponding to the contact block is determined according to the following formula: ， in, Intersection of thin lines P i The gripping force at the intersection of the contact blocks; Intersection P i The stiffness coefficient of the contact block; Intersection of thin lines P i The contact block shape at the location.

4. The fruit adaptive harvesting system according to claim 3, characterized in that, The real-time total grasping force of the grid-like grasping surface on the fruit is determined according to the following formula: ， in, The total real-time grasping force for the fruit when the grid-like grasping surface is engaged. V The volume of the grid-like gripping surface; For the grid-shaped gripping surface in x , y , z Spatial force distribution in direction , Intersection of thin lines P i The component of the gripping force at the intersection point in the x-direction; Intersection of thin lines P i The component of the grasping force at the intersection point in the y-direction; Intersection of thin lines P i The component of the gripping force at the intersection point in the z-direction; i =1, 2, 3...

5. The fruit adaptive harvesting system according to claim 3, characterized in that, The real-time wrapping force of the grid-like grasping surface on the fruit is determined according to the following formula: ， in, For real-time wrapping force of the grid-like grasping surface facing the fruit; S The effective contact area between the fruit and the grid-like grasping surface. , The effective number of contact blocks within the grid-like grasping surface of the fruit. The surface area of ​​each grid within the grid-like grasping surface; For the grid-shaped gripping surface in x , y , z Spatial force distribution in direction , Intersection of thin lines P i The component of the gripping force at the intersection point in the x-direction; Intersection of thin lines P i The component of the grasping force at the intersection point in the y-direction; Intersection of thin lines P i The component of the gripping force at the intersection point in the z-direction; i =1, 2, 3...

6. The fruit adaptive harvesting system according to claim 1, characterized in that, The preset maturity determination module is determined according to the following formula: ， In the formula: Rate the ripeness of the fruit; For real-time total crawling power; For real-time wrapping force; For microslip index; A Effective contact area; The benchmark gripping force is the standard hardness of similar fruits obtained during the trial harvesting phase. The benchmark wrapping force is the standard hardness of similar fruits obtained during the trial harvesting phase. This represents the baseline contact area under the standard hardness of similar fruits obtained during the trial harvesting phase. , , and All are weighted coefficients.

7. The fruit adaptive harvesting system according to claim 6, characterized in that, The maturity level is determined based on the fruit's maturity score and trial harvest threshold. The fruit maturity levels include Class I - immature, Class II - suitable for harvesting, and Class III - overripe, and are determined according to the following formula: Class I - Immature: M > M 1; Class II - Suitable for use: M 2≤ M ≤ M 1; Class III - Overripe: M < M 2; In the formula, M To test the firmness of the fruit during trial harvesting; M 1 represents the upper limit of the fruit firmness threshold during trial harvesting. M 2 represents the lower limit of the threshold for fruit hardness during trial harvesting.

8. The fruit adaptive harvesting system according to claim 6, characterized in that, The real-time execution grasping force is determined according to the following formula: ， in, In the formula, To execute the crawling force in real time; For the preset target grasping force; The total real-time grasping force for the fruit when the grid-like grasping surface is engaged. This is the feedback adjustment coefficient, used to adjust the response of the feedback intensity; This is the baseline proportional coefficient under trial mining conditions; Rate the ripeness of the fruit; This is the center score value for the trial production level; This is the sensitivity adjustment factor.

9. A fruit adaptive harvesting method, implemented according to the fruit adaptive harvesting system described in any one of claims 1-8, characterized in that, Includes the following steps: The actuator module drives two grid-like gripping surfaces to contact the fruit surface. Multiple sensors are used to collect the changes in the magnetic field of the contact blocks at each intersection point on the grid-like gripping surface after force is applied. Based on the change in magnetic field after each contact block is subjected to force, the grasping force at each intersection point is obtained, and the effective contact area and micro-slippage index between the grid-like grasping surface and the fruit are obtained by using the real-time change of the grasping force at each intersection point. The real-time total gripping force and wrapping force of the grid-shaped gripping surface on the fruit are obtained based on the gripping force at each intersection and the effective contact area. Based on the real-time total grasping force, wrapping force, and micro-slippage index, the fruit is input into a preset maturity discrimination module to classify the maturity level. The system matches the maturity level with the preset target grasping force, and obtains the real-time execution grasping force based on the matching result and the real-time total grasping force of the grid-shaped grasping surface on the fruit. The real-time execution grasping force is then fed back to the actuator module to complete the fruit harvesting without damage.

Images