Leaf Pruning End-Effector for Adaptive Positioning at the Branch–Stem Junction of Tomato Plants

Abstract

To address the needs of mechanized tomato leaf pruning, this paper presents the design of an end-effector capable of adaptive positioning at the base of the branch. This design effectively prevents infection at the cut sites of a residual branch and protects the rest of the plant from damage. The design objectives for the pruning actuator were established through the measurement of key parameters related to the morphology and mechanical properties of the lateral branch. Based on this foundation, we developed an innovative gripper featuring a spiral guide groove, enabling simultaneous axial traction and radial cutting of the branch. This design ensures that the branch–stem junction conforms to the cutting blade under shear stress, achieving the required adaptive positioning. By analyzing the mechanical properties of the lateral branch, we modeled the actuator’s traction and cutting forces to determine the key geometric parameters of the spiral fingers and the necessary driving torque. We validated the actuator’s operational effectiveness through discrete element simulation and practical application tests. The experimental results indicate that when removing the branch, a traction force of 32.5 N and a cutting force of 66.3 N are generated. Harvesting effectiveness tests conducted in the tomato field achieved a success rate of 85%. This research offers technical support for the development of handheld pruning devices and pruning robots.

1. Introduction

Tomatoes are a widely cultivated vegetable globally and an important source of human nutrition and food. China ranks first in the cultivation and consumption of tomatoes, with a planting area of about 18 million acres and an annual production of approximately 67 million tons, accounting for about 30% of the global total [1,2]. However, due to the lack of mechanization in critical production processes, labor costs have risen annually, exceeding 60% of the total production costs [3,4]. Pruning and leaf removal, essential steps in tomato cultivation management, are carried out throughout the entire production cycle [5,6,7]. By removing leaves from the area around mature fruits, nutrient and water distribution and transportation can be adjusted, reducing nutrient consumption by branches and leaves and ensuring that water and nutrients prioritize fruit ripening [8]. Additionally, proper pruning of branches and leaves can enhance air circulation, reduce pest and disease density, promote nutrient absorption and tomato root growth, prevent premature aging of tomatoes [9], and improve tomato plants’ ventilation and light exposure, significantly benefiting fruit quality and yield [10].

When pruning tomatoes, cutting at the separation tissue where the branch connects to the main stem is necessary, as the protective layer at the fracture surface can effectively block pathogen infection in the plant. Currently, the selective removal of tomato branches and leaves mainly relies on workers using handheld tools for pruning, requiring weekly pruning operations. This process is labor-intensive and demands high concentration from operators, making it one of the most labor-consuming tasks in the tomato cultivation cycle [11,12]. Existing mechanized pruning equipment for plants utilizes a robotic pruning arm as its critical component, tasked with accurately positioning the end-effector at the designated target location. Zhang et al. [13] developed a pruning machine specifically designed for jujube trees, with key components tailored to the trees’ growth characteristics and specific pruning requirements. The end-effector is equipped with circular saw blades, while the pruning blades on the side and top trimming assemblies are driven in a rotational motion to achieve effective cutting functionality. This pruning machine is capable of rapidly shaping and pruning large-scale, densely planted, dwarf jujube trees. Experimental results demonstrated an average missed pruning rate of 5.46% and an average stubble breakage rate of 5.10%. Ma et al. [14] designed a novel PRRPR-configured robotic arm specifically for mechanical pruning in greenhouse tomatoes. This robotic arm positions the end-effector with the required orientation to handle branch growing in random directions effectively. An inverse kinematics solution for the pruning arm was developed based on a multi-objective optimization algorithm. Experimental results indicate that the optimized robotic arm can reach and prune over 89.94% of branches within the target area, thereby meeting pruning performance requirements. Zhao et al. [15] developed an apple harvesting robot, in which the design of the end-effector takes into account the biological characteristics of the target object (e.g., spherical fruits such as apples). The end-effector features a spoon-shaped two-finger gripper and is equipped with sensors for pressure, position, collision, and vision. Current pruning machinery largely relies on robotic arms to transport pruning end-effectors to designated locations for effective positioning at removal points [16]. However, studies focusing on precise positioning and removal of the target branch and stems directly through the pruning end-effector itself are relatively limited. In most existing pruning machinery, robotic arms are utilized to position pruning end-effectors at designated locations, effectively targeting removal points. However, research on achieving precise positioning and removal of the target branch and stems directly through the end-effector remains relatively scarce.

Robots with autonomous operation capabilities are considered an effective tool to replace manual labor in pruning tasks, with the pruning end-effector being a critical component of pruning robots. Significant progress has been made in developing and applying fruit and vegetable pruning robots, with some products already achieving industrial application. Botteril et al. [17] developed a robotic platform for pruning grapevines specifically designed to reduce visual errors caused by sunlight. The platform’s end-effector is mounted on a 6-degree-of-freedom robotic arm and positioned 1.6 m behind a camera, enabling comprehensive 3D reconstruction of the grapevine before pruning. To facilitate branch pruning, the end-effector is equipped with a milling cutter connected to a high-speed motor. Zahid et al. [18,19] developed a six-degree-of-freedom robotic arm for pruning apple tree branches, equipped with a shear blade as the end-effector. Xiong et al. [20] developed a multi-arm robotic hand for strawberry harvesting, with each arm featuring a three-finger gripper as the end-effector. Existing pruning robots primarily focus on enhancing positioning accuracy through algorithmic improvements, with relatively limited research dedicated to improving precision and pruning efficiency by optimizing mechanical structures. Positioning pruning points solely via visual servoing demands high recognition accuracy and extensive computational resources, yet faces challenges such as limited accuracy and low stability. Thus, adopting an adaptive end-effector positioning approach enables more precise and stable targeting of pruning points, effectively balancing accuracy with stability. Due to the diverse shapes, textures, and mechanical properties of different crops, most end-effectors are specifically designed and developed for particular applications. While these end-effectors have shown promising performance, their effectiveness may be constrained when applied to pruning or harvesting of other types of plants or vegetables.

This paper addresses the efficiency and reliability requirements in tomato pruning operations by conducting a comprehensive measurement and analysis of the mechanical characteristics of branches. We propose a design for a tomato pruning end-effector equipped with adaptive positioning functionality. The innovation of this end-effector lies in its integration of intelligent sensing technology, which enables real-time adaptation to varying stem morphologies, thereby significantly enhancing the precision and efficiency of pruning tasks. A mechanical model was established to describe the relationship between the stem and the end-effector, allowing for calculating and analyzing the essential parameters for the end-effector’s key components. Through a combination of simulation experiments and practical application validation, we evaluated these components’ parameters and operational performance. The findings of this research provide crucial technical support for the advancement of pruning robot technology and serve as a valuable reference for future design optimizations.

2. Measurement and Analysis of the Physical Properties of Branches

2.1. Morphological Characteristics of Branch

Modern greenhouse tomato production is characterized by industrialization, density, and standardization. As shown in Figure 1a, when tomato plants reach a height of 400–500 mm, tipping treatment is initiated to ensure the regularity and stability of subsequent growth. By regularly adjusting the inclination angle of the main stem through pruning, the height of the pruning and harvesting workspace can be maintained within the range of 1600–1800 mm, forming a pruning area.

As shown in Figure 1b, tomato plants grow naturally curved along pre-arranged support wires. The leaves of the tomato are compound, with branches growing alternately along the main stem. The branch connects the main stem to the entire leaf, and the abscission zone is the layer of cells formed at the base of the leaf or leaf stalk before the leaf drops. The base of the branch is the ideal area for pruning operations. Small leafages grow alternately on either side of the branch. The area from the base of the branch to the first leafage is designated as the operational zone for the pruning actuator, where pruning can be performed to avoid interference and collisions with the leafages during the clamping of the branch. Through the measurement and statistical analysis of the morphological parameters of the tomato plant tissues in the pruning operation area, it was found that the average diameter of the main stem $D$ is 10.9 mm, while the average diameter of the branch $d$ is 7.2 mm. The average distance $\Delta l$ between the base of the branch and the first leafage is 53 mm. The spatial angle between the pruning operation area’s branch and the main stem’s central axis ranges from 60° to 85°.

2.2. Measurement of Branch Mechanical Properties

Understanding the mechanical properties of the pruning operation targets is essential for ensuring reliable clamping and cutting. To clarify the relevant material mechanical properties of the branch, this study focuses on measuring and analyzing characteristics such as compressive yield, shear fracture, and friction. Tomato lateral branches in the fruiting stage for 30 days were selected as specimens, each with a sample length of 100 mm, for a total of 120 samples.

2.2.1. Measurement of Compressive Properties

The universal testing machine manufactured by MTS SYSTEMS (CHINA) CO., LTD., Shanghai, China was utilized to measure the compressive properties. The MTS E43 electronic universal testing machine was equipped with two load sensors with ranges of 100 N (±0.30%) and 10 kN (±0.20%) to measure the compressive yield characteristics. The branch was placed on the testing platform (Figure 2a), and the indenter was set to move downward at a speed of 2 mm/s. When a decrease in pressure was detected on the branch (Figure 2b), it was concluded that the internal fibers were damaged. The pressure recorded at the moment of damage provided the minimum pressure required to cause compressive failure in branches of different diameters.

The lateral branches were classified into four groups according to diameter: 6–7 mm, 7–8 mm, 8–9 mm, and 9–10 mm. For each diameter group, five compression yield tests were performed, and the average value was recorded as the final result for each group. With an increasing diameter of branches, the degree of lignification shows a significant increase. Consequently, the pressure required to induce damage rises, as does the compressive force the branch can withstand before radial crushing occurs. Consequently, the compressive force that could be tolerated during radial compression failure also increased. Specifically, the average compressive failure force for branches with a diameter of 6.0–7.0 mm was 16.2 N, for 7.0–8.0 mm it was 21.5 N, for 8–9 mm it was 30.9 N, and for 9–10 mm it was 37.3 N.

2.2.2. Measurement of Shear Properties

As shown in Figure 3a, the target stem is fixed to the shear platform of the universal testing machine to measure the shear fracture properties. The fixed jaws grip the blade to be tested, and the cutting operation is performed on the branch at a constant speed of 2 mm/s until the target stem is completely cut through. The shear fracture force exerted by the cutting blade on the tomato stem is measured.

As shown in Figure 3b, the average shear force required for branches with a diameter of 6–7 mm to fracture is 65.5 N, 7–8 mm is 70.3 N, 8–9 mm is 77.5 N, and 9–10 mm is 85.5 N. This indicates that as the diameter of the branch increases, the shear force required for fracture significantly increases. The lateral branches were classified into four groups according to diameter: 6–7 mm, 7–8 mm, 8–9 mm, and 9–10 mm. For each diameter group, for each diameter range of the branch, five shear tests were conducted, and the average value was taken as the final result. The branches to be tested were divided into four groups based on their diameters, and five shear characteristic tests were performed for each diameter range to obtain the average result.

2.2.3. Measurement of Friction Properties

Considering the need for non-destructive clamping of the branch, this study selected a hard rubber that balances rigidity and toughness as the material for the fingers. A testing setup was established to measure the friction properties of the branch stem, as shown in Figure 4. A rubber sheet was fixed to the indenter of the universal testing machine, and the normal pressure was applied to the branch. The branch is pulled by the traction gauge until it is detached from the indenter. The maximum pulling force measured under this pressure represents the frictional force between the branch and the rubber. This enables the friction coefficient to be calculated, allowing for determination of the minimum gripping force required between the lateral branch and the end-effector.

The lateral branches were divided into four groups, each subjected to normal pressures of 5 N, 10 N, 15 N, and 20 N, respectively. Five friction tests were conducted for each group, and the average of the maximum pulling force and maximum static friction force was recorded, as shown in

Table 1. Measurement Results of Maximum Static Friction Coefficient.

Normal Pressure/N	Friction Force/N	Static Friction Coefficient
5	2.2	0.44
10	4.3	0.43
15	6.5	0.43
20	8.7	0.44

. The test results indicate that, under normal pressures of 5 N, 10 N, 15 N, and 20 N, the average static friction coefficient across the four groups was 0.435. This value was selected for subsequent friction force calculations in this study.

3. Design of Pruning End-Effector

3.1. Overall Composition and Principle

As shown in Figure 5, the pruning end-effector is divided into the lateral branch clamping positioning component and the cutting component. The lateral branch clamping positioning component is mainly composed of two helical roller fingers, which are driven by a DC gear motor to rotate in the opposite direction. When the branch enters between the fingers of the rotating spiral roller, the branch is subjected to two forces at the same time: the friction force acting parallel to the clamped branch, and the vertical traction force exerted on the branch by the helical structure. The finger is output power by the driving motor, and the end finger is driven by the gear set to rotate synchronously in the opposite direction, and the rotation direction is indicated by the green arrow. The end finger surface of the opposite roller is provided with a spiral groove extending in the axial direction and rotating in the opposite direction. In order to avoid damage to the tomato stem epidermis, the roughness of the end finger surface is appropriately increased, and traction is achieved through the friction between the roller finger and the branch. The end of the roller finger is connected with the actuator body through a fixed iron core contained inside, and the outer part of the iron core is wrapped with rubber material. The rubber causes elastic deformation to the roller finger during clamping, which avoids radial extrusion damage of the brittle stem while giving the branch a clamping force.

The cutting component is mainly composed of a side baffle plate and a blade. The side baffle plate is provided with a gap to facilitate the entry of the branch, and the blade is fixed at the gap of the baffle plate. After the friction and traction of the roller finger, the root of the lateral branch contacts the side baffle, and the lateral branch rotates in the spiral grove to the blade—the movement direction is shown by the blue arrow—and is pushed by the groove side until the blade cuts it. In order to ensure that the cut is as close as possible to the root of the branch when cutting, the branch should contact the baffle under the action of friction on the roller finger at the end. The branch should contact the blade under the action of traction—the movement direction of the branch is indicated by the orange arrow—so as to ensure the accurate positioning of the cut.

3.2. Branch Traction Part Design

As a key component of the end-effector, the rubber fingers have their structural design and material selection affecting the gripping performance on the branch. To achieve adaptive positioning of the end-effector on the branch, it is necessary to ensure that the end-effector generates sufficient friction. Therefore, rubber materials that balance elasticity and friction are selected as the surface wrapping material for the fingers. Setting an appropriate distance between the two pairs of roller fingers ensures that the branches are not crushed while providing adequate gripping force. The axial distance between the two fingers is $d$, the distance between the two fingers is $w$, and the finger radius is $r$, as shown in Figure 6.

When a branch with a diameter of $D$ is gripped by the fingers, it undergoes elastic deformation, with a deformation amount of $\Delta d$. Therefore, we have:

$$\left\{\begin{array}{l}D=2r+w\\ \Delta d=d-w\end{array}\right.$$

(1)

To estimate the contact area $A$ of the tomato stem under the gripping action of the end fingers, the contact surface is projected onto a plane and considered an ellipse. Calculating the area of this projected surface allows us to determine the contact area of the tomato stem under the gripping of the end-effector, as shown in Figure 7.

Let the intersection points of the two parabolas with the axis be $x_1$ and $x_2$, where $x_1$ < $x_2$. The contact area can be calculated based on integration as follows:

$$A={\displaystyle {\int }_{x_1}^{x_2}(y_1-y_2)}dx={\displaystyle {\int }_{x_1}^{x_2}[(a_1-a_2)x^2+(b_1-b_2)x+(c_1-c_2)]}dx$$

(2)

This simplifies to:

$$A={\displaystyle \frac{\sqrt{b^2-4ac}\left(b^2-4ac\right)}{6a^2}}$$

(3)

The branch undergoes positive strain under the action of the normal stress σ. The relationship between the elastic modulus of the branch and the compressive force applied to it is as follows:

$$\left\{\begin{array}{l}{F}_N=\delta A\\ \delta =\epsilon E_0\\ \epsilon ={\displaystyle \frac{\Delta d}{d}}\end{array}\right.$$

(4)

In the equation: $E_0$ is the elastic modulus of the branch, and is derived from the Measurement of Compressive Properties analysis. $F_N$ is derived from the Measurement of Compressive Properties analysis. $F_N$ is the compressive force applied to the stem.

The relationship between the compressive force applied to the branch and the deformation of the branch is as follows:

$$F_N={\displaystyle \frac{\Delta d{E}_0}{dA}}$$

(5)

When the fingers grip the branch, the branch comes into contact with the grooves on the finger surface and fits into the grooves. This results in static friction between the branch and the fingers, which represents the force exerted by the end-effector’s fingers on the branch. Therefore, the relationship between the traction force between the branch and the grooves, the gap between the fingers, and the elastic modulus of the branch is as follows:

$$f=\mu F_N=\mu {\displaystyle \frac{d-w}{dA}}{E}_0$$

(6)

3.3. Branch Cutting Part Design

To ensure that the wounds on the tomato stem heal smoothly after the pruning operation and to minimize the cutting area, it is necessary to ensure that the cut surface is as perpendicular as possible to the target branch. The spiral protrusions on the fingers provide a traction force towards the blade direction, resulting in a component of force that helps the base of the branch to detach after being cut by the blade. To ensure that the stem is successfully cut by the blade, it is necessary to perform a force analysis during the cutting process, as shown in Figure 8.

The two fingers are equipped with spiral grooves in opposite directions. During the rotation of the fingers, the side walls of the grooves exert a normal traction force $F_c$ on the branch, which acts as the reaction force to the shearing force applied by the blade. This force can be decomposed into a horizontal traction force $F_a$ and a thrust $F_b$ directed toward the blade, which should satisfy Equation (7).

$$\left\{\begin{array}{l}{F}_b=F_c\cos \theta \\ F_a=F_c\sin \theta \\ T_c=\left(f+F_a\right)r\end{array}\right.$$

(7)

The torque $T_2$ required for cutting satisfies the following equation:

$$T_c=\left(f+F_b\tan \theta \right)r$$

(8)

From the above equation, it can be concluded that the minimum torque required by the drive motor to cut the branch with the blade successfully is $T\ge T_c$.

To ensure that the cut is as close as possible to the main stem of the tomato, the branch should be pulled until it reaches the base before making contact with the blade to initiate cutting. It is assumed that there is no relative sliding between the fingers and the stem. The displacement of the branch away from the main stem under the action of the traction force is equal to the circumference of the fingers, and the normal displacement of the branch against the side walls of the finger grooves corresponds to the distance between the pre-positioned grooves of the fingers and the blade. If the branch moves under the traction force and reaches the base while simultaneously making contact with the blade due to the normal force from the side walls of the finger grooves, let the number of spiral groove turns required for the traction force be $n_1$, and the number of spiral groove turns required for the compressive force towards the blade be $n_2$. The following relationship can be established:

$$\left\{\begin{array}{l}\Delta l=n_{1}2\pi r\\ H=n_{2}2\pi r\tan \theta \end{array}\right.$$

(9)

Thus, the branch first experiences the action of the traction force until it reaches the base, and then it makes contact with the blade under the normal force from the side walls of the finger grooves. This ensures that the branch is pulled to the base before being sheared. It is necessary to ensure that $n_1\le n_2$, which leads to the following equation:

$$\theta \le \arctan {\displaystyle \frac{H}{\Delta l}}$$

(10)

In the equation: H represents the distance between the pre-positioned groove of the finger and the blade.

If the lead angle of the finger’s threaded groove is less than $\theta$, it can ensure that the cut is located at the base of the branch, allowing for rapid healing of wounds after pruning in greenhouse tomato factory cultivation. This helps to avoid pathogen infections and ensures that the cut surface is smooth and neat.

4. Simulation Experiment

4.1. EDEM Model Establishment

The removal process of the pruning end-effector is complex and random, resulting in varying effects on the removal of branches with different diameters. Additionally, small changes in parameters such as the forces and deformations experienced by the stem are difficult to capture and measure. Therefore, the EDEM 2022 simulation is used to model the removal of the branch by the end-effector of the pruning robot. In EDEM, a Single Sphere is selected for simulating the modeling of the tomato stem. The Physical Radius R is set to 0.4 mm, and the Contact Radius R′ is set to 0.8 mm. The relationship between the two is given by R′ ≥ 2R′ [21,22]. The tomato stem is composed of numerous cells that adhere to one another, forming a complex system with intrinsic organic connections. Therefore, in the discrete element simulation, the stem can be regarded as being made up of a large number of interacting discrete particles. To accurately simulate this complex interaction, this study employs the Hertz–Mindlin with bonding contact model [23,24]. This model effectively simulates the complex mechanical behavior resulting from adhesive interactions between particles. Based on the force characteristics, the bonding parameters for the bonding keys in the branch discrete element model are set, as shown in

Table 2. Simulation Parameters.

Material Parameters	Value
Poisson’s ratio of tomato stem particles	0.4
Density of tomato stem particles (kg/m3)	800
Shear modulus of tomato stem particles (Pa)	1 × 107
Contact Parameters	Value
Stem-blade restitution coefficient	0.3
Stem-blade static friction coefficient	0.35
Contact model parameters	Value
Normal Stiffness per unit area (N/m3)	9 × 109
Normal Strength (Pa)	2 × 109
Shear Strength (Pa)	1 × 109

. The three-dimensional model of the tomato pruning robot’s end-effector is created using SolidWorks2022, and the model is imported into EDEM in “step” [25] format to complete the initial setup. The motion settings for the end fingers of the end-effector are configured, with rotation set to reverse synchronous rotation.

4.2. Pruning Simulation

In the tomato pruning process, a time step of 5 × 10⁵ s is set, with data saved every 0.02 s for a total of 500 steps. First, branches with an 8 mm diameter were used to examine the effect of the gap between two pairs of roller fingers on the gripping force. The branch particles are subjected to compressive forces exerted by the roller fingers. The pressure flow diagram is shown in Figure 9a; during the simulation, particles change color from green to red to indicate an increase in force in that area.

Figure 9b shows the pressure experienced by an 8 mm diameter branch during pruning operations under different finger gaps; the gaps between the roller fingers are set to 6 mm, 7 mm, and 8 mm, with roller fingers rotating in reverse at a speed of 5 rad/s. After passing through the pre-positioned grooves of the roller fingers, the branch enters the spiral grooves of the fingers. The compressive force initially increases, reaches a maximum point, then slightly decreases and stabilizes. When the gap between the roller fingers is between 6 mm and 8 mm, the compressive force exerted on the branch does not crush it, resulting in better gripping effects during rotation. Therefore, the selected range for the finger gap is appropriate.

Then, the gap between the two pairs of roller fingers was set to 8 mm, and tests were conducted to assess the gripping force exerted by the roller fingers on branches of varying diameters. The 4 mm, 6 mm, and 8 mm diameter branches are placed into the pre-positioned grooves of the roller fingers of the end-effector. Under the gripping and traction actions of the roller fingers, the branch moves toward the blade, and under the shearing action of the blade, the branch detaches. Figure 10a shows the pressure contour plot during the shearing process, with colors changing from blue to green to red, indicating a gradual increase in the applied force. The figure demonstrates that the branch is subjected to both gripping and shearing forces.

Figure 10b shows the diagram of the shearing forces experienced by branches of different diameters. The graph indicates that the shearing force first increases until it reaches a maximum value, after which the branch detaches and the shearing force decreases sharply. The vertical coordinates of the inflection points in the graph represent the maximum shearing forces experienced by the branch: 65.3 N for an 8 mm diameter branch, 62.8 N for a 6 mm diameter branch, and 53.1 N for a 4 mm diameter branch. This indicates that the design of the end-effector’s two pairs of roller fingers, based on these parameters, can effectively remove branches while avoiding damage to the main stem.

5. Field Test

5.1. Clamping Test

Experiments were conducted on the pruning end-effector to verify the feasibility and practicality of the design plan. As shown in Figure 11, the first step involved using a force gauge to measure the radial pressure exerted by the end fingers of the pruning tool on tomato stems of varying diameters, as well as the pulling force directed away from the main stem. Next, a tie strap was used to connect the distal end of the test stem, away from the root, to a digital force gauge, which was fixed to the main body of the pruning end-effector. During the test, tomato branches positioned at the same distance from the base were placed in the grooves of the end fingers. Finally, a series of non-destructive clamping mechanical property tests were conducted on the tomato branch to measure the actual pulling force generated by the end fingers. The maximum value recorded by the digital force gauge was taken as the pulling force acting away from the main stem under the clamping action of the end fingers.

5.2. Pruning Test

By conducting removal tests on branches of different tomato varieties, the adaptive positioning and non-destructive removal characteristics of the pruning end-effector in practical work are validated. The pruning end-effector was tested for its removal effectiveness. The end-effector, held manually, was used to prune branches within the designated pruning area, and the results were documented. To assess the actual working effectiveness of the pruning end-effector, the standards for pruning operations are unified: a successful removal is determined if the branch residue is within 5 mm and there is no damage to the main stem. Based on these evaluation criteria, field tests are conducted on the tomato pruning end-effector, as shown in Figure 12.

5.3. Results and Discussion

5.3.1. Clamping Test Results

The statistical curve of the clamping characteristics of the end fingers for branches with diameters ranging from 4–8 mm is presented in Figure 13. The graph clearly shows that, within this diameter range, both the radial pressure and the pulling force exerted on the branch by the end fingers increase steadily as the branch diameter increases. For branches with diameters smaller than 4.6 mm, the elastic modulus is below the average value, and surface tissue detachment occurs, leading to slippage and a success rate lower than expected. Conversely, for branches with diameters larger than 7.2 mm, the tissue becomes more lignified, the elastic modulus increases, and effective clamping becomes more difficult. The end-effector performs optimally on branches with diameters between 4.4 mm and 7.2 mm, with experimental results consistent with the anticipated outcomes.

Based on Figure 13, it can be observed that the curve exhibits a noticeable stepwise increase in the range of 4.4 to 7.2 mm on the vertical axis. This suggests that, within this interval, the diameter of the tomato lateral branches is optimal, and the overall performance of the pruning end-effector is at its best. When the diameter of the branch is less than 4.4 mm, its elastic modulus is lower than the average value, and some surface tissue detachment occurs, causing slipping; therefore, the success rate is lower than expected. When the diameter exceeds 7.2 mm, the degree of lignification of the tissue is higher, resulting in a greater elastic modulus and making it difficult to achieve proper gripping. The end-effector shows better gripping performance for branches with diameters between 4.4 mm and 7.2 mm, which meets expectations.

5.3.2. Pruning Test Results

Through testing, the reliability and stability of the pruning end-effector in actual usage scenarios have been verified. Under suitable conditions in a greenhouse environment, handheld pruning end-effectors were used to conduct branch removal tests on tomato plants with branch diameters ranging from 4–12 mm. A total of 100 test samples were categorized by diameter into four equal groups: 4–6 mm, 6–8 mm, 8–10 mm, and 10–12 mm. Experimental data were then collected and analyzed for each group. After multiple tests and summarizing the relevant data, the removal test data for the branch by the pruning end-effector were statistically analyzed. The experimental data are shown in

Table 3. Statistical Analysis of Test Results.

Branch Diameter/mm	Tests Number	Successes Number	Average Residual Length/mm	Success Rate
4–6	25	19	8.5	76%
6–8	25	23	4.4	92%
8–10	25	22	3.8	88%
10–12	25	21	6.5	84%

, indicating that the pruning end-effector achieved a removal success rate of 85% for tomato branches.

The pruning end-effector generally achieves higher removal success rates with smaller-diameter tomato lateral branches. However, as the branch base diameter increases, the success rate declines; incomplete cuts often leave fibers partially connected to the main stem after compression-induced damage. This finding aligns with elastic gripping mechanics analysis, which indicates that the pruning end-effector is less effective at gripping and removing larger-diameter branches while performing better on smaller-diameter branches. Further analysis of shear failure mechanisms suggests that the fragile, succulent nature of tomato branches contributes to this issue. Under radial pressure, branches are prone to damage and juice leakage, which reduces the friction coefficient and leads to slippage between the roller fingers and the branch, resulting in shear failure.

6. Conclusions

This paper introduces a novel tomato leaf pruning end-effector capable of adaptive positioning at the base of the branch, thereby reducing excessive residual branch length and minimizing the risk of infection at the cut site. By measuring the growth morphology and mechanical properties of branches, the design requirements for mechanized pruning have been clarified, including the physical characteristics of stem gripping, compression, cutting, and friction.

For branch stems, the gripping fingers with spiral grooves can achieve bidirectional synchronous drive for radial traction and axial cutting. Additionally, the relationship between the two driving forces and the key geometric parameters of the fingers has been established. Discrete element simulations and physical tests indicate that for branch stems ranging from 4–8 mm, the actuator can generate a traction force of 32.5 N and a cutting force of 66.3 N.

Simulation-based optimization field tests were conducted on tomato stems of different diameters to measure the traction force exerted away from the main stem under the gripping of the end fingers. The test results indicate that the pruning end-effector demonstrates good non-destructive gripping characteristics for smaller diameter tomato stems. After comprehensive removal tests, it was concluded that the success rate of the pruning end-effector for removing branches is 85%. This has preliminarily achieved standardized operations for adaptive positioning of removal points and non-destructive removal of branches, providing valuable references for subsequent design improvements.

The end-effector designed in this paper is limited to the design of the mechanical structure. In future research, the end-effector developed in this study could be integrated into automated tomato pruning equipment, with electrical signal control applied to the rotation of the roller fingers. This advancement would enhance the level of automation in tomato pruning operations.