Design and Experiment of Gripper for Greenhouse Plug Seedling Transplanting Based on EDM

: Many plug seedling grippers have a complex structure and bulky volume in design. They have poor performance in holding substrate and keeping a high integrity rate. In this paper, a new multi ‐ needle seedling gripper is proposed. Based on the elasto ‐ plastic contact model (ECM) in the EDEM, numerous particles with different material properties are modeled to simulate real seedling substrate and to study the interaction between steel needle and growing substrate. Taking the me ‐ dium integrity rate as an evaluation index, explore the influence of different needle diameters, in ‐ sertion depths, and insertion and grabbing speeds on the substrate integrity through the response surface method. The optimized technical parameters of the gripper was obtained through Design ‐ Expert. The results show that the insertion depth of needles has a significant effect on the integrity of the raising medium, and the optimal depth is 40 mm. Insertion and grasping speed and depth have an interaction effect, while other factors have no significant influence on substrate integrity. Optimization results imply that when the diameter of the needle was 3 mm, the insertion depth was 40 mm, and the insertion and grabbing speed was 1 m/s, the growing medium integrity rate was the highest, reaching 89.10%. Under these parameters, prototype tests were conducted, and the highest medium integrity rate was 87.34%, which increased by an average of 7.25% compared with existing studies. The research has reference value for improving the operation quality of the plug seedling transplanter and the design of the seedling gripper. It also shows that solving problems of needle ‐ discrete substrate interaction based on the discrete element method is feasible.


Introduction
Greenhouses are the main production facility for off-season vegetables. According to industry statistics, in 2018, vegetables produced in greenhouses in China accounted for 30% of the total vegetable production that year, which is crucial to meeting people's daily nutritional intake needs [1]. To facilitate the intensive management of vegetable seedlings and improve the space utilization rate and seedling quality, raising seedlings with substrate and plug tray is widely used in greenhouse vegetable production currently [2]. When the seedlings grow to a suitable stage, they will be transplanted from the highdensity plug tray to the low-density one, providing a more independent nutrient environment with more space for the seedlings. Transplanted seedlings can shorten the growth cycle and increase yield effectively [3]. However, manual transplanting has high intensity, low efficiency, and high labor cost [4], and the survival rate of seedlings is hard to guarantee [5]. Therefore, mechanized transplanting of plug seedlings is the future development trend.
In the process of mechanical transplanting, the gripper is a core component for the plug seedling transplanter. As an end effector, grippers are used for seedling grabbing, holding, and releasing, so their design would directly affect the quality and efficiency of the transplanter [6]. Choque M. C. J. [7] proposed a four-needle seedling gripper. For this device, the steel needles were driving to rise and fall in the tube through a steering gear and screw mechanism, completing actions of seedling grasping and releasing, but there is a slow operation speed. To address this issue, Jiang Z. H. [8] used a cylinder to drive steel needles and applied a sensor system to monitor and analyze the adhesion between the substrate and the tray wall as well as the extrusion force on the substrate when grabbing a plug seedling. Jorg O. J. [9] and Li B. [10] designed a new needle-type seedling gripper. In their project, the insertion direction of needles is not parallel to the edges and corners of the tray hole, resulting in a low growing medium integrity of the seedling root system during transplanting. Therefore, this kind of transplanting claw is only suitable for the substrate with strong cohesion. Han L. H. [11] proposed a grasping claw, in which the expansion and contraction of four needles were controlled by four small cylinders separately, so this device is large in volume. In his works, the effects of needle insertion depth, operating pressure, and other factors on the transplanting success rate also are tested.
According to the references mentioned above, the existing seedling grippers have a complex structure and heavy volume shape, and when design researchers have not paid enough attention to maintaining the high integrity rate of the substrate. Additionally, the methods in these reports are experiments, which are time-consuming and labor-intensive, and it is hard to achieve ideal results. The integrity of the medium block is important in transplanting, as the seedlings with more substrate attached have a higher survival rate generally. The growing medium is discrete, and although it forms a fixed shape under the constraints of the tray cell, the stable status is easy to destroy during the transplanting.
The goal of this research is to develop a four-needle seedling gripper and simplify its structural design and volume on the premise of a high integrity rate of the substrate block, and try to study the interaction process of steel needle and substrate based on discrete element simulation and analysis methods, to quickly obtain the optimized technical parameters of the seedling gripper and accelerate the product design. Finally, the transplanting effect of the new seedling gripper based on this rapid design theory is verified by experiments.

Structural Design and Kinematics Analysis of Gripper
The structure of the seedling gripper is shown in Figure 1. Four-pin type, commonly used, was applied as a seedling-grasping method, because it is more stable than two-pin or single pin. The whole gripper includes an air cylinder, cylinder seat, screw rod, push plate, steel needle, spring, guide tube, and fixing plate. The cylinder fixes on the cylinder seat by the screw rod, and its pushrod is linked with the push plate. The guide tube connects with the fixing plate, the steel needle slides into the guide tube, and the spring mounts on the upper end of the steel needle. When transplanting the seedling gripper first moves to the top of a target seedling growing in a high-density plug tray. The cylinder rod controls the push plate and pushes the needle down to insert the seedling substrate, and then the gripper moves up vertically to complete the plug seedling grab. When the gripper moves to the top of an empty cell of the low-density plug tray, the cylinder rod and push plate retract. At this time, the steel needle moves up and returns to the guide tube under the spring force, and the seedling substrate is pushed down from the steel needle and falls into the empty cell under the reverse thrust of the lower end face of the guide tube, then plug seedling released.
This design scheme has a relatively simple structure and a small size for the gripper. In addition, when the steel needle is inserted into the substrate, the elastic force generated by the compression of the spring increases continuously, and the acceleration of the steel needle into the substrate will gradually decrease. When the needle reaches the bottom of the cell and stops moving, then the gripper moves upward, which is beneficial to form a speed buffer and avoid a strong speed impact tearing the substrate block. The focus of the geometric parameter design of the seedling gripper is how to grab as much substrate as possible from the tray cell. Figure 2 shows the overall geometry and connection of the seedling gripper for grasping the seedlings. To maximize the integrity of the substrate, the needle should be as close as possible to the edge of the tray cell while reaching the bottom of the tray cell. Therefore, the best approach is to keep the needle parallel to the edge of the tray cell and grab more of the substrate at the maximum insertion depth. To prevent the needles from interfering with the tray frame, the openings must be smaller than the width of a tray cell. Studies have shown that the optimal value of e0 is 2-3 mm [12], so before the two steel needles are inserted into the substrate, the opening LBE will be 4-6 mm smaller than the width of the tray cell. At the same time, to prevent the steel needle from damaging the plug tray and protect the root of the plug seedling at the cell bottom, there should be a space e1 between the needle tip and the cell bottom after the needle inserts. According to the geometric characteristic in the figure, the initial opening and insertion depth of the needle can be determined as: In the equations, LBE is the initial opening width of the steel needle, and LBB′ is the depth of the needles inserted into the substrate.
The guide tube of the gripper leads the slide direction of the steel needle. To protect the plug seedlings from mechanical damage during the transplanting process, there should be enough space inside the transplanting gripper. The size of the inner space is decided by guide tube length, and the length configuration of the tube needs to refer to the plant height of the plug seedlings. Different vegetable seedlings have different heights when transplanted, and it is hard to meet the size requirements of all plug seedlings in practical design. Therefore, this study takes the most common size requirements of tomato, pepper, and cucumber seedlings as references, and designs the tube length LDE. The LDE is at least: The length of the guide tube calculated by the formula (2) is a critical value. If the connecting neck DD′ is located at the upper end of the guide tube LDE, the stability of this structure is poor. Therefore, the length of the tube LDE is appropriately extended upwards in the design.
When the plug seedlings are released, the cylinder rod drives the push plate to retract. At this time, the steel needle will be retracted into the guide tube under the elastic force of the spring, thereby pulling down the substrate. To meet this working requirement, the elastic force Ft must overcome the vertical component of the gravity G of the steel needle and the frictional resistance Ff of the substrate block. Jiang Z. [8] found that the extrusion force on the steel needle after it inserts into the substrate was between 9.45 and 18.67 N, so the frictional force on the steel needle could be calculated when it was pulled out of the substrate. Comparing the elastic force Ft calculated according to the equation group (3) with the above resistance, the spring that meets the operating conditions can be selected.
In the formula, Ft is the working load (N), s is the spring compression (mm), k is the spring stiffness (N/mm), D is the spring diameter (mm), C is the winding ratio, n is the effective number of turns, and G is the shear modulus of the material (MPa).
As shown in Figure 2, the depth of the steel needle inserted into the substrate is consistent with the s of the spring's compression. After the spring parameters are selected, the force F required to compress the spring can be calculated according to Formula (4), and the cylinder parameters and models can be selected based on this.

Discrete Element Modeling and Parameter Setting
The seedling gripper needs to be modeled before simulation. The key components for grabbing plug seedlings are four steel needles, so the gripper structure can be simplified, as shown in Figure 3. The 3D model of steel needle, fixed plate, and the tray cell is completed by software of Solidworks and imported into the EDEM software in the *.igs file format. The tray cell is a quadrangular pyramid in shape with the size of upper aperture × lower aperture × depth of (23 × 23) mm × (11 × 11) mm × 42 mm, and there is a circular hole with a radius of 2 mm in the center of the lower bottom surface. The accuracy of the substrate particle model has a strong effect on the accuracy of simulation results. The common growing substrate on the market is a mixture of peat, perlite, and vermiculite in a ratio of 3:1:1. Therefore, the substrate particle modeling is based on this object. To make particle models closer to reality, modeling them by referring to the real geometry features. Studies have shown that the basic structure of peat particles is a block, core, and columnar particles [13,14], while perlite and vermiculite are irregular blocks and flakes [15]. Three particle models were established based on the above features shown in Figure 4. Apart from the particles of different materials, there also have seedling roots in the substrate plug. These fibrous roots have different shapes and complex distributions, and it is difficult to model and simulate them based on discrete element methods. The root system can provide cohesion to the substrate, which can strengthen the substrate block and prevent it from breakage and tearing. Therefore, the cohesion effect of seedling roots on the substrate is replaced by improving particle cohesiveness during modeling. Among the EDEM particle factory parameters, contact surface energy represents the viscosity level of particles, and the larger the value, the stronger the viscosity. The contact elastoplastic ratio represents the elastoplastic level of the particles when they are in contact, and 1 means full plasticity, and 0 means full elasticity. According to the research [16], the contact surface energy of peat particles is set to 20 J/m 2 , and its contact elastic-plastic ratio is 0.6. The contact surface energy of perlite particles is 12 J/m 2 , and the contact elastic-plastic ratio is 0.9. The contact surface energy of vermiculite particles is 8 J/m 2 , and the contact elastic-plastic ratio is 0.8.
After the modeling is completed, set particle model, particle factory, kinematic parameters, and simulation parameters in the EDEM preprocessor module. Because the growing medium is sticky, the elasto-plastic contact model (ECM) is applied to describe the practical contact process of medium particles. This model allows particle overlapping and reflects their elastic and plastic deformations, and it is closer to real particle contact situations [17,18]. The parameters of particles include density, Poisson's ratio, shear modulus, etc., it is measured through methods of the ring knife method, the direct shear test, and the triaxial shear test [19]. In order to save time and cost, in this project, the substrate particle parameters are configured based on Gao Guohua's measurement results [20]. The physical parameters of different particles are included in Table 1. When it generates particles, their size is randomly distributed within the actual range to make it closer to the value in the practical situation [21]. Parameters such as motion trajectory, initial speed, and acceleration of the steel needle are set according to the simulation requirements. Finally, in the EDEM solver module, set the simulation step size to 10-6 s, the data saving time interval to 0.01 s, and the computational domain grid size to 2.5 R. When the solver starts to calculate, it generates substrate particles and fills up the tray cells first. After the particles have fallen and the state has stabilized, steel needles start to insert into the substrate and grab it. After the simulation finished, the results were displayed and analyzed in the post-processing module. Before the simulation, through pre-experiment and literature investigation, it was concluded that the parameters affecting the substrate integrity of the plug seedling during transplanting are as follows: (i) Steel needle diameter Theoretically, when a thick steel needle is inserted into the substrate, the extrusion force generated by the interaction with the soil will be greater than the cohesive force of the substrate block, thereby tearing the soil block. If the steel needle is too fine, it is easy to deform, which is not conducive to grasping a plug seedling. According to the traditional transplanting experience, the diameter of the steel needle ranges from 1 to 3 mm.
(ii) Insertion depth The optimal insertion depth of a steel needle is closely related to tray cell depth. If a steel needle is inserted shallowly, the substrate block at the bottom of the tray cell will break when the plug seedlings are pulled out. On the contrary, if the steel needle touches the cell bottom, it may damage the tray frame and the dense seedling roots at the bottom. According to the tray size used in the study (42 mm in depth), the insertion depth of steel needles ranged from 30 to 40 mm.

(iii) Insertion and Grasping speed
Pre-experimental results show that the insertion and grasping speed of steel needles may have an impact on the substrate integrity. Based on testing, the insertion and grasping speed was set between 0.1 and 1 m/s. The simulation scheme is designed based on the Box-Behnken response surface. As a systematic method for establishing models and process optimization, this method can obtain a uniform sampling point and lower running cost than others under the same number factor [22]. The ratio of the substrate mass successfully grasped by the gripper to the total mass of the substrate in the tray cell can express the grasping effect under different operating parameters, so the substrate integrity rate is a test index. The mass sensor in the EDEM analysis module can measure the mass of the substrate successfully grasped by the gripper and the mass of substrate left in the tray cell, and then calculate the substrate integrity rate W according to the formula (5) [23]. Finally, compile the simulation plan according to the horizontal factor table (Table 2) through Design-Expert 11.0.
In this equation, W is the substrate integrity rate (%), mv is the mass of the substrate successfully grasped by the gripper (g); mc is the mass (g) of substrate left in the tray cell.

Gripper Optimization Design and Test
Based on the principle of maximizing the substrate integrity rate (W), the technical and operating parameters of gripper were optimized. The diameter of the steel needle (X1) set within 1~3 mm, the insertion depth (X2) is within 30~40 mm, the insertion and grasping speed (X3) is within 0.1~1 m/s, and the target substrate integrity rate is within 75-100%. The objective function and constraints display in Equation (6).
A prototype was made to verify the transplanting effect of the optimized gripper. A 3D printer was used to print guide tubes, cylinder blocks, and fixing plates, and assemble them with steel needles and cylinders together. A pneumatic system is designed to drive the gripper, and its working air pressure is tested and set to 0.6 Mpa. The insertion/grasping speed of steel needles can be adjusted by a flow regulating valve ( Figure 5). Pepper seedlings cultivated in a laboratory environment were used as transplanting objects. The seedlings grew for 45 days, and the average plant height is 25 mm. The growing substrate was peat, perlite, and vermiculite evenly mixed in 3:1:1, and the substrate moisture content was 75%. The mass of substrate that was successfully grasped by the gripper was weighed and remained in the tray cell, and then the substrate integrity rate was calculated according to Equation (5).

Simulation Results
The simulation results of the transplanting effect of the seedling gripper based on the discrete element method are included in Table 3. From the simulation results alone, when the needle diameter of the transplanting gripper is 1 mm, the insertion depth into the substrate is 40 mm, and the insertion/grasping speed is 0.55 m/s, the substrate integrity rate is the highest, reaching 88.76%. When the needle diameter is 3 mm, the insertion depth is 30 mm, and the speed is 0.55 m/s, the integrity rate is the lowest, only 38.14%. The damage and fracture of the substrate during the simulation are shown in Figure 6. These results are consistent with the situations in which seedling substrates behaved during the practical transplanting process. That is, in a failed grasping, the substrate block will be broken in the middle, resulting in most of the growing material left in the tray cell, and part of the seedling root is exposed. In the high-integrity grab test, only a small amount of material was left at the bottom of the cell.  The Fit Summary module in Design-Expert Version 11.0 was used to fit the data in Table 3, and finally, perform variance analysis based on the suggested linear model. The results are displayed in Table 4. The p-value of the model is 0.0043 < 0.05, indicating that the model is significant and valid. The p-value of the lack of fit term is 0.4061 > 0.05, indicating that the model has a good fit and a small error. It can be seen from the variance analysis that among several factors affecting the substrate integrity rate, the p-value of the needle insertion depth (X2) is 0.0005 < 0.05, which means this factor has a significant impact on the substrate integrity rate. However, the needle diameter (X1) and insertion/grasping speed (X3) had no positive effect on substrate integrity. The contribution that experimental factors give to the index is determined by the F-value. The larger the Fvalue, the greater the contribution of the factor to the index [24,25]. Therefore, the influence order of each factor on the substrate integrity is the insertion depth > needle diameter > insertion/grasping speed. In order to study whether the needle diameter of the gripper and the insertion and grasping speed have an interaction with the insertion depth (significant factor), the response surface models were built. From Figure 7a, there is an interaction between the insertion and grasping speed and the insertion depth. As speed and depth increase, the integrity rate of the substrate increases too. The insertion angle of needles is consistent with the angular line direction of the tray cell. When the needle inserts deeply, the distance between the four tips of the needles becomes smaller, which is helpful to strengthen the gripping force on the substrate, and more substrates can be grabbed successfully. At the same time, as the steel needle was inserted deeply into the substrate, the impact of moving speed on the medium integrity was significantly weakened, and even when the insertion and grasping speed was 1 m/s, higher medium integrity could be obtained. It also indicated that when the needle inserts deeply, it can not only ensure the substrate integrity but also improve the working efficiency effectively. However, when a steel needle is inserted into the substrate shallowly, its grasping speed cannot be too fast, as the substrate will be broken under the condition of unstable grasping. According to Figure 7b, there is no interaction between the needle diameter and the insertion depth. The change in the needle diameter within 1-3 mm has no significant effect on the substrate integrity, which shows that the well-developed root and higher moisture content can provide enough cohesive force on the substrate to overcome the extrusion force generated by the insertion of the needle.

Parameter Optimization and Test Results
Use Design-Expert Version 11.0 to optimize the simulation results, and the first set of optimized values were chosen. Namely, the needle diameter is 2.99 mm, the insertion depth is 39.97 mm, and the insertion/grasping speed is 0.99 m/s. With these parameters, the maximum substrate integrity was 89.10%. Considering that practical manufacture and control accuracy of gripper prototype during transplanting cannot be accurate to two decimal places, the results are rounded to an integer. So, the needle diameter is 3 mm, its insertion depth is 40 mm, and the insertion/grasping speed is 1 m/s. Experiments were carried out to verify the seedling grasping quality of the optimized gripper. A pepper seedling was randomly selected from the tray for grabbing, and then substrate integrity was calculated after the seedling pulls out. The test repeats nine times and the results are displayed in Table 5. The maximum substrate integrity was 87.34%, and the average value was 76.05%. Compared with the simulation optimization results, it is found that the relative error is between 1.98% and 29.45%. This result is better than similar studies. In the ideal case, the substrate integrity rate obtained by the spade endeffector proposed by Tong Junhua [26] fluctuates between 63.6% and 78.8%, and the average integrity rate is 70.8%. Therefore, the transplanting gripper designed in this paper increases the integrity rate of the substrate by an average of 7.25%. From Jin's study, the maximum value of this figure was 83.45% [27], which is 3.89% less than the maximum value tested in this project. Although the grasping performance of the seedling gripper has reached the expected design goal, there is still a large error in the substrate integrity rate for individual tests compared with optimization results. After cleaning the substrate materials on those seedling roots with low substrate integrity, it was found that the larger error only existed for grasping plug seedlings with weak growth. These seedlings have poor root systems and few fibrous roots. As shown in Figure 8, the plug seedlings with less substrate grabbed even have only one taproot. Comparing the root systems of plug seedlings corresponding to different substrate integrity, it was found that the more developed the root system of plug seedlings, the more substrate mass was grasped by the seedling gripper. The reason is that the developed roots have a large extension space in the tray cell and penetrate different areas inside the substrate, thereby improving the cohesion of the substrate block effectively, so the substrate is not easily broken when grasping. From this point of view, the growth level of the plug seedlings used in this experiment is very uneven, which is the reason for the large error in a few experiments. It also indicates that we should pay attention to the combination of agricultural machinery and agronomy when working on the research and development of plug seedling transplanting machines. Only by adopting scientific cultivation techniques and management methods to ensure perfect consistency in the strong level of plug seedlings, can the transplanter perform well [28]. In addition to substrate culture seedlings, there are also tissue culture seedlings in the greenhouse. Tissue culture refers to the process of regenerating a complete plant through part of the tissue. Agar is often used as a cultivation substrate. This kind of seedling has high requirements for the flexibility of the transplanting claws, and the gripper proposed in the paper has good adaptability to the transplanting requirements. This multi-needle gripper is also suitable for situations where the grasping object is allowed to be intruded, such as the movement of a single piece of material in a food factory.

Conclusions
In this paper, the elasto-plastic contact model is used as the particle contact model of the seedling substrate, and multiple growing substrate particle models with different material properties are built to simulate the real seedling growing environment. The influence of different needle diameters, insertion depths, and insertion/grasping speeds of the proposed gripper on the substrate integrity was studied and optimized. The results show that: The insertion depth of the steel needle has a significant effect on the substrate integrity, and this factor and insertion/grasping speed have an interaction, while the other factors have no significant effect on the substrate integrity. The contribution order of factors on substrate integrity is insertion depth > needle diameter > insertion/grasping speed. Based on the simulation results, the optimized parameters of the seedling gripper are: the diameter of the steel needle is 3 mm, the insertion depth is 40 mm, and the insertion/grasping speed is 1 m/s. Under these parameters, the integrity rate of the substrate is 89.10%.
For a practical grasping test, the maximum integrity rate of plug seedling substrate was 87.34%, and the average was 76.05%, which increased by an average of 7.25% compared with existing studies. The relative error range between test results and the optimized results was 1.98-29.45%. The reason for the large error in some experiments is that the strong level of plug seedlings cultivated indoors is not even. Therefore, the mechanized transplanting of plug seedlings should focus on the combination of agricultural machinery and agronomy. This gripper can also be used in tissue culture seedling transplanting and other situations where the grabbing object is allowed to be invaded.
There is a small difference between the simulated data based on the discrete element method and the practical experimental data,which implies that the method is feasible to solve the interaction problem between steel needle's and growing substrate, for it can complete the analysis and demonstration of the problem quickly and effectively.  Acknowledgments: The authors thank the Key Laboratory of Intelligent Horticultural Equipment, the Ministry of Agriculture and Rural Affairs for supporting the experimental conditions of the research. We gratefully acknowledge the experts for editing our paper, as well as the anonymous reviewers for their critical comments and suggestions to improve this article.

Conflicts of Interest:
We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.