Analysis of Mechanical Characteristics of Stereolithography Soft-Picking Manipulator and Its Application in Grasping Fruits and Vegetables

Abstract

Aiming at the issues of complex manufacturing processes and unstable bonding after individual manufacturing in current soft manipulator forming methods, this study investigated the mechanical characteristics of a pneumatically driven soft-picking manipulator formed by the stereolithography (SLA) process and evaluated its application in grasping fruits and vegetables. The soft-picking manipulator mainly consists of three soft actuators designed in an integrated folded structure to simplify the manufacturing process compared to a conventional one. A finite element model (FEM) of the actuator was created to analyze the bending deformation capability under different pressures, and the simulated results match well with the experimental ones. Under the 60 kPa pressure, the maximum grasping force for fingertip- and envelope-grasping is 3.94 N and 8.87 N, respectively. The grasping tests of several fruits and vegetables of different weights and sizes by the soft manipulator were examined, and the results showed that the manipulator has strong adaptability. For spherical and elongated fruits and vegetables, the completion time for grasping is approximately 8.59 ± 1.26 s and 10.99 ± 1.79 s, respectively, and for irregularly shaped ones, the pressure is increased accordingly to the increased grasping stability. This study may provide a basis for the development of a soft manipulator for sorting and picking fruits and vegetables.

1. Introduction

Picking fruits and vegetables is the key link in agricultural production. At present, a large number of agricultural picking robots have realized intelligence, which has greatly improved the efficiency of picking. However, most of the end-effectors use traditional rigid manipulators, which have a complex structure, high rigidity, and low interaction with the environment. It is very easy to damage the grasping target during the picking process. To make the end effector always maintain the output of constant force, Ding et al. [1,2] designed a constant force mechanism based on the combination of positive and negative stiffness to regulate the output range of constant force without using additional sensors and control algorithms. In recent years, scholars at home and abroad have paid great attention to soft manipulators made of flexible materials with high flexibility and safety in human–computer interaction [3,4]. Compared to traditional hard hands, soft and underactuated manipulators have several advantages [5,6,7]. Their own continuous deformation allows them to easily manipulate a broad range of objects and to interact with uncertain environments in a more forgiving manner, especially in the areas of perishable and irregular fruit and vegetable gripping and complex operations in unstructured environments, which have great application potential.

At present, soft actuators can be actuated in many ways, such as tendon actuation using shape-memory alloys [8], electrical actuation using electroactive polymers [9], or dielectric elastomers [10] and fluid actuation via liquids or gases [11]. Among them, the pneumatic soft actuator is characterized by light weight, high efficiency, and non-pollution, which makes it occupy an important position in the field of soft robotics [12,13]. However, limited by the manufacturing technology, pneumatic soft actuators face many difficulties and challenges: complex flexible structures and flexible materials have higher requirements for the manufacturing process, and how to quickly and efficiently process the body structure to meet the specific needs of the problem has become an urgent need to solve.

To solve the molding problem of complex soft structures, numerous scholars have proposed many manufacturing technologies. Among them, the molding technology (Figure 1a) is currently the most widely used. This technology divides the complex mold structure into a large number of simple molds based on the traditional casting method. This reduces mold design and production difficulties [14,15]. However, the complex, multi-step molding and casting manufacturing approach significantly limits the geometric complexity and functionality of the created soft manipulators. In addition, the molding technique also requires an additional bonding process, and problems such as easy cracking at the bonded area may occur [16]. More importantly, the manufacturing scale and the type of structure are also limited by the mold. It will also greatly increase the difficulty of molding when manufacturing soft actuators with complex internal structure characteristics.

In addition, with the development of 3D printing technology and materials, there are many 3D printing processes (such as inkjet printing [17], selective laser sintering (SLS) [18], and fused deposition modeling (FDM) [19], etc.) that can process flexible materials. The soft manipulator structure is fabricated directly, which simplifies the manufacturing process (as shown in Figure 1b). More recently, Peele et al. [20] have created a new type of soft pneu-actuator using SLA. Unlike the usual type of actuators, these types of actuators can be easily reproduced and modified due to the direct fabrication via 3D printing. The usual steps of using specially designed molds and multiple fabrications steps are greatly reduced, saving time and effort. Moreover, Valentine et al. [21] used direct-writing 3D printing technology to print the polyurethane matrix and conductive ink at the same time, and placed tactile sensors into it to create soft actuators with sensing functions. Manuel et al. [22] modified the silicone material to have photosensitive properties and used ultraviolet light to illuminate the material while printing based on DIW technology, so that the material can cure in time and thus create a soft driver that can bend, stretch, and twist. Hong et al. [23] used FDM technology and finite element method to simulate the printing process of the soft actuator, which successfully developed a dual-channel soft actuator, and applied it to wearable devices. Mutlu et al. [24] used TPU wire as manufacturing material and made a bionic gripper hand based on FDM technology. The soft actuator fabricated via the above 3D printing method has good mechanical properties. However, each manufacturing method has its own limitations. FDM requires support structure and is limited by the wire feeding mechanism and thermoplastic materials, and it can only print materials with a Shore hardness above 80 A, so it can only be driven under high pressure [25].

On the other hand, stereolithography (SLA) is currently a relatively mature 3D printing technology, which is widely used in hydrogel printing, flexible capacitors, and medical pre-operation simulation. The materials used in SLA generally consist of light-sensitive polymer precursors, active diluents, cross-linking agents, photoinitiators, and other auxiliary materials [26]. During the printing process, SLA utilizes UV light as the energy source, focuses the laser beam via the scanning galvanoscope, directs it to the surface of the print substrate, and selectively hardens the liquid photo-setting resin layer by layer, where each layer is fully hardened in a few seconds [27]. It has the advantages of high speed, high efficiency, and a wide range of material adaptability [28,29]. However, SLA also faces some problems, e.g., in the process of structural design and manufacturing of the soft actuator, the model structure, placement position, and direction will have a decisive impact on the molding quality of the model. For example, the support structure (as shown in Figure 2a) is required to ensure the complete printing and molding of the model when the model contains a large angle hanging or wide bridge structure (as shown in Figure 2b). Since the support structure is an inseparable part of the overall printed model, a series of post-processing operations are required to remove the support structure after printing is completed and to process the traces left by the support structure. Therefore, we need to combine the characteristics of SLA molding to reduce unnecessary support structures and post-processing to ensure the molding quality of the model in the design and manufacture of soft actuators. In addition, it is also necessary to consider reasonable structural parameters to reduce unnecessary stress concentration and radial expansion, making its bending deformation movement accurate and increasing its service life and performance. At present, most research results only focus on material properties and the SLA molding process [30,31,32], such as rheological behavior, tensile elongation, tunable mechanical behavior, and multi-material printing performance, etc., but there are a few research results on its bending and gripping properties.

Therefore, according to the characteristics of the SLA molding process, this study designs a soft actuator with an integrated folding structure without support during printing and investigates its mechanical characteristics such as the bending deformation ability and output force changes under the sliding state. On this basis, the soft-picking manipulator was developed. A test platform for picking fruit and vegetables was constructed using a kit containing an electronic control circuit and a pneumatic circuit. Finally, according to different sizes and shapes of fruits and vegetables, grasping experiments were conducted to test the grasping success rate, grasping completion time, and driving pressure, etc., to verify the grasping performance of the soft manipulator. For the agricultural picking field, it provides a low-cost, high-precision, and high-surface-quality soft-picking manipulator.

2. Materials and Methods

2.1. SLA Printing System Setup

The experiment was conducted using a UV 270 light-curing 3D printer (Beijing 3D Bot Technology Co., Ltd., Beijing, China). The molding size was 115 mm × 65 mm × 160 mm, the layer thickness was 0.025–0.1 mm, and the XY resolution was 2560 × 1440 pixels. The 3D model was horizontally sliced via a computer-aided design. The slices were then converted into 2D mask images, and the mask images were projected onto the bottom of a resin tank using a light projection device to cure the photopolymer resin. After curing one layer, the platform was first raised to separate the cured layer from the tank bottom and then lowered to create a gap with the tank to cure the next layer. The 3D solid mold was created by selectively curing successive layers of resin, which were repeatedly fused to themselves and to the previous layer under UV light. Figure 3 shows the SLA equipment and the schematic diagram.

2.2. Structure Design of Soft-Picking Manipulator

In order to test the viability of the SLA process for the direct manufacturing of a soft-picking manipulator, we designed a soft-picking manipulator using SoildWorks software (version 2021). The designed soft manipulator is combined with the action of grasping objects with the hand. When grasping objects, the three fingers of the hand can pick up objects with little force. If two fingers are used, the grasping force exerted is large, causing a large deformation of the surface of the object and damage to the fruit, and the grasping stability is poor. If the number of fingers increases, it can improve the grasping stability, but it also increases the difficulty of control and the manufacturing cost. Therefore, three soft actuators are selected and circularly distributed on the flange in this study. The specific structure and main structure parameters are shown in Figure 4. Each soft actuator is mainly composed of strain layer, channel, and the bottom three parts, and each adopts the integrated folding structure design to allow large overall deformation without high local strain. Among them, the strain layer adopts a multi-chamber structure. When the air pressure is increased, the air flows through the channel and fills the chamber, causing the soft actuator to bend and deform.

2.3. Preparation of Soft Actuators

In this study, using silicone rubber photosensitive resin (Resione F80) as the base material, the soft actuators are printed and molded via the SLA method. The specific preparation process is as follows: First, the 3D model of the designed soft actuator is imported, sliced, and edited, and the corresponding molding process parameters are set: the layer thickness is 0.035 mm, the number of layers is eight, the exposure time of the bottom layer is 30 s, and the exposure time of each other layer is 10 s. Then, the sliced file of the soft actuator is exported, transferred to the SLA 3D printer, and the F80 silicone rubber resin is poured into the material tank to start printing. Finally, the printed soft actuator is cleaned with alcohol and placed in the curing oven for secondary curing. After curing, the solid model of the soft actuator can be obtained.

2.4. Finite Element Analysis of the Soft Actuator

In order to analyze the influence of the structure on the bending performance of the soft actuator, the finite element simulation analysis is carried out using ABAQUS software (version 6.14-4) to predict the bending deformation motion of the soft actuator. Considering that the silicone rubber resin used in this study has high elastic nonlinear mechanical properties, the Yeoh model is adopted to describe the nonlinear mechanical behavior of the soft actuator during deformation. The strain energy density function U can be expressed as [33]:

$$U={\displaystyle \sum }_{i=1}^3{C}_{i0}{\left(I_1-3\right)}^i+{\displaystyle \sum }_{i=1}^3\frac{1}{D_i}{\left(J-1\right)}^{2i}$$

(1)

where C_i0 and D_i are the indeterminate coefficients; I₁ is the deformation tensor; and J is the elastic volume ratio, according to the incompressibility of silicone rubber resin, J = 1.

Usually, the strain energy function U in the form of binomial parameters is used, then the binomial parameter form of the Yeoh model is adopted:

$$U=C_{10}\left(I_1-3\right)+C_{20}{\left(I_1-3\right)}^2$$

(2)

where $I_1={\lambda }_1^2+{\lambda }_2^2+{\lambda }_3^2$, λi is the principal strain ratio in all directions and is not dimensioned. By using the uniaxial tensile test on the material, the stress is σ₂ = σ₃ = 0, then

$${\lambda }_2^2={\lambda }_{3 }^2=\frac{1}{{\lambda }_1}$$

(3)

By combining Equations (2) and (3), the relationship between stress σ₁ and the principal strain ratio λ₁ can be obtained [33]:

$${\sigma }_1=2\left({\lambda }_1-\frac{1}{{\lambda }_1^2}\right)\left[C_{10}+6C_{20}+2C_{20}\left({\lambda }_1^2+\frac{2}{{\lambda }_1}\right)\right]$$

(4)

The tensile test is based on standard ASTM D412. The specimens were printed using an SLA printer. Figure 5a shows the printed specimens and print direction, and they are tested via the tensile test, where the speed of the tensile test is 10 mm/min. The fitting curve of the stress and strain was obtained via the tensile test as shown in Figure 5b. The constitutive equation of silicone rubber resin is obtained by using the Yeoh model. The stress–strain data of the tensile test are evaluated. The Yeoh model parameters for the silicone rubber resin are C₁₀ = 0.98 and C₂₀ = 0.37.

In the soft actuator simulation process, the fluid cavity model is used to simulate inflation, and the air pressure input plane is completely fixed and constrained. As the deformation of the soft actuator is a large deformation, it is set to geometric non-linearity in the analysis step. In general, a model that can be subdivided by triangles is not necessarily subdivided by quadrangles. For example, the three corners of a triangle cannot be subdivided into quadrangles. Therefore, for complex geometric structures or models with more subtle features, triangular meshes can be quickly subdivided and accurately described. When setting the mesh cell type, select triangular mesh and set the geometric order to the second order. Since the silicone rubber resin is incompressible, the hybrid cell type is selected. The mesh partition of the soft actuator is shown in Figure 6a.

In simulating and analyzing the soft actuator using ABAQUS software (version 6.14-4), the bending angle θ is taken as the main performance index. The specific measurement method is shown in Figure 6b: The bending angle θ of the soft actuator is the angle between the line from the bottom of the soft actuator to the top and the vertical line.

In this paper, finite element simulation using Abaqus software (version 6.14-4) is used to simulate the bending performance of the soft actuator in order to predict the feasibility of the designed soft actuator structure. We varied the applied pressure from ΔP = 0 to 60 kPa and simulated the bending performance of the actuator, as shown in Figure 7a–d, and the bending angle of the soft actuator gradually increases as the driving pressure increases.

In order to verify the correctness of the simulation results of the bending performance of the soft actuator, we transferred the slicing file to the SLA 3D printer according to the designed soft actuator structure for printing and forming. The soft actuator after printing is shown in Figure 8a, and the printed soft actuator has a good surface quality. Compared with traditional molding methods, the manufacturing process is greatly simplified due to the overall printing of the software driver and the avoidance of the bonding process. However, when the air pressure is not passed through, the soft actuator has a certain initial bending angle (about 2.3°). This may be due to the fact that silicone rubber resin has a certain shrinkage, after the light-curing molding, the internal structure of the soft actuator is hollow and the bottom layer is thicker; the thicker the layer, the greater the shrinkage, and the actuator will bend to the bottom; thus, it has a certain initial bending angle.

We then fixed the printed soft actuator on the test platform and performed a series of bending tests to evaluate the bending deformation ability of the soft actuator. Figure 7e–h shows the deformation of the soft actuator at 15 kPa, 30 kPa, 45 kPa, and 60 kPa pressure. In addition, the deformed profile of the bottom line of the actuator at different pressures was compared between the experimental and FEM results. The deformed profile of this line is represented by a series of equally spaced points that are marked on the deformed curve. As shown in Figure 7i–l, the experimental tests were performed three times and the average values were used for comparison with the FEM results. It indicates that the experimental results match the FEM results. Moreover, the printed pneumatic actuator can produce enough deformation under relatively low air pressure to be suitable for the soft-picking manipulator actuator.

We compared the FEM results of the soft actuator with the pressure of 0~60 kPa with the actual bending test results, as shown in Figure 8b. When the pressure was 0~5 kPa, the soft actuator had a certain initial bending angle due to the SLA process and the structural parameters, resulting in a large error between the FEM results and the test results. When the pressure is 10~60 kPa, the FEM results fit well with the test results, and the maximum error is less than 8.88%. Therefore, we believe that the bending characteristics of the soft actuator can be accurately reflected via the established finite element model.

2.5. Soft Actuator Output Force Test

The output force of the soft actuator directly affects the fruit and veg holding capacity, and in order to characterize the output force of the soft actuator, the test device was built as shown in Figure 9a. The small ball in the figure represents the object to be grasped, with a pressure sensor embedded inside. The ball is moved outward in the direction of the actuator at a speed of 5 mm/s. The pressure sensor is used to measure the change in output force of the soft actuator as the ball slips. The ball is made of polylactic acid (PLA) as a matrix material and is produced on an FDM 3D printer with a diameter of ø75 mm.

According to the above test steps, the output force of the soft actuator in the sliding state was obtained under the pressure of 0~60 kPa, and the test results are shown in Figure 9b. The output force of the soft actuator increases as the driving pressure increases. When the pressure reaches 60 kPa, the output force of the soft actuator can reach 1.9 N. In addition, under the same driving pressure, as the target object slides from the root of the actuator to the tip of the finger, the output force of the soft actuator first increases and then decreases, and the output force is the largest at the position of the last chamber. Since there is no air chamber at the tip of the finger, there is no air pressure input, and the tip of the finger is a soft plate that cannot provide a large output force to the target object; as the object slides to the end of the actuator until it is released from the manipulator, the output force gradually decreases until it reaches zero.

2.6. Test Platform Design

The test platform of the soft-picking manipulator consists mainly of the pneumatic circuit, the electronic control circuit, and the mechanical arm. The pneumatic circuit is connected to the soft-picking manipulator through gas pipes and joints and includes an air pump, pressure regulator, solenoid valve, vacuum generator, etc. The electronic control circuit consists mainly of the main computer, relays, and the DAC module. The control system diagram of the soft-picking manipulator is shown in Figure 10. Among them, the components used in the pneumatic circuit are connected via a pneumatic connector, and the pneumatic pressure gauge data is returned to the programmable logic controller (PLC) and the PC via the signal line, where the electric control circuit is controlled via the PC and PLC solenoid valve for reversing, and then the positive and negative pressure of the driving air pressure is controlled to achieve the grasping and unfolding actions of the soft manipulator.

When controlling the movement of the soft-picking manipulator, the air pump provides the air source for it and adjusts the input pressure via the precision pressure regulator to realize the bending angle control under different pressure. The controller provides negative pressure to the manipulator for the manipulator opening operation when the solenoid valve is combined with the vacuum generator. When the air pressure is directly output from the pressure regulator, the positive pressure is supplied to the manipulator, which is used for the gripping action of the manipulator. By combining the electrical and pneumatic circuits, the soft-picking manipulator can complete the process of gripping the object.

2.7. The Soft Manipulator Grasping Mode

For the soft manipulator designed in this study, there are two grasping methods: the fingertip grasping method and the envelope grasping method, as shown in Figure 11a,b. Fingertip grasping is usually used to grasp small objects through the end of the soft manipulator contact with the target where the contact area is small. Envelope grasping is usually used to grasp larger objects through the inner surface of the soft manipulator, where the whole object is wrapped and the contact area is large. In addition, due to the friction generated by the contact and the bending moment generated by the bending of the envelope, it provides a larger gripping force and makes the gripping process more stable.

In order to test the grasping force of the above two grasping methods, we installed the soft manipulator on the platform and used the dynamometer to test the grasping force under different driving pressures. The grasping test procedure is shown in Figure 11c. In the grasping test, we think that the sum of the maximum indicator number of the dynamometer and the weight of the target object when the target object falls from the soft manipulator is the final grasping force. The target object materials used in the test are all made of polylactic acid and manufactured via the fused deposition modeling (FDM) 3D printer. Among them, the diameter of the object used in the fingertip grasping test is 50 mm and the weight is 23.63 g. The diameter of the object used in the envelope grasping test is 75 mm and the weight is 70.92 g.

3. Results

3.1. Grasping Experiments of Different Grasping Modes

The grasping force results under different grasping methods are shown in Figure 12; as the driving pressure increases, the gripping force of the soft manipulator also increases. When the pressure increases to 60 kPa, the grasping force of fingertip grasping and envelope grasping reaches the maximum, reaching 3.94 N and 8.87 N, respectively. The grasping test results show that the grasping force of envelope grasping is obviously greater than the grasping force of fingertip grasping under the same driving pressure. This is because the contact area between the manipulator end and the target object is small, and the end grasping force is mainly used to overcome the target object’s gravity. In the envelope grasp, the contact area between the soft manipulator and the target object is large, and the friction generated by the contact, coupled with the clamping force generated by the bending of the soft manipulator, makes it have a greater grasping force and further increases the grasping stability.

3.2. Fruit and Vegetable Grasping Experiments

To further verify the feasibility of the soft-picking manipulator for picking fruits and vegetables, this study selected seven fruits and vegetables of different sizes and weights (cherry tomato, orange, tomato, pear, apple, banana, and cucumber) for the grasping experiments. The weight, average diameter and height of the grasping target, success rate, and the air pressure required for grasping are shown in

Table 1. Grasping object characteristics and required air pressure.

Fruit and Vegetable Categories	Weight/g	Height/mm	Mean Diameter/mm	Pressure/kPa	Success Rate/%
Cherry tomato	41.23	50.65	38.97	22	85
Orange	70.86	44.74	58.59	20	100
Tomato	214.60	66.71	75.54	26	100
Pear	283.68	73.66	79.51	36	100
Apple	253.49	75.53	84.33	32	100
Banana	143.79	161.25	37.07	30	80
Cucumber	188.47	151.53	44.39	38	75

. If the fruit did not fall during the grasping process, the grasping was considered successful. The driving pressure was recorded and the grasping success rate was calculated.

As shown in Table 1, we can find that for cherry tomatoes and oranges with small sizes, fingertip grasping is usually adopted. However, because the size of cherry tomatoes is smaller than that of oranges, a larger air pressure is required to drive the soft manipulator to produce a larger bending angle. Therefore, the driving air pressure required for cherry tomatoes is higher than that for oranges. In addition, because the shape of cherry tomato is an irregular heart-shaped structure, it will fall during the process of grasping and transporting, and its grasping success rate is only 85%. For the orange of the same small size, because its shape is similar to the regular sphere, it does not fall down during the test, and the success rate is 100%.

For larger, more closely sized tomatoes, pears, and apples, the driving pressure required to pick them up generally increases with the weight of the target object. Due to the large contact area between these three objects and the inner surface of the soft manipulator, the friction is greater, grasping is more stable, the object does not fall, and the success rate can reach 100%.

Before performing the fruit and vegetable grasping test with the soft-picking manipulator, the first step is to build a mobile test platform (as shown in Figure 13a). The platform mainly consists of a mobile chassis, a robotic arm, a camera for visual recognition, and a soft manipulator. The mobile chassis adopts McNamum wheels, and the four wheels are independently driven. It can realize the omnidirectional combined motion mode such as forward, cross, inclined, and in-situ rotation. Visual recognition cameras are mainly used to identify the fruit and vegetable categories. The soft-picking manipulator consists of three SLA-printed soft actuators equidistant mounted on a flange to form a soft picking manipulator. The pneumatic circuit and electronic control circuit are connected and fixed on the pre-assembled mechanical arm and mobile chassis. Once the mobile test platform has been built, the fruit and vegetable gripping test can be carried out.

As shown in Figure 13b, taking the soft manipulator to grasp an orange as an example, the grasping process is mainly divided into the following steps: Firstly, the pneumatic circuit is controlled to input negative pressure to make the soft manipulator open and move directly above the target object; secondly, the target object is approached and the positive pressure is controlled to make the soft manipulator grip and clamp the fruit or vegetable; finally, the mechanical arm is lifted, the mobile chassis and the soft manipulator are driven to move to the target position, and the soft manipulator opens to place the target object in the specified position to complete the grasping process.

Due to the small size of cherry tomatoes and oranges, they can only be grasped by the fingertips of the soft manipulator (as shown in Figure 13c); when the pressure is increased to 22 kPa and 20 kPa, they can be picked up. At that time, the driving air pressure was the pressure at which the soft manipulator just grasped the fruits and vegetables without falling. Since the grasping is performed by the fingertips of the soft manipulator, it has the disadvantage of insufficient grasping stability, which easily causes the target to fall when the manipulator or test platform moves. To improve the grasping stability of the fingertips of the soft mechanical machine, the driving pressure when grasping cherry tomatoes and oranges is increased to 29 kPa and 27 kPa, respectively. At this time, the success rate of the target during the grasping process will also increase, and there will be no damage to the surface.

For larger tomatoes, pears, and apples, the contact area with the surface of the soft manipulator is large, so the envelope grasp can be used (as shown in Figure 13d). When the driving pressure is increased to 26 kPa, the tomato can be picked up from the table. Pears and apples require 36 kPa and 32 kPa of drive pressure, respectively, to pick them up. Moreover, the envelope grasp makes the soft manipulator fit well with the surface of the target object, and the friction generated by the contact surface further improves the stability of the grasp.

In addition to the above spherical fruits and vegetables, we also conducted grasping tests for two kinds of long fruits and vegetables (bananas and cucumbers), as shown in Figure 13e. From the data in Table 1, we can see that the pressure required to grasp bananas and cucumbers is relatively large, which can reach 30 kPa and 38 kPa, respectively. This is because their diameter is relatively small, their weight is still relatively high, and in order to adapt to the long fruit and vegetable surface, two soft actuators in the manipulator simultaneously bend but also distort, so the pressure required to grasp the banana and the cucumber is larger than that of the tomato with similar weight, and the grasping success rate is relatively low at 80% and 75% respectively.

We also recorded the grasping time of each fruit and vegetable. According to the different grasping stages, the grasping completion time was divided into three stages: in the first stage, the hand claw opened and the mechanical arm moved, which took an average of about 2.41 ± 0.34 s; in the second stage, the hand claw is closed and the mechanical arm is raised. In this stage, it takes about 2.7 ± 0.51 s for spherical fruits and vegetables, while for long fruits and vegetables, the manipulator has to be rotated to an appropriate angle in the grasping stage, so it takes about 5.1 ± 1.04 s. In the third stage, the robot arm moves, the hand claw opens, and the fruit falls into the target area, which takes about 3.48 s on average. Therefore, for spherical fruits and vegetables, the total grasping completion time must be about 8.59 ± 1.26 s, while for long fruits and vegetables, the total grasping completion time is about 10.99 ± 1.79 s.

Overall, the fruit and vegetable grasping test results show that the SLA printing soft manipulator is flexible, has strong adaptability to grasp objects, and is easy to control. It can realize the grasping of fruits and vegetables of different weights and sizes by appropriately adjusting the driving pressure.

4. Discussion

We compared the main performance parameters and characteristics of pneumatic soft manipulators designed via this work with other methods reported in recent years, as shown in

Table 2. The main performance parameters and characteristics of pneumatic soft manipulators.

Manufacturing Method	Output Force	Materials	Manufacturing Process	Manufacturing Costs	Characteristics	Ref. Number
Fused Filament Fabrication (FFF)	4.4 N/100 kPa	Thermoplastic elastomer (TPE)	Simple	Low cost	Rough surfaces; High temperature process; High pressure;	[19]
FDM	17.79 N/300 kPa	TPE	Simple	Low cost	Easy to deform during molding; High pressure;	[23]
DLP	–	Hexane-1,6-diol diacrylate/HEA mixture	Medium complexity	Expensive equipment	Excellent surface quality; Increased performance with multiple materials;	[34]
3D printing	2.45 N/160 kPa	ShOreA30	Simple	–	High pressure; Large bending angle;	[35]
SLS	2.9 N/400 kPa	TPU92A-1	Simple	Expensive equipment and materials	High pressure; Long manufacturing time;	[36]
Molding	–	Dragonskin 10 A/20 A	Complexity	Low cost	More material available; Long manufacturing time; Complex manufacturing process;	[6]
Molding	–	Ecoflex 00-10	Complexity	Low cost	Long manufacturing time; Complex manufacturing process;	[37]
SLA	1.9 N/60 kPa	Silicone rubber resin	Simple	Low cost	Excellent surface quality and precision; Easy to age in the sun.	This paper

. From Table 2, due to the fact that the developed SLA soft manipulator is mainly used for grasping fruits and vegetables, it is necessary to consider its service life while meeting the grasping conditions of the manipulator. Therefore, we did not consider excessive air pressure (0–60 kPa) in our developed manipulator.

However, if the air pressure is increased, the SLA soft manipulator can also achieve a similar output force. Of course, we can also improve the output force of the SLA soft manipulator via the optimization of the SLA process parameters or the optimization of the soft actuator structure. In addition to the output force, compared with other manufacturing methods, the molding process is simple, the price of equipment and materials is inexpensive, the molding accuracy and surface quality are relatively high, and it is especially suitable for industries such as agricultural production that require low cost. Nevertheless, the current study of the SLA soft manipulator did not consider the aging effect of light on silicone rubber resin. Therefore, we found that if the SLA soft manipulator is exposed to the sun for a long time, light aging will occur and even cracks will appear, resulting in a shorter service life of the soft manipulator. Therefore, the SLA soft manipulator should be kept away from light during non-working hours.

In addition, for elongated or irregular objects, the grasping success rate of the SLA soft manipulator designed in this paper is relatively low, only about 80%, and the grasping stability is also insufficient and the adaptability is limited. Therefore, in the future work, we will carry out research on slender or irregular small-size objects and design a special soft manipulator. In addition, the flexible force sensor developed in the previous work [38] will be integrated into the current SLA soft manipulator, so that the soft manipulator can sense and feedback the real-time change information of the grasping force, so as to improve the stability and success rate of grasping fruits and vegetables and expand the application of the SLA soft manipulator in the fields of agriculture, medical rehabilitation, wearable devices, and soft robots.

5. Conclusions

In this study, according to the characteristics of SLA molding technology, we designed and manufactured a kind of soft manipulator for picking fruits and vegetables, which can be formed in one piece without a support structure. The soft manipulator designed can reduce the manufacturing cost, improve the surface quality of molded parts, and eliminate the post-processing process. Through the established FEM of the soft actuator, the bending deformation ability of the soft actuator was evaluated via numerical simulation and experimental verification, and the output force of the soft actuator in the slip state was tested. When the air pressure reached 60 kPa, the maximum output force of the soft actuator was 1.9 N. The grasping force of fingertip- and envelope-grasping can reach 3.94 N and 8.87 N, respectively. We built a set with a pneumatic circuit and an electronic control circuit that can drive the soft-picking manipulator and built a mobile test platform for picking fruits and vegetables to test its ability. This study may provide a basis for the development of a soft manipulator for sorting and picking fruit and vegetables.