An Electric Gripper for Picking Brown Mushrooms with Flexible Force and In Situ Measurement

Abstract

As brown mushrooms are both delicious and beneficial to health, the global production and consumption of brown mushrooms have increased significantly in recent years. Currently, to ensure the quality of brown mushrooms, selective manual picking is required, and the delicate surface of the mushrooms must not be damaged during the picking process. The labor cost of picking accounts for 50–80% of the total labor cost in the entire production process, and the high-humidity, low-temperature plant environment poses a risk of rheumatism for the laborers. In this paper, we propose a novel underactuated gripper based on a lead screw and linear bearings, capable of operating with flexible force control while simultaneously measuring the diameter of the mushrooms. The gripper features three degrees of freedom: lifting, grasping, and rotation, and enabling it to approach, grasp, and detach the mushroom. A thin-film force sensor is installed on the inner side of the fingers to achieve accurate grip force measurement. The use of a PID algorithm ensures precise grip force control, thereby protecting the brown mushrooms from damage. Experimental results demonstrate that the proposed gripper has a static grasping force error of 0.195 N and an average detachment force overshoot of 1.31 N during the entire picking process. The in situ measurement of the mushroom diameter achieves 97.3% accuracy, with a success rate of 98.3%. These results indicate that the gripper achieves a high success rate in harvesting, a low damage rate, and accurate diameter measurement.

1. Introduction

The brown Agaricus bisporus mushroom, commonly referred to as the brown mushroom, is a distinct variant of the Agaricus bisporus portobello mushroom, characterized by its large brown cap. Due to its tender and juicy texture, reminiscent of a thick and succulent steak, it is colloquially known as the “mushroom steak” globally [1]. Brown mushrooms are rich in bioactive compounds, including fibers, polysaccharides (such as $\beta$-glucans), selenium, and vitamins, which play a crucial role in preventing chronic diseases and promoting healthy aging [2,3].

China is the world’s largest producer of Agaricus bisporus portobello mushrooms, contributing 54% of global production, amounting to 2.37 billion kilograms [4]. According to a statistical survey conducted by the China Edible Fungi Association, the total output of edible fungi in China in 2020 was 40.61 million tons (fresh products), with a total output value reaching 346.565 billion CNY, reflecting an increase of approximately 10.84% compared to 312.667 billion CNY in 2019 [5]. In the United States, mushrooms hold significant economic importance, particularly in Pennsylvania, which produced nearly two-thirds of the nation’s mushrooms, totaling 572 million pounds in 2018 [6,7].

In modern Agaricus bisporus factory plants, mushrooms are grown on multilayer standard shelves and continuously cultivated day and night. Brown mushrooms, in particular, are classified by the size of their caps and require selective picking around the clock [8]. Traditional manual picking methods necessitate workers climbing the multilayer shelves [9,10,11] and bending their upper bodies into the interlayers to selectively pick brown mushrooms that meet the required size criteria [12]. This process demands great care to avoid leaving scratches where the fingers touch the mushrooms, making it both time-consuming and labor-intensive, with labor costs accounting for 50–80% of the total production costs. Moreover, the high-humidity and low-temperature environment maintained to promote mushroom growth poses a risk of rheumatism for the laborers [13]. This discourages younger workers from participating in the industry, potentially hindering its development. With the gradual decline of the demographic dividend and increasing labor costs, the push for high automation in agriculture is intensifying, with robots poised to replace manual labor in the future [14,15,16,17,18].

Early mushroom harvesting machines used blades to sweep across mushroom beds, indiscriminately harvesting mushrooms [19]. However, this method is unsuitable for brown mushrooms, which require selective picking without damaging the caps, as any damage devalues them. Currently, the most widespread method for picking mushrooms involves using a suction device. A mushroom-picking robot with a dual suction cup end-effector was developed in 2001, capable of automatically locating, grasping, transporting, and trimming mushrooms, achieving an 80% success rate in commercial farms [20,21]. Yang et al. designed a novel mushroom-picking end-effector based on vacuum negative pressure, which used a suction cup to pick white mushrooms, achieving an 88.2% success rate and a 2.9% damage rate in a factory environment [22]. However, suction cups have limitations. Other studies have shown that suction devices generate pressure, slippage, and shear forces during the adsorption and twisting of mushrooms, leading to surface discoloration due to bruising [23,24]. Noticeable white rings appear on the mushroom caps three days after picking [25], reducing their economic value.

To achieve nondestructive picking, it is essential to analyze the forces involved in grasping [26,27,28]. Some researchers have studied the forces exerted by the thumb, index finger, and middle finger when manually picking Agaricus bisporus mushrooms with an average diameter of about 4 cm, as well as the mushroom’s rotation angle and picking time. They concluded that traditional picking methods (twisting, bending, and lifting) and simple bending are more effective [29]. Other researchers have conducted mechanical property analyses for shiitake mushroom picking, including stipe compressive stress tests and stress relaxation tests, and have initiated finite element simulations [30,31].

Our lab has conducted extensive work on brown mushroom picking. As early as 2018, Lu et al. [32] proposed an in situ measurement method for brown mushrooms using the SR300 depth camera Additionally, Lu et al. [33] developed an advanced integrated method for in situ recognition, measurement, and positioning of brown mushrooms based on YOLO v5 transfer learning combined with three-dimensional edge information for dynamic diameter estimation. This method achieves an average processing time of 50 ms per frame, a 100% recognition success rate under various lighting conditions, and an average diameter measurement accuracy of 97.28%. Ji et al. [34] also designed and fabricated a pneumatic soft gripper capable of adaptive grasping and in situ measurement, with a measurement accuracy greater than 95%.

Nondestructive picking is always a key issue in research on fruit and vegetable harvesting, requiring flexible force control. Pneumatic soft grippers [35,36,37] have been developed for their natural softness. Force sensors and flexible force control methods are applied to rigid grippers, enabling them to output flexible force [38]. Particularly, grippers often adopt underactuated structures, capable of being opened and closed by controlling a single motor [39]. With force sensors and control methods, underactuated grippers can perform delicate operations [40]. Xu et al. [41] designed an underactuated broccoli-picking manipulator, achieving an 84% success rate in picking and a 100% lossless rate. Xiong et al. [42] proposed a cable-driven gripper capable of picking isolated strawberries with a near-perfect success rate (96.8%).

Although pneumatic flexible grippers can offer more degrees of freedom and safer picking characteristics, the low-temperature and high-humidity environment in mushroom houses causes silicone rubber to age more quickly, leading to air leaks in pneumatic flexible grippers and reducing their reliability and durability. Moreover, pneumatic grippers require extensive peripheral air circuit equipment, including noisy air pumps and high-pressure air storage tubes, which pose safety hazards in mushroom houses.

In this paper, we present an underactuated gripper with adaptive force control, more suitable for industrial environments, capable of selective picking of brown mushrooms and accurately controlling the picking force to ensure quality. This paper contributes the following: (1) designing an underactuated electric picking gripper capable of lifting, grasping, and rotating functions; (2) achieving nondestructive and efficient picking of brown mushrooms based on precise force feedback control; and (3) proposing an in situ measurement method that allows automatic acquisition of the brown mushroom diameter during grasping.

2. Materials and Methods

2.1. Design Basis

The cross-sectional model of a single brown Agaricus bisporus mushroom is shown in Figure 1a. The main body of the brown Agaricus bisporus mushroom consists of the cap and the stem. When mature, its surface appears dark brown, and the cap is large and hemispherical.

The brown mushroom needs to be classified by the size of the cap before being picked. There is no clear standard for grading the size of the cap. We apply a classification method, as shown in

Table 1. Mushroom Varieties and Characteristics.

Variety	Cap Diameter	Open Cap	Classification
Cremini mushroom	<6 cm	No	I
Portabellini mushroom	6–8 cm	No	II
Portobello mushroom	8–10 cm	No	III
Continue to grow	>10 cm	Yes	Low value

, which is used by Yangzhou Aojit Biotechnology Co. LTD, Zhenjiang, China, a large Agaricus bisporus mushroom production company in Yangzhou, Jiangsu Province, China. The cremini mushroom is the child form of the brown Agaricus bisporus mushroom, with a cap size of less than 6 cm. The portobello mushroom is the mature form, characterized by its large size, dark color, and rich flavor, with a cap size of 8 cm to 10 cm. The portobellini mushroom is intermediate between the cremini and portobello mushrooms, with a cap size of 6 cm to 8 cm. If the mushroom is not picked in time and the cap continues to grow and open, it decreases in commercial value.

The optimal picking time for portobello mushrooms is generally when the cap diameter $d$ is 8–10 cm, the height above the substrate $H$ is 5–7 cm, and the cap height $h$ is 2–3 cm. At this stage, the cap has just detached from the stem and is inwardly curled, with a loose and thick flesh, making it the best stage for consumption. The detailed mechanical properties of the portobello mushroom are discussed in Section 3.1.

Picking requires delicate actions to avoid damaging the mucous-covered surface on the cap of the brown Agaricus bisporus mushroom. If this layer is damaged, the mushroom surface quickly blackens, affecting its quality. Thus, to ensure the quality of the mushrooms, traditional manual picking employs standardized three-step actions (Figure 1b):

• Spread three or four fingers open to wrap the edge of the mushroom cap.
• Rotate the mushroom cap once to the left and once to the right.
• Quickly and non-destructively detach the stem from the substrate.

According to this standardized picking action, the designed gripper needs to have three degrees of freedom: grasping, rotation, and lifting. In detail, the gripper should have the ability to open and close to achieve the picking and releasing of the mushroom; it should have the capability to rotate entirely, enabling the twisting action required to harvest the mushroom after gripping it; and it should have the ability to lift and lower, allowing the mushroom to be raised after picking and transferred into a tray. Moreover, the gripper should have the capability to measure the size of the mushroom cap, perform in situ classification, and selectively pick up the mushrooms.

2.2. Design and Fabrication

Considering that the growth temperature of mushrooms is approximately 17–18 °C, with a humidity level of 85–90% [43], the main body of the gripper is made of 304 stainless steel. According to the standardized three-step actions, the designed gripper consists of a finger drive module, a rotation module, and a lifting module, as shown in Figure 2a. The finger drive module employs an underactuated structure, using a through-type stepper motor (UMot28HS4401 (Chongqing Umot Technology Co. Ltd., Chongqing, China)) to drive the lead screw for synchronous movement of four fingers in the vertical direction. The soft rubber gloves installed on the finger tips are 3D printed using rubber with a Shore A hardness of 70 (WeNext Technology Co., Ltd., Shenzhen, China). During the picking process, the small mass of the brown mushroom and the low-speed requirement prioritize the size and weight of the drive elements over power. Thus, this through-type stepper motor has a voltage of 12 V, a torque of 90 mN·m, a lead screw length of 90 mm, a lead screw specification of Tr5, a lead of 4 mm, a maximum vertical load of 1.5 kg, a maximum no-load vertical speed of 80 mm/s, and an operating speed of 30 mm/s at 450 rpm. Additionally, this motor is equipped with an encoder capable of recording the rotation degree of the screw, allowing the open size of the gripper to be calculated. The detailed Mathematica derivation process is shown in Section 2.4.

Specifically, the lower end of the lead screw is rigidly connected to the joint bearing mounting base. The fingers are sleeved with linear bearing sleeves, which are hingedly connected to the joint bearing mounting base, allowing free rotation. One end of each finger is linked to the base via a chain connection, while the other end is fitted with flexible finger sleeves. The full close and full open states of the gripper are shown in Figure 2b and Figure 2c, respectively. The synapses on the flexible finger sleeves increase the stability of the grip. The overall dimensions of the gripper are 170 × 100 × 190 mm³, indicating that the gripper is flexible.

The rotation module uses a stepper motor to drive a crank–rocker mechanism, achieving a $\pm {45}^{°}$ rotation of the gripper.

The lifting module utilizes a lead screw mechanism to achieve vertical lifting of the gripper, with a vertical lift height of 100 mm. The ball screw used is a FUYU brand(Chongqing Fuyu Technology Co. Ltd., Chongqing, China), model FSK30, with specifications Tr8*4—100 mm.

The full picking action process is as follows:

• The lifting module fixes the gripper at an appropriate height.
• The finger drive motor operates to wrap the fingers around the mushroom.
• Measure the size of the cap and classify the mushroom in situ.
• The rotation motor drives the gripper to rotate, detaching the stem.
• The lifting module operates to lift the mushroom.

2.3. Kinematic Modeling and Measurement Methods

There are various methods for robot modeling, such as the Denavit–Hartenberg (DH) model and the F/S model. Considering that the gripper is composed of four symmetrical fingers, with each finger’s drive part abstracted as a closed-loop five-bar linkage, the simplest form of expression is the closed-loop chain method [44]. As shown in Figure 3a,b, the kinematic equation of the five-bar linkage structure of the gripper is as follows:

$$l_{\mathrm{AB}}{e}^{i{\theta }_{\mathrm{AB}}}+l_{\mathrm{BC}}{e}^{i{\theta }_{\mathrm{B}}}+l_{\mathrm{CD}}{e}^{i{\theta }_{\mathrm{C}}}=l_{\mathrm{AE}}{e}^{i{\theta }_{\mathrm{AE}}}+l_{\mathrm{ED}}{e}^{i{\theta }_{\mathrm{B}}},$$

(1)

where $A,B,C,D$, and $E$ are the five rotational points. The angles are defined by starting from the positive direction of the x-axis, and counterclockwise rotation is considered positive. The angles $\angle ABC$, $\angle BCD$, and $\angle EAB$ are right angles; hence, ${\theta }_A={\displaystyle \frac{3}{2}}\pi ,{\theta }_B=0$, and ${\theta }_C={\displaystyle \frac{1}{2}}\pi$. The lengths $l_{\mathrm{AE}}$, $l_{\mathrm{ED}}$ and $l_{\mathrm{DC}}$ are determined by the mechanical parameters of the gripper and are fixed constants. The length $l_{AB}$ is determined by the stepper motor and is given by

$$l_{\mathrm{AB}}=L+l_m,$$

(2)

where $L$ is the length of $l_{\mathrm{AB}}$ when the gripper is fully open, which is the minimum length of $l_{\mathrm{AB}}$, i.e., $L=\min \left\{l_{\mathrm{AB}}\right\}$. The length l_m is the distance moved by the lead screw driven by the stepper motor,

$$l_m={\displaystyle \frac{1}{360}}N\times {\theta }_{\mathrm{step} }\times P,$$

(3)

where $N$ is the number of pulses, ${\theta }_{\mathrm{step}}$ is the step angle of the stepper motor, and $P$ is the lead screw pitch.

Since the diameter is an important basis for mushroom classification, another feature of this gripper is its ability to measure the mushroom diameter while grasping it. Assuming the gripper ideally grips the mushroom, the grasping radius $R$ can be inferred as

$$R=l_{AE}+l_1+l_2,$$

(4)

$$=l_{AE}+l_{EF}\sin (\alpha )+l_{FG}\sin (\beta ),$$

(5)

As shown in Figure 3, the $\alpha =\pi /2-{\theta }_E$, and the $\beta =\angle EFG-\pi /2-{\theta }_E$, the $R$ can be rewritten as

$$R=l_{AE}+l_{EF}\cos \left({\theta }_E\right)+l_{FG}\sin \left(\angle EFG-\pi /2-{\theta }_E\right),$$

(6)

The angle ${\theta }_E$ can be calculated by substituting the parameters into the kinematic model Equation (1), and the other parameters are constant due to the mechanism of the gripper. Thus, the gripper diameter, namely the diameter of the mushroom cap $D=2R$, can be measured. The Equation (6) supports the diameter measurement in Section 3.3 and Section 3.4.

For convenience, the forward kinematics of the gripper is denoted as

$$h(\theta (t)),$$

(7)

where $\theta (t)$ is the rotation angle of the motor.

The velocity vector of the finger motion, $\dot{x}$, is given by

$$\dot{x}(\theta (t))=J(\theta (t))\dot{\theta }(t),$$

(8)

where $J(\theta (t))$ is the Jacobian matrix, and $\dot{\theta }(t)$ is the motor speed.

2.4. Dynamic Model

Based on the principle of virtual power, the input and output virtual power calculations are as follows:

$${\Psi }^T\dot{\theta }=f^{T}v,$$

(9)

where $\Psi$ is the input tor picking time que vector of the stepper motor, $\theta$ is the corresponding angular velocity vector of the motor, $f$ is the normal contact force between the gripper and the mushroom, and $v$ is the corresponding contact velocity vector.

Equation (9) presents the relationship between the motor’s rotational speed and the fingertip force, providing the mathematical basis for accurately controlling the force applied by the gripper on the mushroom.

2.5. Soft Control

To achieve precise and nondestructive picking of brown mushrooms, we have affixed high-sensitivity thin-film pressure sensors (IMS-C10A, WAAAX Inc. (Waaax Electronic Technology Co. Ltd., Luoyang, China)) to the inner side of the gripper using glue (Sil-Poxy, Smooth-on Inc., Macungie, PA, USA), as shown in Figure 4a. These sensors have a circular sensitive area with a diameter of 10 mm and a force range of 0.5–20 N. These sensors operate on the principle of resistance change and can accurately monitor the force variations when in contact with the mushroom. A conditioning circuit was designed to convert these contact force changes into digital signal. As shown in Figure 4b, the flexible thin-film force sensor forms a single-arm bridge circuit with resistor R1, resistor R2, and potentiometer R3. By adjusting potentiometer R3, the initial output of the bridge can be set to zero. Subsequently, the signal from the resistance change of the bending sensor is amplified and filtered through a two-stage amplification circuit and a T-type filter circuit before being output. Then, an analog-to-digital converter (ADC) of a microcontroller reads the voltage signal and converts it into a digital value.

As Figure 5 shows, the digitized signal serves as the input for a PID (Proportional–Integral–Derivative) controller running on the microcontroller. Based on these inputs, the PID controller calculates an output value, which is then converted into pulse signals to drive the stepper motor. The stepper motor, in turn, rotates the lead screw to precisely control the gripping force of the gripper, thereby achieving a constant force grip on the mushroom. This method ensures that the mushroom is neither damaged by excessive force nor fails to be grasped due to insufficient force.

To further enhance system stability and prevent oscillations during control, an error threshold is set. When the actual contact force applied by the gripper falls within this error threshold range of the preset contact force, the stepper motor halts its movement to avoid over-adjustment and ensure a stable grip. This mechanism not only guarantees the precision of the gripping action but also ensures operational smoothness, significantly increasing the success rate and safety of brown mushroom picking. Through this technological implementation, efficient, precise, and automated picking of mushrooms can be achieved without causing damage.

3. Experiment and Results

3.1. Mushroom Mechanism Test

The basic mechanical parameters required for the measurement of brown mushrooms include the elastic modulus, Poisson’s ratio, and the coefficient of sliding friction between the mushroom and flexible materials. These parameters are useful for modeling and simulating the gripping process of the mushrooms.

We took portobello mushrooms to test the mechanism parameters. Within the elastic limit of the portobello mushroom, the elastic modulus E is calculated from the normal stress $\sigma$ and normal strain $\epsilon$, as follows:

$$E={\displaystyle \frac{\sigma }{\epsilon }}={\displaystyle \frac{FL}{S\Delta L}},$$

(10)

where $F$ is the applied external force, $L$ is the initial length of the sample, $S$ is the initial cross-sectional area of the sample, and $\Delta L$ is the deformation under the applied force.

The negative ratio of the strain in the direction perpendicular to the load to the strain in the direction of the load defines the Poisson’s ratio $v$ of the portobello mushroom, which can be calculated based on Boussinesq’s theory,

$$v=\sqrt{1-{\displaystyle \frac{2rE\Delta D}{F}}},$$

(11)

where $r$ is the radius of the compression film, and $\Delta D$ is the deformation in the material.

The coefficient of sliding friction $\mu$ is an important parameter that measures whether relative sliding occurs between the portobello mushroom and the gripper’s glove. It depends on the roughness of the contact surface and is calculated as follows:

$$\mu ={\displaystyle \frac{T/R}{Q}}={\displaystyle \frac{T}{RQ}},$$

(12)

where $T$ is the friction torque, $Q$ is the applied force on the sample, and $R$ is the outer radius of the cylindrical sample [45].

Samples are prepared from the cap of the portobello mushroom. A TMS-PRO (Food Technology Co. Ltd., Sterling, VA, USA) texture analyzer with a compression film radius $r_m=2 \mathrm{cm}$ cylindrical mold is used to perform compression tests, measuring the force–displacement curve. Samples are cut radially from the cap of the portobello mushroom to create cylindrical samples of flexible material with an outer radius $R_m=2 \mathrm{cm}$. Wear tests are conducted using an M-2000-type (Jinan Hengxu Technology Co. Ltd., Jinan, China) wear testing machine.

The mushrooms were delivered from Yangzhou Aojit Biotechnology Co., LTD., Zhenjiang, China within 48 h after being picked. Eight portobello mushrooms were selected, without any visible damage, as judged by eye. All the measurement tests were conducted 8 times. The average values were calculated to determine the mechanical parameters of the portobello mushrooms, as shown in

Table 2. Mechanical Parameters of Portobello Mushrooms.

Parameter	Value
Gravity G/N	2 ± 0.05
Poisson’s Ratio v	0.325 ± 0.01
Elastic Modulus E/MPa	1.49 ± 0.03
Coefficient of Sliding Friction μ	2 ± 0.05
Tensile Force at Root fg/N	1.8 ± 0.12
Rupture Force of Cap fs/N	25 ± 0.37
Tensile Force of Cap fi/N	8 ± 0.24
Torque for Soil–Stem Separation T1 (N·m)	0.006 ± 0.0005
Torque for Cap–Stem Separation T2 (N·m)	0.048 ± 0.03

.

As the torque required for soil–stem separation is much smaller than that for cap–stem separation, the stem will separate from the soil first, and the rotation torque will return to zero to avoid detaching the cap from the stem. Additionally, the force should grasp the mushroom firmly without damaging the cap. Thus, the grasping force $F_g$ should satisfy the condition $G/\mu <F_g<f$, namely $1 \mathrm{N}<F_g<25 \mathrm{N}$. To a practice consideration, a middle value of $F_g=10 \mathrm{N}$ can be chosen in Section 3.4.

3.2. Force Control Accuracy

An experiment was conducted to verify the force control accuracy. As shown in Figure 6a, the gripper and the force gauge (ZTS-50N, IMADA Inc., Northbrook, IL, USA) are fixed to ensure measurement accuracy. The mechanical structure of the gripper guarantees consistent force output from each finger. To avoid collision between the fingers and the gauge, three fingers were removed, leaving only one finger installed. The force gauge was used to provide the ground truth. The target force was set from 1 N to 15 N in 1 N increments, and each target was tested three times.

The PID controller parameters were adjusted using the Ziegler–Nichols tuning method [46]. The resulting PID parameters were fine-tuned through iterative testing to optimize system performance, resulting in a rise time of 1.1 s, a settling time of 1.7 s, and an overshoot of 3.2%. As shown in Figure 6b, the controlled force is consistent with the ground truth, with an average error of 0.195 N. Due to the small error, the error bars are almost indistinguishable and appear as transverse lines in the figure.

3.3. Diameter Measurement Accuracy

The measurement accuracy was verified using standard cylinders with diameters ranging from 30 mm to 100 mm in 10 mm increments, produced by a 3D printer. During the gripper’s grasp of the cylinder, the grip force increases to a preset threshold indicating a steady grasp. The diameter of the cylinder can then be calculated by the method proposed in Section 2.3. As shown in Figure 7, the measurement results are consistent with the ground truth, with an average error of 0.83 mm. The main errors may arise from the lead screw gap. All measurements were conducted three times, and the error bars are present in the figure but are too small to be clearly visible.

3.4. Field Picking Test

In a test group, 10 portobello mushrooms and 10 portobellini mushrooms were picked. The experiment was repeated three times, with the average grasping force set at 10 N. The metrics calculated included detachment force overshoot, damage rate, picking success rate, diameter accuracy, and picking time.

The detachment force overshoot is an important parameter because, during the step of rotating the mushroom to detach it from the substrate, the force may overshoot in a short time. The detachment force overshoot is calculated as the maximum instantaneous force minus prefixed grasping force, namely 10 N. The damage rate is defined as the proportion of mushrooms that exhibit peeling or noticeable scratches on their surface due to the action of the gripper during picking, calculated by dividing the number of damaged mushrooms by the total number picked. Successful picking is defined as the robotic gripper completing the steps of approaching the mushroom, grasping it, and extracting it from the soil. The picking success rate is then calculated by dividing the number of successfully picked mushrooms by the total number of mushrooms. The accuracy of the diameter is determined by dividing the diameter measured by the gripper by the manually measured diameter. Since the mushrooms are not ideally circular, manual measurements involve measuring the diameter twice. The first measurement direction was chosen to be the largest dimension, and the second direction was perpendicular to the first. The average of these two measurements is then calculated to obtain the final result for the manual measurements. To illustrate the different shapes of the mushrooms, they are fitted as ellipses using the first direction length and the second direction length (Figure 8).

Illustrated in Figure 9a are the steps involved in the robotic picking of mushrooms. The time required to complete all the steps in Figure 9a once is referred to as the picking time. In the initial phase, the gripper opens and slowly approaches the target mushroom, generating no contact force as it has yet to touch the mushroom. As the gripper reaches the predetermined position, its four fingers begin to slowly close, gradually increasing the force until reaching the preset grasping force, ensuring a secure grip on the mushroom without causing damage. Subsequently, the gripper performs a rotation operation. Through flexible force control, the mushroom’s stem is gently yet firmly twisted off, avoiding unnecessary damage to the cap of the mushroom.

According to industry practices, harvested mushrooms are typically put on shelves for sale within 48 h and sold within 24 h. Thus, the 72 h [25] mark is critical to determine if there is any damage. If no significant color change is observed after 72 h, it can be concluded that there is no visible damage. In this study, no quantitative measurement was conducted to evaluate the bruise severity level. As shown in Figure 9b, the picked mushroom shows no obvious color change, indicating a harmless picking process.

The results of the field picking test can be seen in

Table 3. Experimental Results for Brown Mushroom picking.

Parameter	Test 1	Test 2	Test 3	Average Value
Detachment Force Overshoot	1.40 N	1.21 N	1.32 N	1.31 N
Damage Rate	0%	0%	5%	1.67%
Picking Success Rate	100%	100%	95%	98.3%
Diameter Accuracy	98.1%	97.4%	96.5%	97.3%
Picking Time	8.05 s	8.15 s	8.22 s	8.14 s

. The average detachment force overshoot is only 1.31 N. As the rupture force of the cap is 25 N in Table 2, 11.45 N is quite a save force. And all the measurements’ accuracy values are greater than 96.5%, with average value 97.3%. Regarding the damage rate and the picking success rate, both Test 1 and Test 2 achieved no mushroom damage and a 100% picking success rate. However, in Test 3, a picking failure occurred (Figure 9c). This failure not only resulted in an unsuccessful pick but also left scratches on the surface of the mushroom, directly affecting its quality. A detailed analysis of this failure revealed that despite timely cleaning of the mucus on the finger sleeves to maintain optimal gripping, by the third experiment, the accumulation of mucus on the finger sleeves reached a critical point, causing slippage and affecting the success rate of picking. The average picking time is 8.14 s.

4. Discussion

To more clearly reflect the contribution of this article, the features of our previous gripper and other related grippers are detailed in

Table 4. Comparison of the gripper for picking the mushroom.

Author	Structure	Actuation	In Situ Measurement	Force Control	Success Rate	Damage Rate
This work	Rigid	Motor-driven	Yes	Yes	98.3%	1.67%
Ours [34]	Rubber gripper	Positive pneumatic	Yes	Yes	No application	No application
Yang et al. [22]	Suction cup	Negative pneumatic	No	No	88.2%	2.9%
Reed et al. [20]	Suction cup	Negative Pneumatic	No	No	80%	Not mentioned
Huang et al. [47]	Suction cup	Negative Pneumatic	No	No	100%	Not mentioned

. The pneumatic gripper made of silicone rubber [34] is capable of in-measurement and force control, but it remains a prototype and has not been used in actual mushroom cultivation. Yang et al. [22] utilized a suction cup to grip the mushroom cap, achieving an 88.2% success rate and a 2.9% damage rate under a mean allowable pressure of 13.82 N. Although the suction force can be adjusted by varying the negative pressure, the actual force cannot be measured directly in real time due to the difficulty of installing a force sensor on the cap. The proposed gripper in this paper, however, is capable of force control, achieving a higher success rate of 98.3% and a lower damage rate of 1.67%. Additionally, the proposed gripper can achieve in situ measurement with an average accuracy of 97.3%, while the suction caps cannot. The other two grippers [20,47] using the suction cup face the same problem: they are incapable of in situ measurement and force control.

However, the experiments revealed some issues, particularly concerning the impact of mucus on the mushroom surface on picking efficiency and mushroom quality. The mucus on the mushroom surface adheres to the silicone finger sleeves of the gripper, reducing the friction between the gripper and the mushroom and increasing the risk of surface damage during grasping. Current solutions, such as manually wiping the finger sleeves or replacing them periodically, can mitigate this issue to some extent. However, these practices increase labor costs and are inefficient, hindering the realization of fully automated production.

In response to this issue, adding an automatic finger cleaning device appears to be a direct and effective method. Such a device would automatically clean the gripper after picking each batch of mushrooms, ensuring optimal conditions for each grasp and effectively avoiding slippage due to mucus accumulation. Moreover, considering that a cleaning device might increase the complexity and failure rate of the machine, using nano or other polymer materials to create self-cleaning finger sleeves represents another innovative direction. These materials could automatically repel mucilage through specific surface treatments, such as superhydrophobic or super oleophobic coatings, thereby reducing mucilage adhesion and decreasing the frequency of cleaning.

5. Conclusions

This paper introduces a novel flexible gripper based on force control, powered by electricity, designed for picking delicate brown mushrooms. It achieves precise grasping forces during the picking process by utilizing force sensor and operational amplifiers to provide force feedback, with a PID algorithm enhancing the accuracy of the grip. An in situ measurement method is proposed based on the kinematic model of the gripper. The experimental results demonstrate that the proposed gripper has the static grasping force error 0.195 N, and the average detachment force overshoot during the whole picking up is 1.31 N. The in situ measurement of the diameter of the mushroom achieves 97.3% accuracy. The success rate is 98.3%, with only one mushroom sustaining a scratch.

In future work, a comprehensive mushroom force mechanism model will be established, enabling the optimization of the grasping force. The designed flexible gripper will be installed on our in-development brown mushroom-picking robot, which will perform actual picking operations in a mushroom cultivation room, achieving fully autonomous operation.