In the realm of agricultural robotics, the development of specialized end effectors for harvesting delicate crops remains a significant challenge. This article presents a comprehensive design and analysis of a three-finger pull-type end effector tailored for the mechanical harvesting of safflower. The design is inspired by the manual harvesting process, where human fingers effectively gather and pluck the safflower filaments. Through detailed material property analysis, structural design, kinematic simulation, and experimental validation, we demonstrate that this end effector can achieve high harvesting efficiency with minimal damage to the crop. The insights gained from this work contribute to advancing automation in safflower harvesting, potentially reducing labor costs and improving yield. Throughout this discussion, the term “end effector” will be frequently emphasized to underscore its critical role in robotic harvesting systems.
Safflower, a valuable herbaceous plant known for its medicinal, dye, and feed applications, requires timely harvesting to prevent waste. Currently, harvesting is predominantly manual due to the lack of efficient mechanical solutions. Existing machines, such as pneumatic or portable roller-type devices, have not significantly improved efficiency, necessitating the development of advanced robotic end effectors. In recent years, agricultural robots have seen rapid growth, with end effectors designed for fruits and vegetables like tomatoes, strawberries, and citrus. However, safflower presents unique challenges as a clustered, flexible material that requires gathering and pulling actions for separation. This gap in end effector technology motivates our design, which mimics the human three-finger technique to achieve effective harvesting.
The manual harvesting of safflower involves a precise sequence of actions: the thumb, index, and middle fingers vertically approach the safflower filaments, converge to gather them, apply a clamping force using the finger pads, and finally perform an upward pulling motion to separate the filaments from the seed head. This process ensures high harvest purity and low seed head damage. By analyzing this biomechanical process, we derived the fundamental requirements for our end effector. The end effector must replicate the three-finger coordination, provide sufficient clamping force, and execute a smooth pulling action. This human-inspired approach forms the basis for our end effector design, aiming to bridge the gap between manual dexterity and robotic automation.
The overall design of the three-finger pull-type end effector integrates several key components to achieve the desired functionality. The end effector is mounted on a robotic arm, allowing for precise positioning. It consists of a frame, a clamping mechanism, a motor, and three fingers. The clamping mechanism includes an end cover, springs, guide chutes with tapered slots, a connection plate, and a positioning tube. The working principle involves: first, the robotic arm moves the end effector above the safflower seed head using machine vision for localization; second, the end effector descends vertically until the seed head enters the positioning tube and contacts the finger pads; third, the motor lifts the end cover connected to the guide chutes, causing the fingers to converge inward due to the taper, thus clamping the filaments; and fourth, the robotic arm performs an upward pull to separate the filaments. This sequential operation ensures that the end effector mimics human actions while leveraging mechanical precision.
To ensure effective harvesting, a detailed mechanical model was developed for the end effector. During clamping, the three fingers apply normal forces (\(F_n\)) on the safflower filaments, generating frictional forces (\(F_s\)) that overcome the separation force (\(F_l\)) between the filaments and the seed head. The relationship is governed by the following equations, where \(\mu\) is the coefficient of friction between the rubber-coated finger pads and the filaments:
$$F_s = F_n \mu$$
The total pulling force (\(F_t\)) is the sum of the frictional forces from all three fingers:
$$F_t = 3F_n \mu$$
Through experimental measurements, the maximum separation force \(F_l\) was found to be 25 N. The static friction coefficient \(\mu\) was determined using a static sliding friction angle measuring instrument, yielding an angle \(\alpha = 19.29^\circ\). The coefficient is calculated as:
$$\mu = \tan \alpha = 0.35$$
Substituting into the equations, the required normal force per finger is:
$$F_n = \frac{F_l}{3\mu} = \frac{25}{3 \times 0.35} \approx 23.8 \, \text{N}$$
Thus, the end effector must generate at least 23.8 N of normal force per finger to ensure successful separation. This mechanical analysis provides a foundation for designing the clamping mechanism and selecting appropriate actuators for the end effector.
The structural design of the end effector was informed by detailed measurements of safflower morphological characteristics. Key dimensions, such as the corolla diameter, seed head diameter, neck diameter, and neck height, were measured from samples harvested during the peak season. The variability in these parameters necessitates design tolerances to accommodate different safflower sizes. The positioning tube, where the fingers are mounted, must allow the seed head to enter and position the neck at the optimal clamping height. The design criteria include: the installation height of the fingers (\(D_1\)) must be greater than the neck height (\(L_a\)) plus a safety margin, and the diameter of the positioning tube (\(D_2\)) must exceed the maximum seed head diameter (\(d_{gq,\text{max}}\)) with an allowance. These relationships are summarized as:
$$D_1 > L_a + \Delta_1, \quad D_2 > d_{gq,\text{max}} + \Delta_2$$
where \(\Delta_1\) and \(\Delta_2\) are design margins. Additionally, the finger pads are designed with a spherical shape to minimize gaps during clamping. The maximum gap (\(d\)) between fingers when converged is related to the finger pad radius (\(R\)) by:
$$d = \tan 30^\circ \cdot R$$
To prevent slippage and damage, \(d\) is kept much smaller than the minimum neck diameter, and the finger pads are coated with flexible rubber material. The clamping mechanism uses a tapered guide chute system, where the finger displacement (\(\Delta l\)) is linked to the lift height of the chute (\(\Delta h\)) and the taper angle (\(\theta\)):
$$\tan \theta = \frac{\Delta h}{\Delta l}$$
This allows for tuning the speed and force profile of the end effector based on motor selection. For instance, a smaller \(\theta\) increases the finger speed for a given \(\Delta h\), enhancing responsiveness. The design parameters are optimized through kinematic simulations, as discussed later. To summarize the measurement data, Table 1 presents the statistical results for key safflower dimensions, which guided the end effector’s dimensional specifications.
| Parameter | Mean (mm) | Standard Deviation (mm) | Maximum (mm) | Coefficient of Variation |
|---|---|---|---|---|
| Corolla Diameter (\(d_{hg}\)) | 35.19 | 1.12 | 37.03 | 0.031 |
| Seed Head Diameter (\(d_{gq}\)) | 25.91 | 1.98 | 29.77 | 0.076 |
| Neck Diameter (\(d\)) | 6.01 | 0.88 | 7.40 | 0.146 |
| Neck Height (\(L_a\)) | 15.71 | 0.88 | 16.88 | 0.056 |
| Filament Length (\(L_b\)) | 23.77 | 2.11 | 26.33 | 0.088 |
Kinematic simulation using Adams software was conducted to validate the motion and force characteristics of the end effector. A 3D model of the end effector was built and imported into Adams, where constraints and contacts were applied, such as moving pairs between the fingers and positioning tube, and contact forces between the rollers and tapered chutes. The contact force model follows the Hertzian contact theory combined with a continuous impact model, expressed as:
$$F_{\text{contact}} = k (x – x_0)^e – c v$$
where \(k\) is the contact stiffness, \(e\) is the elasticity exponent, \(c\) is the damping coefficient, \(x\) is the penetration depth, \(x_0\) is the initial gap, and \(v\) is the relative velocity. A sinusoidal drive function was applied to the tapered chute to simulate the reciprocating motion, and reset springs were added between the chute and fingers to return the end effector to its initial position after harvesting. The simulation parameters included a gravity acceleration of 9806.65 mm/s², 200 simulation steps, a step size of 0.01 s, and a total time of 3.5 s. The results, shown in the displacement and force curves, indicate that the finger displacement reaches a maximum of 20 mm at approximately 1.6 seconds, corresponding to the peak spring force of 22 N. The force-displacement relationship adheres to Hooke’s law:
$$F = K \Delta x$$
where \(K\) is the spring stiffness and \(\Delta x\) is the compression. The simulation confirms that the end effector can achieve the required clamping force and motion profile, ensuring effective filament grasping and pulling. This analysis is crucial for optimizing the end effector’s performance before physical prototyping.
An experimental testbed was constructed to evaluate the harvesting performance of the end effector. The control system hardware includes an STM32 microcontroller, stepper motors (model: 57BYG56), stepper drivers (model: TB6600), and proximity sensors (model: LJ12A3-4-Z/BX). The workflow involves: the STM32 sends signals to the stepper drivers to control the motors; motor 1 rotates forward to lift the tapered chute, causing finger convergence until a displacement of 20 mm is detected by the sensor, at which point the clamping force meets the requirement; motor 2 then activates to perform the upward pull; finally, both motors reverse to reset the end effector for the next cycle. This automated sequence ensures consistent operation of the end effector during testing.
The experiments were conducted on safflower samples from a test field, with seed head diameters ranging from 18.54 to 40.54 mm, neck diameters from 2.90 to 8.24 mm, and neck heights from 2.04 to 4.20 mm. A precision electronic balance (accuracy: 0.01 g) was used to weigh the harvested filaments and any residues on the seed heads. Two metrics were defined to assess the end effector’s performance: harvest purity rate (\(Y\)) and seed head damage rate (\(X\)). The harvest purity rate is calculated as:
$$Y = \frac{M_a}{M_a + M_b} \times 100\%$$
where \(M_a\) is the mass of separated filaments and \(M_b\) is the mass of residual filaments. The seed head damage rate is:
$$X = \frac{N_s}{N_s + N_w} \times 100\%$$
where \(N_s\) is the number of damaged seed heads and \(N_w\) is the number of intact seed heads. Tests were performed on 20 safflower samples, and the results are summarized in Table 2. The end effector achieved an average harvest purity rate of 92.47% with a low standard deviation, indicating consistent performance across different samples. The seed head damage rate was only 1.24%, demonstrating that the end effector minimizes crop injury. These results validate the design’s effectiveness and its potential for practical application in safflower harvesting.
| Parameter | Mean (%) | Standard Deviation (%) | Maximum (%) | Coefficient of Variation |
|---|---|---|---|---|
| Harvest Purity Rate (\(Y\)) | 92.47 | 3.27 | 98.36 | 0.035 |
| Seed Head Damage Rate (\(X\)) | 1.24 | 0.45 | 1.68 | 0.36 |
In conclusion, the three-finger pull-type end effector designed for safflower harvesting successfully addresses the challenges of automating the harvest of clustered, flexible materials. By emulating human finger actions, incorporating material property analysis, and optimizing structural and kinematic parameters, the end effector achieves high harvest purity and low damage rates. The mechanical model ensures sufficient clamping force, while the simulation and experimental results confirm its practical viability. This end effector represents a significant step toward reducing labor dependency and enhancing efficiency in safflower cultivation. Future work could focus on scaling the design for commercial use, integrating advanced sensors for real-time adaptation, and extending the concept to other similar crops. The development of such specialized end effectors is pivotal for the advancement of agricultural robotics, offering sustainable solutions for modern farming challenges.
The design process highlighted several key considerations for end effector development in agricultural contexts. First, understanding the biological characteristics of the crop is essential for tailoring the end effector’s geometry and mechanics. Second, friction and force analysis must account for variability in material properties to ensure reliable operation. Third, simulation tools like Adams provide valuable insights into motion dynamics, enabling optimization before costly physical trials. Fourth, modular testbeds allow for iterative refinement of the end effector based on empirical data. These principles can guide future innovations in end effector technology, not only for safflower but for a wide range of delicate crops. As robotics continues to evolve, the role of intelligent end effectors will become increasingly central to achieving precision agriculture and food security goals.
Moreover, the economic implications of deploying such end effectors are substantial. By automating safflower harvesting, farmers can reduce labor costs, which often constitute a major portion of production expenses. The end effector’s high efficiency and low damage rate also contribute to better yield quality and higher market value. In regions like Xinjiang, where safflower is extensively cultivated, mechanization could transform local industries and support rural development. However, challenges remain, such as the initial investment in robotic systems and the need for training operators. Continued research into cost-effective and user-friendly end effectors will be crucial for widespread adoption. Collaborative efforts between academic institutions and agricultural sectors can accelerate this transition, leveraging expertise in both robotics and agronomy.
From a technical perspective, the end effector’s design incorporates several innovative features. The three-finger configuration, inspired by human dexterity, offers a balance between simplicity and functionality. The use of tapered guide chutes for finger actuation allows for compact and efficient force transmission. The flexible rubber coating on the finger pads enhances grip while preventing crop damage. These elements collectively make the end effector robust and adaptable to field conditions. Additionally, the integration with a robotic arm and machine vision system enables autonomous operation, though this aspect was not deeply explored in the current study. Future enhancements could include adaptive control algorithms that adjust clamping force based on real-time feedback from vision or force sensors, further improving the end effector’s performance.
The experimental validation phase provided critical data on the end effector’s real-world performance. The harvest purity rate of over 92% meets practical requirements, while the minimal seed head damage ensures that the crop remains viable for subsequent uses, such as seed oil extraction. The low variability in results across different safflower sizes demonstrates the end effector’s robustness. However, limitations were noted, such as occasional slippage in very dry conditions or with overly compacted filaments. These issues can be addressed by refining the finger pad material or adjusting the clamping sequence. Overall, the positive outcomes underscore the potential of this end effector to revolutionize safflower harvesting, paving the way for fully automated systems that combine multiple end effectors on harvesting platforms for large-scale operations.
In summary, this work presents a holistic approach to designing a specialized end effector for safflower harvesting. The combination of biomechanical analysis, mechanical modeling, simulation, and experimentation ensures that the end effector is both theoretically sound and practically effective. The frequent emphasis on the end effector throughout this discussion highlights its centrality in robotic harvesting systems. As agriculture faces increasing demands for productivity and sustainability, innovations in end effector technology will play a vital role. The three-finger pull-type end effector described here serves as a case study in how human-inspired design can bridge the gap between manual and mechanical harvesting, offering a scalable solution for crops that require delicate handling. With further development, such end effectors could become standard tools in precision agriculture, contributing to a more efficient and resilient food system.