In modern agriculture, the automation of fruit harvesting is a critical step toward increasing efficiency and reducing labor costs. Apple harvesting, in particular, presents significant challenges due to the delicate nature of the fruit and the need for precision to avoid damage. Traditional manual harvesting is labor-intensive, time-consuming, and often hazardous, especially when dealing with high branches. To address these issues, I have developed a novel cutting end effector for apple harvesting robots. This end effector is designed to be simple, efficient, and non-damaging, eliminating the need for gripping or sucking mechanisms that can cause bruising or other injuries to the fruit. The core innovation lies in its cutting mechanism, which severs the stem directly without applying pressure to the apple itself. This article details the design, parameter calculations, and motion simulation of this cutting end effector, providing a comprehensive analysis of its performance and feasibility.
The apple harvesting robot system typically consists of a mobile base, a robotic arm, a vision system, a collection device, and the end effector. The end effector is the critical component that interacts directly with the fruit, and its design profoundly impacts the overall harvesting quality and speed. Existing end effectors often rely on gripping or suction, which require complex control systems to regulate force and are susceptible to interference from leaves and debris. My design bypasses these issues by focusing solely on cutting the stem. The workflow of the robot is as follows: the vision system identifies the position and orientation of an apple, the robotic arm maneuvers the end effector into the appropriate position, the end effector performs the cutting action, and the apple falls through a collection tube into a storage container. This process is repeated for continuous harvesting.
The proposed cutting end effector comprises three main subsystems: the body, the cutting mechanism, and the transmission mechanism. The body serves as the structural frame, connecting to the robotic arm’s wrist and enclosing the apple during operation. Its inner cavity is lined with a soft protective layer to prevent any impact damage to the fruit. The cutting mechanism includes two symmetrical blade assemblies, each consisting of a cutting blade, a blade connecting rod, a connecting rod gear, and a sector gear. The blades are mounted on the connecting rods, which are fixed to the connecting rod gears and sector gears. These gears share a common rotation axis fixed to the body. The transmission mechanism consists of a drive motor, a motor gear, transmission gears, and a drive shaft. The motor’s output is transferred through the motor gear to the transmission gears, which are connected by the drive shaft to ensure synchronized movement of the two blade assemblies on either side. The power is then transmitted to the sector gears and subsequently to the connecting rod gears, causing the blades to swing inward and cut the stem. The entire operation is divided into a cutting stroke and a return stroke, controlled by sensors that detect the blades’ positions.
To ensure effective stem cutting, the design parameters are derived from the physical characteristics of apples. Typical apple diameters range from 60 mm to 100 mm. Therefore, the internal diameter of the end effector body is set at D = 120 mm to accommodate most apples comfortably. The distance between the centers of the two connecting rods is S = 145 mm. The length from the blade edge to the pivot center of the connecting rod is L = 70 mm. The blade width is 15 mm, and the connecting rod width is 8 mm. For complete stem severance, the blades must overlap by 2–3 mm at the end of the cutting stroke. When the connecting rods are near the vertical position at the end of the cut, the center distance between them is approximately 19–20 mm. This dictates the pitch diameter of the connecting rod gears to be about 19–20 mm. Considering the minimum number of teeth to avoid undercutting (Z_min ≥ 17), I initially select the number of teeth for the connecting rod gear as Z1 = 20. The center distance is 20 mm, leading to a module m calculated as:
$$ m = \frac{\text{Pitch Diameter}}{Z} = \frac{20 \, \text{mm}}{20} = 1 \, \text{mm} $$
Thus, the gear module is standardized to m = 1 mm. The required cutting time for one complete operation is t ≤ 1 s to maintain high harvesting efficiency. The designed swing angle for the connecting rod is θ = 55°. The angular velocity ω of the connecting rod gear must satisfy:
$$ \omega \geq \frac{2\theta}{t} $$
Converting to rotational speed n1 (in revolutions per minute, rpm):
$$ n_1 \geq \frac{2\theta}{t} \times \frac{60}{2\pi} = \frac{2 \times 55^\circ}{1 \, \text{s}} \times \frac{60}{2\pi \times \frac{180^\circ}{\pi}} = \frac{110 \, \text{deg}}{1 \, \text{s}} \times \frac{60}{360 \, \text{deg}} = 18.3 \, \text{rpm} $$
I choose a DC motor with an output power of 8 W, a rated speed of n0 = 52 rpm, and a maximum torque of 5400 N·mm. The speed reduction ratio i from the motor to the connecting rod gear is:
$$ i = \frac{n_0}{n_1} = \frac{52}{18.3} \approx 2.84 $$
To distribute manufacturing errors and simplify the transmission, the motor gear and the transmission gear are assigned the same number of teeth. For compactness, the transmission gear is positioned near the edge of the body, with a center distance between the transmission gear and the connecting rod gear of approximately a1 = 57 mm. The relationship for spur gears is:
$$ a_1 = m \cdot \frac{Z_2 + Z_3}{2} $$
where Z2 is the number of teeth on the connecting rod gear (80, as derived from the sector gear design), and Z3 is the number of teeth on the transmission gear. Solving for Z3:
$$ 57 = 1 \times \frac{80 + Z_3}{2} \Rightarrow 114 = 80 + Z_3 \Rightarrow Z_3 = 34 $$
However, to achieve the desired speed ratio and ensure smooth meshing, I adjust the gear teeth counts through iterative design. After optimization, the final gear parameters are summarized in Table 1.
| Component | Number of Teeth (Z) | Module (m) [mm] | Pitch Diameter [mm] | Function |
|---|---|---|---|---|
| Connecting Rod Gear | 20 | 1 | 20 | Drives the blade connecting rod |
| Sector Gear | 80 | 1 | 80 | Transmits motion from transmission gear |
| Transmission Gear | 30 | 1 | 30 | Transfers torque from motor gear |
| Motor Gear | 30 | 1 | 30 | Connected directly to the motor shaft |
The actual speed ratio is recalculated based on these gear teeth. The transmission from the motor gear to the sector gear involves two stages: motor gear to transmission gear, and transmission gear to sector gear. The overall speed ratio i_total is:
$$ i_{\text{total}} = \frac{Z_{\text{sector gear}}}{Z_{\text{motor gear}}} = \frac{80}{30} \approx 2.67 $$
This yields a connecting rod gear speed of:
$$ n_1 = \frac{n_0}{i_{\text{total}}} = \frac{52}{2.67} \approx 19.5 \, \text{rpm} $$
which meets the requirement of n1 ≥ 18.3 rpm. The cutting time for a 55° swing is:
$$ t = \frac{2\theta}{\omega} = \frac{2 \times 55^\circ}{19.5 \, \text{rpm} \times \frac{2\pi}{60} \, \text{deg/s}} \approx \frac{110}{19.5 \times 0.1047} \approx 0.54 \, \text{s} $$
This is well within the 1 s constraint, ensuring rapid operation of the end effector.
Next, I calculate the torque requirements to ensure the motor can provide sufficient cutting force. According to research data, the cutting force required to sever an apple stem varies with apple diameter. For a typical apple diameter of d = 75 mm, the stem cutting force is approximately F = 30 N. The worst-case scenario occurs when the apple is positioned against one side of the end effector cavity, placing the stem closer to one blade. In this case, the torque on the connecting rod gear nearest to the stem is higher. The torque M1 on the connecting rod gear is given by:
$$ M_1 = F \cdot L \cdot K $$
where L = 70 mm is the lever arm, and K is the load factor accounting for various dynamic and static loads in gear transmission. The load factor K is calculated as:
$$ K = K_A \cdot K_v \cdot K_\alpha \cdot K_\beta $$
where KA is the application factor (taken as 1.0 for uniform load), Kv is the dynamic factor (1.05 for low-speed gears), Kα is the load distribution factor across teeth (1.0 for precise gears), and Kβ is the load distribution factor along tooth width (1.2 for moderate face width). Thus:
$$ K = 1.0 \times 1.05 \times 1.0 \times 1.2 = 1.26 $$
I conservatively use K = 1.3 for safety. Therefore:
$$ M_1 = 30 \, \text{N} \times 0.07 \, \text{m} \times 1.3 = 2.73 \, \text{N} \cdot \text{m} = 2730 \, \text{N} \cdot \text{mm} $$
The torque on the transmission gear M2 and motor gear M3 are derived considering the gear ratios. The transmission gear torque is:
$$ M_2 = M_1 \cdot \frac{Z_3}{Z_2} = 2730 \times \frac{30}{80} = 1023.75 \, \text{N} \cdot \text{mm} $$
Since the drive shaft connects two transmission gears, the total torque required from the motor gear is essentially the sum for both sides, but in the worst-case scenario with asymmetric loading, the motor must supply torque for the side with higher load. The motor gear torque M3 is:
$$ M_3 = M_1 \cdot \frac{Z_4}{Z_2} = 2730 \times \frac{30}{80} = 1023.75 \, \text{N} \cdot \text{mm} $$
However, this calculation only accounts for one side. To ensure the motor can handle the peak load, I consider the total torque when both blades are cutting simultaneously under maximum force. But since the design ensures that the blades cut sequentially or symmetrically, the motor torque requirement is dominated by the single-side worst case. The selected motor with a maximum torque of 5400 N·mm far exceeds 1023.75 N·mm, providing a substantial safety margin. The required motor speed is already verified as n0 ≥ 48.9 rpm (from n1 × i_total), and the chosen 52 rpm motor meets this. Table 2 summarizes the torque calculations.
| Component | Torque (M) [N·mm] | Calculation Formula | Remarks |
|---|---|---|---|
| Connecting Rod Gear | 2730 | M1 = F × L × K | Worst-case single-side load |
| Transmission Gear | 1023.75 | M2 = M1 × (Z3/Z2) | Per side, assuming force on one gear |
| Motor Gear | 1023.75 | M3 = M1 × (Z4/Z2) | Required motor output torque |
| Motor Available | 5400 (max) | From specifications | Provides safety factor >5 |
To validate the kinematic performance of the cutting end effector, I conducted a motion simulation using ADAMS software. A three-dimensional model of the transmission and cutting mechanisms was created in SolidWorks and imported into ADAMS. The materials were assigned: steel for the blades, drive shaft, and connecting rods, and nylon for the gears to reduce weight and noise. Constraints were applied: fixed joints for stationary parts, revolute joints for rotating gears, and contact forces between mating gear teeth. The drive shaft was converted to a flexible body to account for deformations during motion, enhancing simulation accuracy. The motor gear was given a rotational speed drive of 52 rpm. The simulation was run for 5.2 seconds with 1000 steps to capture a complete cutting cycle.
To analyze the cutting action, I placed marker points on the cutting edges of both blades: Marker102 on the left blade and Marker104 on the right blade. The Y-direction displacement of these markers was monitored to determine when the blades overlap. Figure 1 shows the simulation model setup in ADAMS, illustrating the gear assemblies and blade positions.
The displacement results indicate that the relative Y-distance between Marker102 and Marker104 reduces to zero at approximately 0.46 seconds, confirming that the blades intersect and fully cut the stem. This cutting time is even faster than the calculated 0.54 seconds, likely due to the dynamic effects and flexibility in the simulation. The angular velocity of Marker104 was also plotted, showing an average value around 117 degrees per second with minor periodic fluctuations. These fluctuations are caused by the inherent impacts during gear meshing as teeth engage and disengage, but they are within acceptable limits and do not affect the cutting reliability. The average angular velocity corresponds to a rotational speed of:
$$ n = \frac{117 \, \text{deg/s} \times 60}{360 \, \text{deg}} = 19.5 \, \text{rpm} $$
which matches the design target. The simulation thus verifies that the cutting end effector operates as intended, with smooth motion and precise stem severance within the required time frame.
Further analysis involves the stress on gear teeth during operation. Using the calculated torques, the tangential force Ft on the gear teeth can be derived. For the motor gear with torque M3 = 1023.75 N·mm and pitch radius r = m × Z4 / 2 = 1 × 30 / 2 = 15 mm:
$$ F_t = \frac{M_3}{r} = \frac{1023.75 \, \text{N} \cdot \text{mm}}{15 \, \text{mm}} = 68.25 \, \text{N} $$
The bending stress σ_b on the gear tooth can be estimated using the Lewis formula:
$$ \sigma_b = \frac{F_t}{b \cdot m \cdot Y} $$
where b is the face width (assumed as 10 mm for compactness), and Y is the Lewis form factor (approximately 0.3 for 30 teeth). Thus:
$$ \sigma_b = \frac{68.25}{10 \times 1 \times 0.3} = 22.75 \, \text{MPa} $$
For nylon gears, the allowable bending stress is typically around 40–60 MPa, indicating a safety factor greater than 2. This confirms the gear strength is sufficient for the application.
Another critical aspect is the energy consumption of the end effector. The motor power P is related to torque and speed by:
$$ P = \frac{M_3 \cdot n_0}{9549} = \frac{1023.75 \times 52}{9549} \approx 5.57 \, \text{W} $$
This is below the motor’s rated 8 W, ensuring efficient operation without overheating. The low power requirement aligns with the goal of developing an energy-efficient harvesting robot.
The design of this cutting end effector offers several advantages over traditional gripping or suction-based end effectors. First, it eliminates the risk of fruit damage from excessive pressure or impact, as the apple is never clamped. The soft inner lining of the body provides cushioning. Second, the mechanism is mechanically simple, with fewer moving parts and no need for complex force sensors or vacuum systems. This reduces cost and maintenance. Third, the cutting action is fast and reliable, enabling high-speed harvesting. The end effector can be adapted to other fruits with similar stem characteristics, such as peaches or pears, by adjusting the internal diameter and blade parameters.
Potential improvements include integrating a stem detection sensor to ensure the blades align precisely with the stem, especially for apples at varied orientations. Additionally, the gear transmission could be optimized for weight reduction using advanced materials like carbon fiber composites. The motion simulation could be extended to include dynamic analysis of the entire robotic arm to study vibrations during harvesting.
In conclusion, I have successfully designed and analyzed a cutting end effector for apple harvesting robots. The end effector features a novel cutting mechanism that severs the stem without contacting the fruit, thereby minimizing damage. Through detailed parameter calculations, I determined the gear specifications, torque requirements, and speed parameters, all of which satisfy the operational constraints. The ADAMS motion simulation confirmed that the cutting action is completed within 0.46 seconds, with smooth and synchronized blade movement. The gear stresses and power consumption are within safe limits. This cutting end effector represents a significant step forward in agricultural robotics, offering a robust and efficient solution for automated fruit harvesting. Future work will involve prototyping and field testing to validate performance under real-world conditions.
The development of such end effectors is crucial for advancing precision agriculture. As robotics technology evolves, integrating AI-based vision systems with intelligent end effectors will enable fully autonomous orchards. My design contributes to this vision by providing a reliable and simple end effector that can be scaled and modified for various crops. The principles outlined here—direct cutting, mechanical simplicity, and rigorous simulation—can guide the design of end effectors for other agricultural tasks, such as pruning or selective harvesting. Ultimately, the goal is to reduce labor dependency, increase productivity, and promote sustainable farming practices through innovative robotic solutions.
To further elaborate on the design considerations, I analyzed the kinematic chain of the end effector. The relationship between the motor rotation and the blade tip displacement is derived using geometric principles. Let φ be the rotation angle of the motor gear. The rotation angle of the sector gear ψ is:
$$ \psi = \phi \cdot \frac{Z_4}{Z_2} = \phi \cdot \frac{30}{80} = 0.375\phi $$
Since the connecting rod gear is fixed to the sector gear, its rotation angle is also ψ. The displacement of the blade tip in Cartesian coordinates (assuming initial position at θ0 = 0) can be modeled as:
$$ x = L \cdot \sin(\psi) $$
$$ y = L \cdot \cos(\psi) $$
where L is the connecting rod length. The velocity of the blade tip is obtained by differentiation:
$$ v_x = L \cdot \cos(\psi) \cdot \dot{\psi} $$
$$ v_y = -L \cdot \sin(\psi) \cdot \dot{\psi} $$
where ˙ψ is the angular velocity of the connecting rod gear. Given the motor angular velocity ω_motor = 52 rpm = 5.45 rad/s, we have:
$$ \dot{\psi} = \omega_{\text{motor}} \cdot \frac{Z_4}{Z_2} = 5.45 \times 0.375 = 2.04 \, \text{rad/s} $$
The maximum blade tip speed occurs when sin(ψ) or cos(ψ) is 1, yielding v_max ≈ L × ˙ψ = 0.07 × 2.04 = 0.143 m/s. This moderate speed ensures precise cutting without causing excessive inertia forces.
For the stem cutting dynamics, the force required to cut through the stem depends on the blade sharpness and stem material properties. Assuming a shear strength τ for apple stems of about 2 MPa, and a stem cross-sectional area A_stem ≈ π × (2 mm)^2 = 12.56 mm², the theoretical cutting force F_cut is:
$$ F_{\text{cut}} = \tau \cdot A_{\text{stem}} = 2 \times 10^6 \, \text{Pa} \times 12.56 \times 10^{-6} \, \text{m}^2 = 25.12 \, \text{N} $$
This aligns with the empirical value of 30 N used in calculations, validating the design force. The blade design includes serrated edges to enhance cutting efficiency by concentrating stress.
In terms of control, the end effector operates with a simple on-off control for the motor, triggered by position sensors. The sensors detect when the blades reach the fully open or fully closed positions, sending signals to reverse or stop the motor. This binary control reduces complexity compared to force-controlled grippers. However, for adaptive harvesting, future versions could incorporate torque feedback to detect cutting completion or jams.
The end effector’s compatibility with various robotic arms is ensured by standard interface dimensions. The body is designed to mount on common robotic wrist flanges, allowing integration with existing harvesting robot platforms. The weight of the end effector is estimated by summing component masses. Using density values (steel: 7850 kg/m³, nylon: 1150 kg/m³), the total mass is approximately 0.5 kg, which is lightweight for most robotic arms.
Table 3 provides a summary of key performance metrics for the cutting end effector, highlighting its efficiency and robustness.
| Metric | Value | Unit | Comment |
|---|---|---|---|
| Cutting Time | 0.46 | s | From simulation, within design limit |
| Motor Power Used | 5.57 | W | Below motor rating, efficient |
| Maximum Blade Speed | 0.143 | m/s | Ensures precise operation |
| End Effector Mass | 0.5 | kg | Lightweight for easy manipulation |
| Gear Safety Factor | >2 | – | Based on bending stress analysis |
| Stem Cutting Force | 30 | N | Empirical value, design basis |
The innovation of this end effector lies in its simplicity and effectiveness. By focusing on the cutting action alone, it avoids the pitfalls of multi-function mechanisms. The use of standard gear components makes it cost-effective and easy to manufacture. The motion simulation validates the design, providing confidence in its real-world performance. As agricultural robotics continues to grow, such dedicated end effectors will play a pivotal role in automating specific tasks.
In summary, this article presents a comprehensive design and analysis of a cutting end effector for apple harvesting robots. The end effector demonstrates how mechanical design principles, combined with simulation tools, can yield efficient solutions for agricultural challenges. The methodologies described—parameter calculation, torque analysis, kinematic simulation—serve as a template for developing end effectors for other crops. The ultimate aim is to contribute to the advancement of robotic harvesting, making it more accessible and effective for farmers worldwide. Through continuous improvement and integration with smarter systems, end effectors like this will help shape the future of automated agriculture.