The pursuit of efficient, non-destructive harvesting of fresh-market apples presents a significant challenge for robotics. An ideal harvesting end effector must fulfill a demanding set of criteria: it must grasp with high efficiency, adapt to the natural variation in fruit size and shape, apply forces gently to prevent bruising, and maintain a simple mechanical and control architecture for reliability in field conditions. While various solutions exist, from vacuum suction to tendon-driven grippers, many introduce complexity in structure, control, or weight. Our research explores a fundamentally different approach inspired by soft robotics. We present the design and experimental analysis of a novel pneumatic apple-picking end effector that utilizes the principle of a flexible silicone membrane expanding under controlled air pressure to conform to and securely hold the fruit.
The core innovation of our end effector lies in its simplicity. Instead of intricate multi-jointed fingers or complex linkage mechanisms, the grasping function is achieved through two inflatable silicone airbags housed within a simple clam-shell structure. This design philosophy directly addresses the need for a gentle, conformable grip. When pressurized, the thin working face of each silicone bladder expands, moulding itself precisely to the contour of the apple’s surface. This distributed, compliant contact area significantly reduces localized pressure points that are the primary cause of fruit damage during mechanical handling. The inherent adaptability of the silicone material allows a single end effector design to accommodate a wide range of apple diameters without requiring complex sensing and control algorithms for finger positioning. The physical structure is lightweight and compact, comprising only the rigid outer shell, the flexible bladders, and the necessary pneumatic fittings, making it highly suitable for integration with a robotic manipulator arm without imposing excessive payload demands.
Developing a functional harvesting end effector requires a robust and responsive control system. The pneumatic system we implemented is designed for precise operation. It consists of an air pump, a filter/regulator unit to provide clean air at a set pressure, and a network of solenoid valves controlled by a microcontroller (Arduino). A pressure sensor provides real-time feedback on the state of the silicone airbags. The operational sequence is straightforward: upon receiving a harvest signal (e.g., from a vision system), the microcontroller activates the inflation valve, filling the bladders. The pressure sensor monitors the internal pressure until a pre-determined threshold is reached, at which point the inflation valve closes and a separate lock valve engages to maintain a constant pressure, ensuring a stable grip during the detachment and transfer motions. After the fruit is placed in a collection bin, a deflation valve opens to vent the air, releasing the apple. This closed-loop pressure control is crucial for ensuring consistent performance regardless of minor leaks or temperature variations.
Evaluating the performance of any harvesting end effector hinges on two critical mechanical metrics: the holding force and the clamping pressure. The holding force is the maximum axial force the gripper can resist before the fruit slips out; it must exceed the natural pedicel (stem) detachment force of the apple. The clamping pressure (or force per unit area) is the normal force exerted by the gripper’s fingers onto the fruit’s skin; this must be kept below the threshold that causes bruising. We conducted systematic experiments to characterize our pneumatic end effector against these metrics. Artificial apple models with diameters of 65.5 mm and 82.4 mm were used to represent a typical size range. The internal pressure of the silicone bladders was varied across five levels: 15.0, 17.5, 20.0, 22.5, and 25.0 kPa. For holding force tests, a force sensor measured the peak force required to pull the apple vertically from the energized gripper. The results clearly show a strong positive correlation between applied air pressure and the generated holding force, which can be modeled by a linear relationship for the operational range:
$$ F_h = k P + c $$
where $F_h$ is the holding force, $P$ is the applied air pressure, and $k$ and $c$ are constants dependent on the fruit size and silicone material properties. The data is summarized in the table below:
| Fruit Diameter (mm) | Pressure (kPa) | Holding Force – Test 1 (N) | Holding Force – Test 2 (N) | Holding Force – Test 3 (N) | Mean ± Std Dev (N) |
|---|---|---|---|---|---|
| 65.5 | 15.0 | 22.5 | 25.1 | 20.8 | 22.8 ± 1.79 |
| 17.5 | 28.3 | 30.5 | 26.7 | 28.5 ± 1.56 | |
| 20.0 | 34.2 | 31.8 | 32.1 | 32.7 ± 1.02 | |
| 22.5 | 37.5 | 35.0 | 38.9 | 37.1 ± 1.65 | |
| 25.0 | 40.1 | 42.3 | 39.5 | 40.6 ± 1.22 | |
| 82.4 | 15.0 | 28.7 | 32.1 | 30.5 | 30.4 ± 1.37 |
| 17.5 | 34.8 | 33.2 | 36.0 | 34.7 ± 1.20 | |
| 20.0 | 38.2 | 35.1 | 36.7 | 36.7 ± 1.29 | |
| 22.5 | 41.5 | 43.8 | 39.9 | 41.7 ± 1.67 | |
| 25.0 | 45.0 | 47.1 | 44.5 | 45.5 ± 1.16 |
To measure the actual clamping force exerted on the fruit’s surface, we employed an array of thin-film force sensors attached to the artificial apples. Prior to testing, each sensor was meticulously calibrated. The calibration process involved applying known masses and recording the corresponding output voltage. The relationship was highly linear, as described by the equation derived from the sensor’s specifications:
$$ U_o = \left(\frac{F}{F_{max}}\right) \times 5.0 $$
where $U_o$ is the output voltage (V), $F$ is the applied force (g), and $F_{max}$ is the sensor’s maximum capacity (1500g). Rearranging gives the force calculation formula:
$$ F = \frac{U_o \times F_{max}}{5.0} = \frac{U_o \times 1500}{5.0} = 300 \times U_o $$
The calibration data for six representative sensors confirmed excellent linearity, with coefficients of determination ($R^2$) exceeding 0.998. With calibrated sensors, we measured the clamping force at three key pressure levels (15, 20, 25 kPa) on both fruit sizes. The maximum force reading from the six sensors during each trial was recorded. The results, summarized in the following table, indicate that clamping force also increases with pressure and is generally higher for larger fruit due to greater contact area with the expanding silicone. However, the absolute values remain very low, in the range of 0.5 to 1.2 Newtons, which is promising for non-destructive handling.
| Fruit Diameter (mm) | Pressure (kPa) | Clamping Force – Test 1 (N) | Clamping Force – Test 2 (N) | Clamping Force – Test 3 (N) | Mean ± Std Dev (N) |
|---|---|---|---|---|---|
| 65.5 | 15 | 0.51 | 0.59 | 0.54 | 0.55 ± 0.033 |
| 20 | 0.97 | 0.98 | 0.98 | 0.98 ± 0.005 | |
| 25 | 1.12 | 1.09 | 1.11 | 1.11 ± 0.012 | |
| 82.4 | 15 | 0.95 | 0.96 | 0.98 | 0.96 ± 0.013 |
| 20 | 1.06 | 1.10 | 1.07 | 1.07 ± 0.017 | |
| 25 | 1.16 | 1.17 | 1.19 | 1.17 ± 0.012 |
The experimental data provides a clear basis for selecting the optimal operational parameters for our pneumatic end effector. The primary constraint is that the holding force must reliably exceed the apple’s stem detachment force, which is typically around 30 N. From the holding force table, it is evident that for the smaller (65.5 mm) fruit, a pressure of 20.0 kPa yields a mean holding force of 32.7 N, which meets the requirement. At 17.5 kPa, the force is borderline and less reliable. Simultaneously, we aim to minimize the clamping force to reduce any risk of bruising. The clamping force data shows that at 20.0 kPa, the force on the fruit is approximately 1.0 N, which is very low. Increasing pressure to 25.0 kPa provides a larger safety margin for holding force but also increases the clamping force by about 10-15%. Therefore, 20.0 kPa is identified as the optimal working pressure, striking the best balance between securing the fruit firmly and handling it gently. This pressure setting ensures the end effector is functional across the tested size range while prioritizing fruit integrity.
In conclusion, the design and testing of this pneumatic apple harvesting end effector demonstrate a highly effective and simplified approach to a complex agricultural robotics problem. By leveraging the compliant nature of pressurized silicone, the end effector achieves adaptive, form-fitting grasps on apples of varying sizes. The systematic evaluation of holding and clamping forces confirmed that the device can securely detach fruit while applying minimal contact pressure. The selected operational pressure of 20 kPa ensures reliable performance aligned with the physiological constraints of the fruit. This pneumatic end effector exemplifies how principles from soft robotics can be translated into practical, field-worthy agricultural tools. Its compact structure, simple control paradigm, and inherently gentle gripping action offer a compelling alternative to more complex mechanical grippers, providing a new and promising direction for the development of non-destructive harvesting end effectors.