Mechanical harvesting has been widely used in recent years, due mainly to the ever-increasing fruit demand and the labor shortages. Fortunately, the numerical simulation can be expected to analyze the response of the fruits to the compression, vibration, and impact under various mechanical loads and environment. Among them, finite element analysis (FEA) can be utilized to predict the complex mechanical behavior, thereby providing for an efficient, low-cost, and high-precision solution for fruit harvesting. In this study, a systematic review was presented on the research progress of the fruit harvesting using FEA. Firstly, the current status of the mechanized harvesting was outlined, including the vibration harvesting, harvest assist platforms, and picking robots. Secondly, the research process of the finite element modeling was presented during fruit harvesting. In geometric modeling, 3D scanning and reverse engineering were typically adopted to reconstruct the accurate models of the fruits, branches, and harvesting equipment. The fruit physical properties were acquired to determine the key indicators, such as the elastic modulus, Poisson's ratio, and yield strength after mechanical tests (e.g., uniaxial compression, tensile, and impact tests). These parameters were directly determined the accuracy of the simulation. Mesh generation was often required to balance the computational efficiency and accuracy for the reliability of the simulations. Thirdly, two fields of the FEA application were focused on the fruit rheological properties and the optimization of the harvesting equipment. In rheological properties, the FEA was used to simulate the static stress (e.g., fruit stacking during storage) and dynamic stress (e.g., impact during vibration harvesting or robotic picking). The stress distribution was then obtained to identify the high-risk damage areas. In terms of the equipment optimization, the FEA was employed to adjust the parameters of the vibration harvesting (frequency, amplitude, and excitation direction), in order to reduce the fruit damage. The structural components of the robotic end-effectors (e.g., flexible grippers made of silica gel or rubber) were optimized to minimize the contact stress. The picking paths were also simulated to improve the operational efficiency of the robotic arms. Finally, the current challenges were summarized for the future directions, such as the balance between model simplification and simulation accuracy, insufficient adaptability of the boundary conditions in the dynamic operation scenarios, and the lack of standardized databases for the fruit attributes and working parameters. The efficiency of the FEA was then restricted in the practical applications. Multi-physics coupling, artificial intelligence (AI) and FEA were integrated for the databases of the cross-variety fruit attributes in the future. Greater breakthroughs were found, such as the accurate prediction on the harvesting damage under complex working conditions, intelligent optimization of the adaptive picking equipment, and the performance of the multi-machine collaborative operation. Furthermore, the model iteration and verification were combined with the field test, in order to strengthen the connection between numerical simulation and engineering practice. More solid theoretical and technical support were also provided for the large-scale application of the mechanic harvesting. The fruit industry can transform and then upgrade towards the high efficiency, low damage and intelligence. Numerical simulation can also provide the efficient and feasible solution in fruit harvesting.

Apple grading is one of the critical steps to maximize returns. Among them, the high-quality apples can be sold as fresh at the premium prices, while the lower-grade apples can be used in process applications, such as juice, fresh-cut, or sauce. The apple grading is often conducted indoors using large-scale commercial sorting systems. However, the diseased apples can be mixed with the healthy ones during post-harvest transportation and storage, potentially leading to cross-contamination and economic losses. Apple field sorting equipment can be expected to perform the real-time grading immediately after harvest. The external quality can be identified, such as the fruit size, color, and surface defects, in order to reduce the risk of cross-contamination. Precise apple conveying and rotation can be found in the grading systems using machine vision. However, the limited space and high throughput have confined to capture the complete surface image for each apple in the field. It is also required that the conveying system with a single-row separation, smooth conveying, and uniform rotation. In this study, a variable-pitch spiral roller conveyor was designed to integrate the fruits separation, conveying, and rotation. The kinematic simulations of the variable-pitch spiral roller were conducted using ADAMS software. An iterative approach was employed to monitor the force conditions during apple conveying. A systematic investigation was also made to explore the impact of the different spiral roller speeds on the vibration during apple conveying. The optimal speed range was determined to be 2-3 r/s. A color-based method was also developed to calculate the cumulative coloring rate of the apple surfaces. Six apples were selected with the fruit diameters of (94.0 ± 0.5) mm (large), (84.0 ± 0.5) mm (medium), and (74.0 ± 0.5) mm (small). Each apple surface was uniformly divided into six vertical regions, and then coated with the six watercolor paints (red, white, yellow, blue, green, and purple) to cover the original peel color. The images were captured during the spiral drum conveying using motion image acquisition. The HSV color segmentation and morphological denoising techniques were also employed to extract each color region. Subsequently, the unfolded 2D images of the apple surface were generated for further analysis. A surface coloring rate algorithm was developed to calculate the cumulative coverage rate. A comparison was also made on the area of each colored region in the moving image with the area in the unfolded reference image. This ratio was used to estimate the number of rotations each apple made in the field of the camera’s view. A systematic analysis was then implemented on the relationship between fruit size and rotational behavior. The results show that: 1) A single apple rotated 3 times and 2 times, respectively, when the conveyor shaft operated at the speeds of 2 and 3 r/s, respectively. The imaging system simultaneously captured the image information from 12 apples within 1 second. 2) The horizontal displacement distance for each apple rotation increased by 53%-63% when the spacing increased from 8 to 12 mm. Overall, the spiral drum conveyor reduced the image information leakage, fully meeting the requirements for the fruit separation, uniform rotation, and high-throughput grading under field conditions with limited space. This finding can also provide a strong reference to explore the motion mechanisms of the fruits on the spiral drums.