Hui Zhang

Tomato harvesting remains one of the most labour‐intensive operations in horticultural production, particularly in protected cultivation systems where fruit maturity, cluster structure, canopy occlusion and damage‐sensitive handling strongly influence marketable yield and postharvest quality. Robotic harvesting has been widely investigated as a potential solution, but its practical adoption depends not only on advances in machine vision, manipulation and end‐effector design, but also on how well these technologies respond to crop‐specific biological and production‐system constraints. Using a structured scoping approach covering publications from 1990 to 2026, 54 studies were included in the qualitative synthesis. This review evaluates progress in fruit detection and maturity assessment, three‐dimensional localisation, navigation and motion planning, low‐damage detachment, end‐effector design, integrated harvesting performance, fruit quality preservation and adaptation of production systems for robotic operation. Two principal outputs are proposed: an integrated metric framework linking engineering performance with horticultural outcomes and a recommended minimum reporting dataset covering crop and canopy conditions, robotic configuration, performance definitions, fruit damage, postharvest quality and commercial indicators. Together, these outputs are intended to improve cross‐study comparability and support the transition from experimental prototypes towards robot‐ready and commercially relevant tomato‐production systems.

Early and accurate detection of mechanical damage in prunes is crucial for preserving postharvest quality and enabling automated sorting. This study proposes a practical and reproducible method for identifying cumulative bruising in prunes using visible–near-infrared (Vis–NIR) reflectance spectroscopy coupled with machine learning techniques. A self-developed impact simulation device was designed to induce progressive damage under controlled energy levels, simulating realistic postharvest handling conditions. Spectral data were collected from the equatorial region of each fruit and processed using a hybrid modeling framework comprising continuous wavelet transform (CWT) for spectral enhancement, uninformative variable elimination (UVE) for optimal wavelength selection, and support vector machine (SVM) for classification. The proposed CWT-UVE-SVM model achieved an overall classification accuracy of 93.22%, successfully distinguishing intact, mildly bruised, and cumulatively damaged samples. Notably, the results revealed nonlinear reflectance variations in the near-infrared region associated with repeated low-energy impacts, highlighting the capacity of spectral response patterns to capture progressive physiological changes. This research not only advances nondestructive detection methods for prune grading but also provides a scalable modeling strategy for cumulative mechanical damage assessment in soft horticultural products.

Near‐Infrared (NIR) Spectroscopy has become a pivotal non‐destructive technology in agricultural production and fruit quality assessment, particularly in the pear industry. This review comprehensively examines recent advancements in NIR Spectroscopy for evaluating critical quality parameters such as Soluble Solids Content (SSC), firmness, and moisture content, which are essential for determining optimal harvest timing, improving storage practices, and extending shelf life. The evolution of portable NIR spectrometers is explored, emphasizing their increasing suitability for real‐time, in‐field applications while addressing challenges such as calibration variability and environmental factors. Additionally, the integration of NIR spectroscopy with imaging technologies like Hyperspectral Imaging (HSI) and Multispectral Imaging (MSI) is highlighted for their potential to enhance the precision and scope of fruit quality assessment. The review further considers incorporating low‐cost technologies, such as electronic noses and machine vision, with NIR spectroscopy to create cost‐effective and robust multi‐sensor platforms. By synthesizing these developments, the review identifies gaps in current methodologies, such as limited model generalization across diverse pear varieties, and proposes solutions like data fusion and advanced machine learning approaches. These strategies are poised to improve model adaptability and accuracy, ensuring NIR spectroscopy remains at the forefront of pear quality assessment. Future research directions emphasize the need for integrating cutting‐edge technologies with NIR spectroscopy to achieve more sustainable and efficient agricultural practices.

X-ray imaging technology, with its exceptional capability for visualizing internal structures, plays a critical role in the nondestructive testing of agricultural products such as walnuts and almonds. To address the current challenges of high noise levels in walnut X-ray images and the difficulty in preserving critical image details after denoising, this study proposes an improved Restormer denoising model for clearer processing of walnut X-ray images. First, batch normalization layers were introduced into the Multi-Dconv Head Transposed Attention mechanism to enhance the model's understanding of image features. Second, a vertical Total Variation loss function was integrated to suppress high noise levels and extract clearer image features. Experimental results demonstrated that the improved Restormer model achieved a PSNR of 37.30, an SSIM of 0.9358, and an information entropy of 6.5600, representing improvements of 0.16 dB, 0.0002, and 0.0014, respectively, compared to the original Restormer model. In terms of PSNR, SSIM, information entropy, and visual quality, the proposed model outperformed six commonly used network models, including DnCNN and Uformer, delivering clearer images with richer details. Furthermore, on the local dataset, the model also exhibited excellent processing performance, generalization ability, and stability, making it a highly effective solution for walnut X-ray image denoising. The research results can offer a theoretical basis for the efficient image denoising method on walnut X-ray and provide valuable insights for denoising research in other imaging fields.

The purpose of this study was to achieve non-destructive detection of the internal defects of in-shell walnuts using X-ray radiography technology based on improved Faster R-CNN network model. First, the FPN structure was added to the feature-extraction layer to extract richer image information. Then, ROI Align was used instead of ROI Pooling for eliminating the localization bias problem caused by the quantization operation. Finally, the Softer-NMS module was introduced to the final regression layer with the predicted bounding box for improving the localization accuracy of the candidate boxes. The results of the study indicated that the proposed network model can effectively identify internal defects of in-shell walnuts. Specifically, the discrimination accuracies of the in-shell sound, shriveled, and empty-shell walnuts were 96.14%, 91.72%, and 94.80%, respectively, and the highest overall accuracy was 94.22%. Compared to the original Faster R-CNN network model, the improved Faster R-CNN model achieved an increase of 5.86% in mAP and 5.65% in F₁-value. Consequently, the proposed method can be applied for the in-shell walnuts with shriveled and empty-shell defects.

The purpose of this work was to achieve non-destructive detection of the internal defects of in-shell walnuts using X-ray radiography technology based on improved Faster R-CNN network model. First, the FPN structure was added to the feature extraction layer to extract richer image information. Then ROI Align was used to instead of ROI Pooling for eliminating the localization bias problem caused by quantization operation. Finally, the Softer-NMS module was introduced to the final regression layer with the predicted bounding box for improving the localization accuracy of the candidate boxes. The results indicated that the internal defects in intact walnuts with shells could be identified effectively using the proposed improved network model in this study. Specifically, the discrimination accuracy of the in-shell sound, shriveled and empty-shell walnuts were 96.14%, 91.72% and 94.80% respectively, and the highest overall accuracy can reach 94.22 %. Contrasted with original Faster R-CNN network model, the mAP and F1-value of the improved Faster R-CNN model increased by 5.86% and 5.65%, respectively. Consequently, the proposed method can be applied for the in-shell walnuts with shriveled and empty-shell defects.

Conventional racket sports training highly relies on coaches' knowledge and experience, leading to biases in the guidance. To solve this problem, smart wearable devices based on Internet of Things technology (IoT) have been extensively investigated to support data-driven training. Considerable studies introduced methods to extract valuable information from the sensor data collected by IoT devices. However, the information cannot provide actionable insights for coaches due to the large data volume and high data dimensions. We proposed an IoT + VA framework, Tac-Trainer, to integrate the sensor data, the information, and coaches' knowledge to facilitate racket sports training. Tac-Trainer consists of four components: device configuration, data interpretation, training optimization, and result visualization. These components collect trainees' kinematic data through IoT devices, transform the data into attributes and indicators, generate training suggestions, and provide an interactive visualization interface for exploration, respectively. We further discuss new research opportunities and challenges inspired by our work from two perspectives, VA for IoT and IoT for VA.