In the mechanical harvesting process, pineapple fruits are prone to damage. Traditional detection methods struggle to quantitatively assess pineapple damage and often operate at slow speeds. To address these challenges, this paper proposes a pineapple mechanical damage detection method based on machine vision, which segments the damaged region and calculates its area using multiple image processing algorithms. First, both color and depth images of the damaged pineapple are captured using a RealSense depth camera, and their pixel information is aligned. Subsequently, preprocessing techniques such as grayscale conversion, contrast enhancement, and Gaussian denoising are applied to the color images to generate grayscale images with prominent damage features. Next, an image segmentation method that combines thresholding, edge detection, and morphological processing is employed to process the images and output the damage contour images with smoother boundaries. After contour-filling and isolation of the smaller connected regions, a binary image of the damaged area is generated. Finally, a calibration object with a known surface area is used to derive both the depth values and pixel area. By integrating the depth information with the pixel area of the binary image, the damaged area of the pineapple is calculated. The damage detection system was implemented in MATLAB, and the experimental results showed that compared with the actual measured damaged area, the proposed method achieved an average error of 5.67% and an area calculation accuracy of 94.33%, even under the conditions of minimal skin color differences and low image resolution. Compared to traditional manual detection, this approach increases detection speed by over 30 times.

Sunscorch, a common disease in pineapple cultivation, is caused by direct sunlight on the surface of lodging pineapple fruits. To mitigate this issue, pineapple fruits are often bagged or supported with fixing rods. This paper presents the design of an anti-lodging machine, comprising an anti-lodging device and an AGV (Automated Guided Vehicle), and develops a machine vision system to precisely aim the anti-lodging device at the crown buds. The YOLOv5s algorithm is enhanced for detecting crown buds. The improvement of the YOLOv5s algorithm includes the introduction of the attention mechanism CBAM to enhance feature extraction, the addition of BiFPN and GhostNet to strengthen feature fusion and to reduce detection time and computational cost. By transforming pixel coordinates, the relative coordinates between the crown bud and the device are calculated to achieve accurate targeting. The experiments demonstrate that the precision, recall, and map_0.5 of the improved YOLOv5s algorithm are 91.89 %, 92.25 %, 94.01 %, and the detection speed is 15.71 ms/image. The average detection speed is 1.14 ms/image and 9.45 ms/image faster than that of SSD and Faster R-CNN, benefiting from a parameter number only 3/4 and 3/7 of SSD and Faster R-CNN, respectively. The results of contrast experiments indicate that, although the lodging prevention rate of this intelligent anti-lodging method is 4 % lower than the manual binding, its operating speed is 2.7 times faster.

The unmanned pineapple harvester is required to operate efficiently along crop rows. This study presents a novel approach for extracting visual navigation line in pineapple fields using an enhanced YOLOv5 algorithm. The objective is to overcome challenges related to positioning accuracy rate and stability encountered by unmanned pineapple harvesters. The improvements made to the YOLOv5 algorithm include the addition of a small object detection layer in the neck layer. Furthermore, modifications were made to the original loss function to enhance training stability and increase training speed. Experimental results demonstrate that the enhanced YOLOv5 model achieved a 3.62 % increase in training precision rate and a 2.15 % increase in recall rate compared to the original model. With an image resolution of 960 x 1080 pixels, the average detection speed reached 17.35 frames per second (fps). The enhanced YOLOv5 algorithm was utilized to recognize and extract feature points representing pineapple rows. A clustering algorithm was employed to classify these feature points by row, while a modified shortest distance algorithm was applied to fit the pineapple crop row centerline and calculate the optimal navigation line. Field experiments revealed an average extraction accuracy rate of 89.13 % and 85.32 % for sunny and cloudy weather in high-density crops, respectively, with an average accuracy rate of 85.74 %, row recognition accuracy rate of 89.29 % and angle error of 3.54 degrees in different density crops, respectively. This algorithm offers a method for obtaining navigation paths for unmanned pineapple harvesters.

Fruits are susceptible to damage from external loads during harvesting, transport and storage. Fruit damage is determined by its tissue mechanical properties. The aim of this study was to evaluate two theories (Hooke's and Hertz's Laws) on the apparent elastic modulus of pineapple and predict the internal mechanical damage of pineapple under compression. Two multi-scale finite element models (FEMs) were developed to predict the pineapple internal damage. One model, based on Hooke's Law contained three layers, including peel, pulp, and core. Another model, based on Hertz's Law, was a two layers model, including peel and pulp (containing core). The difference of these two FEMs was evaluated in terms of different compression displacements. The results showed that, when the compression displacement <= 10 mm, the Hooke three-layer model was closer to reality. But when the compression displacement >10 mm, the Hertz two-layer model was better in reflecting real pineapple compression. Simulated data confirmed the experimental results and predicted the internal mechanical damage of pineapple. Finite element results indicated that, when a force was applied to the fruit, pulp tissue suffered mechanical damage before peel and core tissue. When the compression level was <= 5% (5.5 mm; 86 N), there was no damage to pulp, core, and peel. But, if compression level >5%, the fruit was damaged. These results indicated that pineapple placed horizontally with an allowable numbers of stackable pineapples must be less than eight. Otherwise, the maximum allowable force of stacked fruit on the lowest one will reach the pulp failure force.

针对我国竹垫人工组装劳动强度大、效率低的问题,提出一种基于改进YOLOv3深度卷积网络的竹垫智能装配方法,通过智能识别、定位、摆放和组装实现卡扣式竹垫的装配生产模式.YOLOv3深度卷积网络的改进包括:首先通过在原YOLOv3的特征提取网络Darknet-53中加入DenseNet密集型连接网络结构,提高特征提取能力;再根据实际检测需求减少网络预测尺度,提升检测精度;最后采用K-means聚类算法对YOLOv3中的初始锚点框参数进行优化,以加快识别速度.笔者还设计了试验机进行装配生产试验.该试验机使用CCD相机采集不同图案的竹垫样品图像,对竹垫样品图像进行识别定位处理,获得目标竹垫零件的位置和颜色信息,然后控制摆放系统对目标竹垫零件进行吸取摆放,再启动传送机构输送竹垫零件,最后通过组装系统实现对目标竹垫零件的组装,实现了竹垫检测、摆放、传送、组装自动化.试验结果表明,视觉系统在GPU和CPU下识别定位竹垫零件的平均时间为16.7和105.3 ms,识别均值平均精度MAP为99.86％,平均组装一行竹垫零件的时间为24.63 s,验证了本方法的可行性.

The detection and localization of pineapple fruit must be successfully conducted to realize intelligent picking. This paper proposed a method for detecting and localizing pineapples in natural environments based on binocular stereo vision and an improved YOLO (You Only Look Once) v3 model. In comparison with the original YOLOv3, the improved algorithm did the following two improvement: the DenseNet was added into the Darknet-53 backbone network to optimize the 13 × 13 and 26 × 26 feature layer, and the SPP-net was fused in the 52 × 52 dimension detection module to strengthen the information representation ability of feature map. A binocular camera acquired left and right images. The left image was then input to the improved YOLOv3 model to obtain the position information of pineapples in the image. Stereo matching and parallax calculation of the target pineapple region were completed via the stereo matching algorithm. Finally, the three-dimensional co-ordinates of pineapples were calculated based on the triangulation principle of binocular stereo vision. A series of experiments were run to compare the detection result of this method against YOLOv3, Faster-RCNN (Region Convolutional Neural Network) and Mobilenet-SSD (Single Shot MultiBox Detector) as per their respective F1 score (Balanced Score) and AP (Average Precision) values. On test set with slight occlusion, the F1 score and AP values of the improved YOLOv3 model were 93.18% and 97.55%, respectively. As the occlusion grew severe, the F1 score and AP values decreased to 89.15% and 91.47%, respectively. The improved YOLOv3 model developed in this study had the best detection effect among all models tested. The binocular stereo vision localization experiment showed an average absolute error of 24.414 mm and average relative error of 1.17% at a distance of 1.7–2.7 m. The proposed method may thus be suitable for picking robots detecting and localizing pineapple fruit in natural environments.

As an excellent algorithm in gesture recognition, three frame difference detection can solve the “double shadow” problem of two frame difference method to a certain extent. But the three frame difference detection will have holes when recognizing gestures, and it cannot adapt to sudden changes in lighting and so on. To solve this problem, the three frame difference gesture recognition method based on mixed bone features was proposed. Firstly, the binary size of the data set was unified, and the background color inter-ference and computation were reduced. Secondly, the three frame difference of mixed bone features was used to detect and track gestures. Finally, the neural network was used for gesture recognition and classification. This method could effectively train hand characteristics, significantly reduce the interference of background on gesture recognition, and improve the recognition efficiency. The experimental results showed that the minimum recognition rate of this method was 93.38% and the maximum was 99.99% in complex background, which could meet the requirement of robustness. This method provides a new idea for gesture recognition in complex image background.

Research on the mechanical harvesting of pineapples is currently in its early stages. The purpose of this study is to provide a design and configure a method for multi-flexible-fingered roller pineapple harvester. Depending on the physical and mechanical characteristics of pineapples, the evaluation function for the critical damage condition of the fruit was established. Our experimental results revealed the optimal parameters for pineapple harvesting were as follows: the rollers of the harvesting mechanism should be inclined at 35°, the left flexible fingers should be 120 mm long, the gap between each of the left flexible fingers should be 30 mm, the length of the right flexible fingers should be 150 mm long, and the gap between each of the right flexible fingers should be 10 mm. The harvesting rate was 85% and the damage rate was 5% in the laboratory; in the natural environment, harvesting rate and damage rate were 78% and 8% respectively, and the harvesting speed was about 1 s per fruit, which demonstrated the harvesting machinery could sufficiently meet the usage demand of pineapple harvesting. In the cases of unsuccessful harvesting, failure resulted from mismatched flexible finger length, fruit size, and harvesting posture and position.

菠萝机械化收获的研究目前处于初期阶段.该研究设计了一种多柔性指滚筒菠萝采收机构,模拟人工采收方式形成折断力矩使水果与植株分离.首先,提出一种通过两组柔性指与菠萝作用形成折断力矩使菠萝花萼处脱落层断裂的采收方法,并测量了该采收方法下脱落层的折断力矩;其次基于柔性指与菠萝作用时产生大变形的特点,确定了以伪刚体模型表征柔性指大变形的方法;根据采收机构模型和菠萝物理、力学特性建立了采收力学模型,确定了菠萝临界损伤条件下的收获评价函数,并基于所建立的模型求出了采收机构的最佳结构参数和收获可能性区域.最后,结合理论分析试制了样机进行台架试验,试验结果表明:当菠萝处于收获可能性区域内时,两滚筒相对倾角为35°,左侧柔性指长度为120 mm、相邻两指间隙为30 mm,右侧柔性指长度为150 mm、相邻两指间隙为10 mm的采收机构能成功采收菠萝,收获率为85％,损伤率为5％,单个果实平均采收时间约为1 s.该研究结果可用于指导双柔性指滚筒菠萝采收机构的开发.

This study designed a sub-pixel precision edge detecting algorithm to enhance contour smoothness. First, the coordinate value of RGB pixel is projected on the space line of R=G=B to obtain gray image. Then, pixel edges are located using a Canny detector. Next, the edge width is thinned to a single pixel using a morphological thinning operation. Finally, sub-pixel-level smooth contours are extracted by interpolation. In this sub-pixel level contour extraction process, a Single-Pixel-Multi-Point interpolation method was developed to enhance edge smoothness and obtain high precision in edge estimation. This method divides edges in a $3\times 3$ pixels block into nine arrangement modes. According to the arrangement of the eight neighborhoods of a centered edge pixel, different locations of interpolated sub-pixel points are calculated by interpolation with Bezier curves. For symmetrically arranged linear edge pixels, this method can be used to determine the exact contour. Experimental results showed that the proposed algorithm can improve the smoothness of image edge contour. As the curvature of the edge increases, the maximum systematic error will increase. For the edge pixel centered in the $3\times 3$ pixels block and two pixels located at the corner of one side, the max systematic error is 0.5 pixel. For two edge pixels aligned in a single row or column with one located at a corner, the max systematic error is 0.25 pixel.

The defect rate of initially produced block bamboo (Bambusoideae) parts is >20 percent. Sorting out these defective parts manually is a highly time-consuming and tedious process. An intelligent sorting system was developed based on machine vision using a Radial Basis Function (RBF) neural network learning algorithm in this study. First, a high-speed charge-coupled device camera was used to obtain a series of images of perfect and defective block bamboo parts. Next, the RBF neural-network learning algorithm was applied to obtain defect characteristics and to locate defective parts moving forward on a conveyor belt. An array of air jets was designed to force defective parts off the belt. Experimental results showed that the average defective part removal rate of the proposed system was 91.7 percent.

研究一种能够自动抓取和自动对心的智能剖竹机上料机械手的设计优化方法.首先,根据生产技术要求设计并确定了机械手的整体结构;然后,利用多变量遗传算法优化了机械臂的尺寸参数,使机械手实际工作空间与期望工作空间重合,且结构长度最小.利用随机概率蒙特卡洛算法模拟的机械手目标空间初步验证了机械手可以抓取到不同直径毛竹.上料试验结果表明该机械手抓取不同直径毛竹的平均对心率为90.45％.抓取成功率为100％.

A nonautomatic bamboo-splitting machine must charge with material and change tools manually. However, manual charging is very dangerous. An intelligent bamboo-splitting machine can feed automatically and change tools intelligently and has broad application prospects. A charging manipulator is an important part of an intelligent bamboo-splitting machine. The size of the manipulator was optimized here using a genetic algorithm. The capture rate, centering rate, and dynamic characteristics of an intelligent bamboo-splitting machine charging manipulator, in which key factors were considered, were experimentally studied. First, three different manipulators, with arm lengths at 210, 220, and 230 mm, were developed. Then, the bamboo materials were divided into three gradients (60–85, 85–110, and 110–135 mm) according to diameter ranges. Accelerators were used to measure the manipulator arm dynamic characteristics, and a high-speed charge-coupled device was used to record the grasping process. Experimental results showed that the manipulator capture rate with an arm length of = 220 mm was as high as 100%, but that of manipulators with arm lengths of = 210 and 230 mm was 96 and 98.67%, respectively. Thus, the manipulator with a 220 mm arm length showed better performance than the other two manipulators. Trend curves of the influence of material diameter on capture time were similar to an exponential function.

剖竹机传统的手工上料方法非常危险,智能剖竹机能够实现自动上料、智能换刀,应用前景广阔.上料机械手是智能剖竹机的重要部件.上料的成功率和对心率是开发毛竹上料机械手必须考虑的因素.文中通过试验研究一种自主开发的气动毛竹上料机械手的抓取成功率、对心率和抓取动态特性.研究过程中根据直径大小将毛竹毛坯分成60～85,85～110和110～135 mm 3个梯度,开发了大臂长度分别为210,220和230 mm的机械手进行试验,并利用加速度传感器测量抓取动态特性.试验结果表明臂长为220 mm的机械手平均抓取成功率为100％,但臂长为210 mm和230 mm的机械手分别只有96％和98.67％.因此,臂长为220 mm的机械手能可靠地抓取各种直径的毛竹.拟合出机械手抓取响应时间-胚料直径关系曲线与指数函数曲线近似,平均加速度-胚料直径曲线与幂函数曲线近似.

针对目前我国柑橘人工采摘作业效率低、成本高、劳动力需求大的问题,设计了一种柑橘梳剪采收末端执行装置,阐述了该装置的结构及工作原理.对果柄剪切的曲柄摇杆机构进行优化设计,优化后机构最小传动角的最大值γmin=48.99°.设计了剪切机构的控制电路,可实现剪切机构的准确启动与停止,并避免堵转.本设计结构简单,与机械臂连接可以构成一个完整的柑橘采收机械,实现柑橘的机械化半自动采收,大幅度提高采收效率,降低了采收成本.

In precision agriculture, identification of fruit in trees is important. Furthermore, it is also essential for estimating the yield, targeting the exact location for a harvesting robot and selectively harvesting the fruit. An elliptical boundary model-based machine vision algorithm was developed to identify immature and mature pomelo fruit in trees. In the proposed solution, the images were converted from RGB space to Y′CbCr space. Then, ordinary least-squares (OLS) was introduced in fitting implicit second order polynomials of elliptical boundary models in the Cr–Cb color space for segmenting immature green fruits, mature green, green partial white, green partial yellow and green partial red fruits. Those elliptical boundary models along with area opening mathematical morphology and diameter thresholding were applied in the identification procedure. The algorithm was tested on a set of 200 validation images acquired under natural illumination conditions. The results of the validation test showed that the total correct identification rate was 93.5%. The total false positive, missed rate, repeated rate and merged rate were equal to 8.2, 6.5, 10.2 and 10.6%, respectively. The proposed method performed better in detecting mature fruit, the color of which is different from green, than in detecting immature green fruit. On average, the segmenting time for 640 × 480 and 1280 × 960 images were 0.134 and 0.200 s, respectively and the total identification time for 640 × 480 and 1280 × 960 images were 0.240 and 0.362 s, respectively.

The shaking functionality of different types of shaking tines for citrus canopy shakers was evaluated. Shaking responses were simulated using ANSYS/LS-DYNA, and a tractor-mounted experimental shaking machine developed for conducting field trials. Straight nylon, bent nylon, straight PVC, and straight steel tines were used as shaking rods and, in experiments, accelerometers placed at branches and fruit to detect vibrational acceleration. Simulation results indicated that tine materials affected the vibrational response of trees and the acceleration peaks of branches and limbs exerted by a steel tine were similar to 2.0 times as high as exerted by a nylon tine. Both simulation and experimental results demonstrated that the tine bending stiffness had a positive effect on the generated acceleration. Tree injuries caused by a straight nylon tine were much less than by other tines. The results suggested that a shaking tine should have high stiffness, but its surface should be soft. These simulation and experiment methods can be used for improving canopy shaker applications.

The overall low detachment percentage in citrus mechanical harvesting is a concerning problem. Studies of the effects of tine-shaking frequency and penetrating depth on fruit detachment for citrus canopy-shaker harvesting have not been reported to date. The objective of this study was to examine how tine-shaking frequency and penetrating depth affect fruit detachment based on simulation and pertinent field experiments for a citrus canopy shaker that inserts a row of shaking tines into the tree canopy. According to evaluation of the branch/stem elasticity, density, and fruit detachment force, a cantilevered limb model, including a periodic shaking force, was constructed to simulate the shaking process in citrus canopy shaking. Simulation results demonstrated a positive correlation between the shaking frequency and maximum stress at the fruit end of the stem, and a 5 Hz shaking frequency found to be sufficient for fruit removal. It was also observed that the penetrating depth ensured that, when shaking spot was close to the junction of the limb and stem, the maximum stress increased at the fruit end of the stem. Field trial results agreed with the simulation results, with both simulation and experiments indicating highly significant effects (p < 0.01) from both the tine-shaking frequency and penetrating depth on fruit removal. The simulation method used here can be utilized for improvements in canopy-shaker applications.

Intelligent detection is a key technology in precision agriculture. As items of different color cluster in different non-overlapping elliptical regions, this study proposed a method for constructing a multi-elliptical boundary model in Cr-Cb co-ordinates to detect citrus fruit and tree trunks in natural light environments. Here, the detected citrus variety was spring sweet tangerine, and the parameters of the elliptical boundary models for detecting these fruit and tree trunks solved by color-space transformation and ellipse fitting. A series of image detection experiments were performed to evaluate the method’s performance. The experimental results showed that the correct and false positive percentages in fruit identification from images were 90.8 and 11.2%, respectively. The number of correctly detected images in distinguishing tree trunks from background was 44 of 50 images.