基于逆运动学降维求解与YOLO v4的果实采摘系统研究

为提高采摘设备的执行效率，采用六自由度机械臂、树莓派、Android手机端和服务器设计了一种智能果实采摘系统，该系统可自动识别不同种类的水果，并实现自动采摘，可通过手机端远程控制采摘设备的起始和停止，并远程查看实时采摘视频。提出通过降低自由度和使用二维坐标系来实现三维坐标系中机械臂逆运动学的求解过程，从而避免了大量的矩阵运算，使机械臂逆运动学求解过程更加简捷。利用Matlab中的Robotic Toolbox进行机械臂三维建模仿真，验证了降维求解的可行性。在果实采摘流程中，为了使机械臂运动轨迹更加稳定与协调，采用五项式插值法对机械臂进行运动轨迹规划控制。基于Darknet深度学习框架的YOLO v4目标检测识别算法进行果实目标检测和像素定位，在Ubuntu 19.10操作系统中使用2000幅图像作为训练集，分别对不同种类的果实进行识别模型训练，在GPU环境下进行测试，结果表明，每种果实识别的准确率均在94%以上，单次果实采摘的时间约为17s。经过实际测试，该系统具有良好的稳定性、实时性以及对果实采摘的准确性。     

Characterizing Spatial Variability of Soil Textural Fractions and Fractal Parameters Derived from Particle Size Distributions

Soil particle size distribution (PSD) is a fundamental physical property affecting other soil properties. Characterizing spatial variability of soil texture is very important in environmental research. The objectives of this work were: 1) to partition PSD of 75 soil samples, collected from a flat field in the University of Guilan, Iran, into two scaling domains using a piecewise fractal model to evaluate the relationships between fractal dimensions of scaling domains and soil clay, silt, and sand fractions and 2) to assess the potential of fractal parameters as an index used in a geostatistical approach reflecting the spatial variability of soil texture. Features of PSD of soil samples were studied using fractal geometry, and geostatistical techniques were used to characterize the spatial variability of fractal and soil textural parameters. There were two scaling domains for the PSD of soil samples. The fractal dimensions of these two scaling domains (D₁ and D₂) were then used to characterize different ranges of soil particle sizes and their relationships to the soil textural parameters. There was a positive correlation between D₁ and clay content (R² = 0.924), a negative correlation between D₁ and silt content (R² = 0.801), and a negative correlation between D₂ and sand content (R² = 0.913). The geometric mean diameter of soil particles had a negative correlation with D₁ (R² = 0.569) and D₂ (R² = 0.682). Semivariograms of fractal dimensions and soil textural parameters were calculated and the maps of spatial variation of D₁ and D₂ and soil PSD parameters were provided using ordinary kriging. The results showed that there were also spatial correlations between D₁ and D₂ and particle size fractions. According to the semivariogram models and validation parameters, the fractal parameters had powerful spatial structure and could better describe the spatial variability of soil texture.     

Exogenous application of humic acid and salicylic acid to alleviate seedling drought stress in two corn (Zea mays L.) hybrids

Efficacy of 1?mM humic (HA) and salicylic (SC) acids on SC 260 and SC 705 corn seedlings to alleviate drought stress via polyethylene glycol was studied via hydroponics at Shiraz University, Iran in a factorial, randomized design, with four replicates each. Under stress, SC 260 had higher electrolyte leakage compared to SC 705, and exogenous application of HA combined with SA decreased SC 705 electrolyte leakage. As a general trend, photosynthetic pigment content, relative water content, root and shoot length, mean number and diameter of central and peripheral root metaxylem, and K⁺ accumulation were higher in SC 705 treated with HA and SA compared to SC 260. Application of HA with SA could be an effective and low cost approach to ensure seedling establishment and plant growth in fields affected by soil drought in the early season, especially for the SC 705 corn hybrid in semi-arid regions.     

土地利用方式对雅江中游土壤理化性质及 颗粒分形特征的影响

采取野外调查和室内试验相结合方法,分析了雅江中游4种土地利用方式基本理化性质,并运用土壤颗粒体积分形维数模型研究了不同土地利用方式土壤颗粒分形特征及与土壤粒径、性状关系。结果表明:(1)不同土地利用方式0~30 cm土壤理化性质差异显著(P0.05),土壤容重表现为稀疏灌丛农耕地荒草地人工林地,随着土壤层次增加,土壤容重呈增大趋势。总孔隙度、毛管孔隙度、饱和持水率、毛管持水率均与容重规律相反;(2)4种土地利用方式主要以粉粒、砂粒为主,黏粒含量仅占到0.36%~4.27%,土壤质地偏砂性;(3)0~30 cm土壤颗粒体积分形维数平均值的顺序为:农耕地荒草地人工林地稀疏灌丛。人工林地具有较好的土壤结构,相对疏松、通透性好。而稀疏灌丛受到砾石影响,分形维数最小,但在保水保肥能力上最弱;(4)土壤颗粒体积分形维数与砂粒体积百分含量呈极显著负相关(r=–0.873,P0.01);与粉粒体积百分含量呈极显著正相关(r=0.860,P0.01);与黏粒体积百分含量呈极显著正相关(r=0.913,P0.01)。这表明,一定程度上,研究区分形维数可作为评估水土流失的指标;(5)土壤颗粒体积分形维数与各土壤单一物理性质具有不同程度相关性,其中与总孔隙度、毛管孔隙度、饱和含水量、毛管含水量均为正相关关系,而与容重、有机质呈负相关关系。     

艾比湖湿地不同盐渍化土壤粒度组成及可蚀性研究

为了研究内陆干旱区不同盐渍化土壤的粒径组成及其可蚀性的空间分布,在不同盐渍化程度上探讨其与土壤中盐分含量的相互关系,本研究以新疆维吾尔自治区艾比湖湿地为研究靶区,利用2015年5月获取的66个表层土壤样品,对土壤的粒度组成、含盐量、分形维数和可蚀性因子K值进行了研究。结果表明:样品中盐土占比达到59.09%,平均含盐量达131.59 g/kg,研究区内盐渍化现象既普遍又严重;土壤颗粒以粉粒和砂粒为主,随着盐渍化程度的加剧黏粒含量不断增加,砂粒含量则相反;土壤颗粒的分形维数D值介于2.14~2.60,平均值为2.45;随着盐渍化程度加剧,分形维数逐渐变大,K值也随之逐渐增大,盐土的K值最大达0.093;土壤可蚀性因子K值与含盐量的相关性较高(r=0.596**),在一定程度上可以说明,土壤的含盐量水平越高,K值越大,越容易被侵蚀。     

地貌发育阶段判识方法研究——以泥河沟流域和枣子沟流域为例

为更加准确判识小流域地貌发育阶段，以黄土高原综合治理小流域——陕西省淳化县泥河沟流域和乾县枣子沟流域为例，分别应用盒分维方法和Horton分维方法进行水系分维值的计算，并利用面积高程积分和地貌信息熵进行验证。结果表明，泥河沟和枣子沟流域水系盒分维值分别为1.886和1.827。根据Strahler曲线积分值和地貌信息熵方法验证结果表明，盒分维方法计算的水系分维值对流域地貌发育阶段的判识更加准确、可靠；这2个流域地貌处于侵蚀发育阶段的壮龄期，其生态敏感性整体上处于较低水平，生态环境抗干扰能力增强并趋于更加稳定的状态。     

杨树低效林下土壤粒径分布与分形特征

[目的]研究鄂尔多斯地区杨树低效林下土壤粒径分布与分形特征,为该地区杨树低效林的成因分析及改造提供基础依据,并为杨树人工林的科学管理提供参考与决策支持。[方法]使用Mastersizer3000激光衍射粒度分析仪测量土壤粒径,采用土壤颗粒体积分形维数模型计算分形维数D值,并测定土壤养分。[结果]①鄂尔多斯地区杨树低效林下土壤颗粒组成主要以细砂粒(42.33%)、极细砂粒(15.72%)、中砂粒(19.77%)为主;主要土壤质地类型为砂土;土壤颗粒的分形维数D值在1.222 5～2.204 5范围内,处于较低水平。②土壤颗粒分形维数D值与粉粒、极细砂粒、细砂粒含量呈显著正相关关系,与中砂粒、粗砂粒、极粗砂粒呈显著负相关关系。③随着土层加深,粒径250μm的土壤颗粒含量增加,粒径250μm的土壤颗粒含量减少;土壤颗粒分形维数D值随土层加深而增大,增大幅度不大。④土壤颗粒分形维数D值与土壤有机质、全氮、速效氮、速效磷含量的关系均未达到显著水平,土壤各养分含量的变化对分形维数D值没有显著影响。[结论]鄂尔多斯地区杨树低效林下土壤颗粒组成以砂粒为主,土壤分形维数较低,土壤结构极不均匀,这样的土壤结构对杨树的正常生长发育极为不利,不良的土壤结构也在一定程度上导致了低效林的形成,因此在对杨树低效林进行预防和改造时也要改善林下土壤结构。     

北美规格材机械分等综述

规格材的机械分等方法是北美一种很重要的分等体系,即采用机械应力测定设备对木材进行非破坏性试验,按测定的木材弯曲强度和弹性模量确定木材的材质等级。用于机械分等规格材的分选参数可以因不同工厂机械的具体情况而进行调整,较目测分等更能适应现代化木材工业的发展要求。文中从分等方法的诞生、机械分等与目测分等相互关系、机械分等方法和标准、分等机构设置等方面介绍了北美规格材机械分等,在此基础上分析了中国规格材分等的现状并对中国规格材分等提出了参考建议。     

Impacts of fractal features of soil on moisture infiltration capacity of typical stands in Jinyun mountain of Chongqing city

The soil structure was expressed with fractal dimensions of particle size distribution (PSD), aggregate size distribution (ASD), and soil pore size distribution (SPD). The effect of soil fractal features on soil infiltration velocity and process was studied. The result of the fractal feature shows that fractal dimensions of PSD are obviously greater than those of ASD and SPD, and in different soil genetic horizons, the fractal dimension of ASD has the greatest variability, and shows a downtrend on the top-to-bottom genetic horizon. According to the soil infiltration process curve, the infiltration process was divided into three phases: (1) the initial phase (0–5 min), (2) the transition phase (5–30 min), and (3) the stable phase (30–180 min). In the initial phase of infiltration, the soil structure of soil genetic horizon A was the major influencing factor; in the transition phase of infiltration, the pore distribution of soil horizon AB and soil structure of horizon B were the major influencing factors; in the stable phase of infiltration, the soil structure of horizon C was the major influencing factor to the infiltration velocity. Soil infiltration process is influenced comprehensively by soil PSD, ASD, and SPD. In the overall soil water infiltration, the infiltration in shrub forest land was much faster than that in other land uses, and in the initial infiltration phase, arable land soil infiltration was much faster than that in forest land, and in the stable infiltration phase, the infiltration velocity in forest land was faster than that in arable land. __________ Translated from Journal of Beijing Forestry University, 2006, 28(3): 73–78 [译自: 北京林业大学学报]     

Contributions of Extractives and Lignin to the Fractal Geometry of Internal Wood Surfaces

Internal wood surfaces can be treated as fractals, which are between Euclidean geometry and complete randomness. The fractal dimension Dfs is very informative in investigating the roughness of the internal surfaces of wood. In this study, the water sorption isotherms, including adsorption and desorption isotherm, of untreated, benzene-alcohol extracted and delignified (after benzene-alcohol extracted) spruce (Cuninghamia lanceolata) were measured at 30℃. On the basis of these isotherms, the Dfs values were calculated by FHH equation, which is based on multimolecular sorption. The results showed that both groups of Dfs values (respectively calculated from adsorption and desorption isotherms) of untreated, benzene-alcohol extracted and delignified wood have same order, that is, untreated > benzene-alcohol extracted - delignified. Therefore, the conclusion can be made that the benzene-alcohol extractives have significant contribution to the fractal geometry of internal wood surfaces. Lignin also has influ