基于导频跟踪的果蔬采摘机器人夜间识别方法研究

为了提高果实采摘机器人在夜间作业的工作效率,提出了一种适用于果实采摘机器人机械手夜间果实定位识别的方法,引入了导频追踪技术,并在此基础上提出了一种基于直接判决的快时变信道估计方法。该方法采用了直接判决算法进行信道估计,根据夜间果实图像的二值化处理阈值,标定发出指令的信道,从而对是否采摘进行快速判决,提高了机器人的工作效率。为了验证该方法的有效性和可靠性,对机械手的夜间作业的效率和准确性进行了测试。通过测试发现:该方法可以快速地对果实进行定位,并通过校正,其有效地降低夜间误操作率,为果实采摘机器人的研究,提供了一种新的可参考的导频技术方法。     

一种多机械手编码控制系统的果蔬采摘机器人设计

为了提高果蔬的采摘效率,对果蔬采摘机器人进行了改进,设计了一种新的多机械手的编码控制采摘机器人。通过对机器人功能和结构的设计,使机器人具有了利用机器视觉技术的图像边缘提取来划分每个机械手的作业区间的功能,并可以利用编码器对每个机械手进行编码,从而完成多机械手的协同作业。对机器人的采摘作业性能进行了测试,首先利用机器视觉模块完成了苹果采摘区间的划分,并预设了每个机械手的采摘作业轨迹,利用编码器对预设轨迹进行了追踪。通过测试发现:机器人多机械手的实际追踪轨迹和预设轨迹的误差很小,满足设计需求,多机械手的协同采摘平均速度可以达到80个/min以上,具有高效的果蔬采摘性能。     

基于改进动态扩展和位置服务的农产品追溯系统优化

农产品追溯系统是解决食品安全问题的重要手段,目前常见的追溯系统往往具有适用品种单一、精度不高、可应用性弱等特点,无法满足大规模推广应用的需要。该文以建立适用面广、性能高效的农产品追溯系统为目标,分析了常见追溯系统的软件框架与追溯信息数据结构,重点研究可支撑多品种追溯的动态扩展数据模型优化技术,同时以提升追溯精度、降低追溯成本为目标研究了基于位置服务(location based service,LBS)的追溯精度提升技术。针对追溯系统适用范围窄的问题,分析了追溯信息指标模型的数据结构特征,对比了常见的4种动态扩展数据模型,采用EAV与XML数据模型相结合的方法,通过可视化操作界面实现了不同品种追溯信息的动态扩展,通过性能分析发现该模型较其他两种模型时间缩短65%,查询效率提升了175%;针对追溯信息中的产地环境信息,分析了各类环境数据的可用来源和数据特征,制订了数据匹配的最小空间单元和实现方法,运用LBS技术实现了产地环境信息的自动获取,通过追溯系统之间的精度对比发现,该技术显著提高了追溯系统的宽度、深度、精确度,从而整体提升了系统追溯精度并降低了数据采集成本。该系统已在青海省应用,结果表明该系统较其他系统在适用范围、系统性能、追溯精度与追溯成本上具有明显优势。     

Multi-target tracking algorithm based on adaptive sampling interval in wireless sensor networks

Multi-target tracking is a hot topic of current research on wireless sensor networks (WSN). Based on adaptive sampling interval, we propose a multi-target tracking algorithm in order to save energy consumption and prevent tracking lost for WSN. We contrast the targets moving model by using the position metadata, and predicte the targets moving status based on extended Kalman filter (EKF).we adopt the probability density function (PDF) of the estimated targets to establish the tracking cluster. By defining the tracking center, we use Markov distance to quantify the election process of the main node (MN). We comput targets impact strength through the targets importance and the distance to MN node, and then use it to build tracking algorithm. We do the simulation experiment based on MATLAB, and the experiment results show that the proposed algorithm can accurate predict the trajectory of the targets, and adjust the sampling interval while the targets were moving. By analyzing the experiments data, we know that the proposed algorithm can improve the tracking precision and save the energy consumption of WSN obviously.     

Optimal placement of the region of interest for bolus tracking on brain computed tomography angiography in Beagle dogs

This study aimed to determine the optimal placement of the region of interest (ROI) among four anatomical sites—pulmonary artery (PA), pulmonary vein (PV), aortic arch (AA), and carotid artery (CA)—in computed tomography (CT) brain angiography with automatic bolus tracking in healthy beagle dogs. Six beagles were included, and CT brain angiography was performed four times for each dog, to cover each ROI. The scan parameters, amount, and injection rate of the contrast medium were the same. The major intracranial arteries were selected for quantitative and qualitative evaluation: caudal cerebellar artery (CcA), basilar artery (BA), rostral cerebellar artery (RcA), caudal cerebral artery (CCA), middle cerebral artery (MCA), and rostral cerebral artery (RCA). Quantitative evaluation showed significantly higher CT attenuation values for the RcA, CCA, and MCA in the PA group and RcA and MCA in the PV group than in the CA group. Qualitative analysis revealed significantly higher scores for the BA, CCA, and MCA in the PA and PV groups than in the CA group. Venous contamination did not differ significantly among the ROIs, but the mean scores of the AA and CA groups were higher than those of the PA and PV groups. CT brain angiography using bolus tracking in the beagle dogs showed that the ROI should be placed at the PA or PV rather than at the CA for optimal images with strong contrast enhancement of the BA, RcA, CCA, and MCA and minimal venous contamination.     

Effects of contour hedgerows on water and soil conservation,crop productivity and nutrient budget for slope farmland in the Three Gorges Region (TGR) of China

Soil erosion has long been recognized as a major environmental problem in the Three Gorges Region (TGR) where slope farming is commonly practiced but the local topography is hilly. In consideration of the poor socioeconomic position of local farmers, low cost hedgerows had been introduced as a soil conservation measure to the TGR in the late 1980s. A collaborative research programme was initiated by the Chinese University of Hong Kong, the Chinese Academy of Science, the Huazhong Agricultural University, and the Bureau of Soil and Water Conservation of Zigui County to study the potential of adopting hedgerows in the TGR. Six experimental plots (10 m × 2 m, gradient = 25°) were constructed at Zigui County, Hubei Province to study effects of hedgerows on erosion, nutrient loss and crop productivity. Results indicated that there were significant relationships between rainfall and runoff, and rainfall and soil loss, respectively. Conventional slope farming could not be considered a sustainable agricultural practice because it resulted in severe erosion and low crop yield. Hedgerows per se seemed not to be effective in reducing soil loss and boosting crop productivity, but performances could be greatly improved when they combined with the use of fertilizers. Current farming and fertilization practices, however, generally did not meet N demands of crops. Results and findings of this paper will contribute towards a technical reference for the promotion and adoption of hedgerows in the TGR.     

水稻穴直播机自主驾驶控制系统设计

为满足水稻穴直播机自主作业的高性能需求,对穴直播机的自主驾驶系统进行了自主化设计与验证。根据洋马VP6型穴直播机的结构特点,对穴直播机自主驾驶进行机电改造,设计了自主驾驶总控制器和CAN总线化电动转向、油门及挡位的分控制系统,满足了自动驾驶的功能要求。经过分析穴直播机的运动学模型及控制模型,分别建立了外环轨迹跟踪和内环转向角度跟踪的双闭环控制算法策略。经过反复试验,确定合适的控制器参数为k_p=1、k_d=3.5。水田环境的试验结果表明:自动驾驶控制系统达到了横向偏差的绝对值不大于8 cm的跟踪效果。     

基于趋近律滑模控制的智能车辆轨迹跟踪研究

针对目前已有智能车辆轨迹跟踪控制存在跟踪精度低、鲁棒性弱等问题,结合滑模控制响应迅速、抗干扰能力强的优点,提出一种基于趋近律的滑模控制方法。提出的趋近律通过特殊幂次函数和反双曲正弦函数的组合,提高了系统状态的趋近速度并且平滑和限制了抖振现象,可实现控制车辆平顺快速跟踪参考轨迹。在Simulink软件上搭建了车辆运动学模型并进行轨迹跟踪仿真实验,通过与双幂次趋近律滑模控制进行对比验证了控制效果。仿真实验结果表明,相对于双幂次趋近律滑模控制,提出的趋近律滑模控制的车辆能更快地跟踪到参考轨迹,横向和纵向误差收敛速度明显增快,航向角抖振现象减弱,系统具有更快的趋近速度,并且抖振现象被削弱。     

基于非线性模型的农用车路径跟踪控制器设计与试验

为提高农用车辆路径跟踪性能,提出一种基于非线性模型预测的路径跟踪控制方法。该方法将路径跟踪问题转换为求解满足速度、转角约束的最优值问题。首先将农用车的非线性运动学模型进行离散化推出递推模型,作为控制器的预测方程;然后建立以农用车运动学模型控制量为状态量的目标函数,设计各个变量的约束条件,把预测方程代入目标函数将其转化为基于递推序列的二次规划法响应问题,在此基础上进行梯度计算解决非线性的约束优化;最后,利用实时反馈与滚动优化实现控制器的闭环校正;重复以上过程,完成预测控制。Matlab仿真结果表明:非线性模型预测控制器能够实现对所设计路径的有效跟踪。相对应的场地试验结果表明:试验小车以2 m/s的速度跟踪参考路径时,最大横向偏差为-4.28 cm;3 m/s跟踪参考路径时,最大纵向偏差为-6.61 cm,可以满足农用车辆对于路径跟踪的精度要求。与线性模型预测控制器的对比试验表明:以3 m/s的速度跟踪圆形路径时,设计的控制器跟踪横向偏差降低了36.8%,纵向偏差降低了32.98%。