提出了一种基于气电混合驱动且能够全天候工作的高效率苹果收获机器人.该机器人机械臂包含5个自由度,混合使用电动和气动两种驱动方式,同时保证机械臂的准确定位和末端执行器对果实的快速柔性抓取.机器人视觉识别系统结合了机器视觉和深度神经网络方法,在提高系统鲁棒性的同时优化了系统的整体检测速度.此外,机器人配备的夜间照明系统使其能够实现全天候工作.在实验室环境下进行了机器人视觉检测试验和苹果采摘试验,结果表明,视觉系统定位苹果的平均时间为44 ms,机器人的采摘率为81.25％,平均每个苹果的采摘时间为7.81s.

The existing tractor autopilot system relies on the built-in path tracking algorithm to approach a target point on the global path when the driver starts the navigation mode at any position in the field.The tracking path will produce a shock phenomenon near the target point,it will affect the path tracking accuracy,and the entire approach process is uncontrollable.In order to solve this problem,a guided trajectory planning method based on third-order B-spline theory was proposed.The trajectory with the minimum length was taken as the desired path,the maximum curvature constraint,the maximum steering angle constraint and the heading of starting goal point constraint were taken into account.Quantum genetic algorithm (QGA) was used to optimize the control points,and finally the B-spline theory was used to generate a smooth desired trajectory.Four kinds of operating conditions were tested in Matlab environment.The simulation results showed that the B-spline theory based on the QGA could be used to obtain a trajectory satisfying the multiple nonlinear constraints,and the trajectory curvature change was continuous,which was beneficial to the path tracking controller.A field test was carried out,and the results showed that the pure pursuit algorithm alone had a travel distance of 78.6 m,and the travel distance based on the guided trajectory planning method was 23.7 m.Compared with the existing autopilot system,the trajectory planning method based on B-spline theory was helpful to control the shape of tracking path and improve the controllability of approach process.

Unmanned tractors have become the focus of research in the field of intelligent agricultural machinery in recent years, which can effectively improve the agricultural productivity and the accuracy of operations. Automatic steering control system was a prerequisite for driverless tractors, and the key technologies included steering wheel angle detection method and angle tracking control algorithm. Steering wheel angle measurement results were the direct factors affecting the navigation effect. Encoder measurement and angular rate measurement were 2 commonly used front wheel angle measurement methods. Absolute angle measurement method had higher detection accuracy, but there were too many mechanical connectors using this method, and the calibration work was complex. The angular rate measurement method generally used inertial devices, which were easy to install and had a long working life. However, there was a random drift in the gyroscope and the accumulated error would affect the measurement accuracy. In practical applications, whatever the above-described measuring methods used, the angle measuring device was the most easily damaged component of the entire control system. For example, the angle measuring mechanism was easily damaged during operation by crops when it was installed in the position of the tractor front axle, it would cause signal output failure, and the reliability and safety of the unmanned system would be affected. In addition, the steering control system would be subject to a variety of non-linear factors such as mechanical clearance and hydraulic system lag during the working process, resulting in poor control effects. It was necessary to compensate for the uncertainties to further enhance the effect of angle tracking control. In order to solve the problems above, an RBF adaptive sliding mode control method with the ability of fault-tolerant detection of front wheel angle was proposed. First of all, a discrete state equation of the control system was deduced according to the structure of the tractor steering control system, and then the relationship between the front wheel angle and inertial information of the tractor's steering center including the lateral acceleration and the yaw rate was deduced respectively from the linear two-degree-of-freedom vehicle model. Secondary, in order to obtain a high redundancy value of the front wheel angle, 2 estimated values of the front wheel angle were obtained by Kalman filter, and a fault diagnosis algorithm and a fault-tolerant output algorithm were designed by comparing the residual threshold between the angle encoder value and these 2 estimated values. Thirdly, an adaptive sliding mode control method of the front wheel angle was proposed, which used an RBF neural network to identify the uncertain disturbance in order to ensure that the steering control system could accurately and timely track the expected rotation angle under the uncertainty interference factors. Finally, the fault tolerance test and automatic control test of the front wheel angle were carried out. The experimental results showed that the maximum error of the angular estimation based on the lateral acceleration was 2.94° and the root mean square error was 0.81°, while the maximum error of the angular estimation based on the yaw rate was 1.73° and the root mean square error was 0.12°. The fault-tolerant algorithm could automatically switch to the estimation value when the encoder artificially exerted an interference signal, and it could effectively replace the role of encoder and improve the reliability of the tractor automatic driving system. The experimental results of the angle control system showed that the performance indicators of the adaptive sliding mode control algorithm based on RBF network were better than the traditional PID (proportion, integration, differentiation) control method, and it could track the desired angle quickly with small overshoot, and the maximum error of angle control was 0.21° and the root mean square error was 0.07°. Test results showed that the fault-tolerant adaptive sliding mode control method can improve the reliability and accuracy of the automatic steering control system and help to solve the problem of high failure rate of the front wheel rotation angle measuring device of the tractor autopilot system.

针对由于转向机制不同导致的自动导航系统无法在轮式与履带式拖拉机上通用的问题,提出了一种基于履带转速与虚拟驱动速度和虚拟转向角转换模型的控制方法.通过对轮式车辆中性转向二轮模型和履带车辆转向模型的分析,推导出履带车辆两侧履带卷绕速度与虚拟转向角和虚拟驱动速度的表达式,为自动导航驾驶系统构建完整的反馈控制.进行实车转向试验,采集数据,对提出的转换模型进行验证.结果显示,该转换模型下理论值与实际参考值具有很好的对应关系,该转向控制方法可行.

为了提高自动驾驶系统对车身纵向速度和滑移工况等影响因素的自适应能力,提出了一种基于航向预估模型的路径跟踪控制算法.首先对NF-752型履带式拖拉机进行适应性改造,将其手动转向机构和无级变速机构改造为大扭矩舵机控制,通过绝对值编码器测量两侧履带转速,利用RTK-GNSS定位系统测量车辆地理位置和车身速度,结合车载计算机和底层控制器搭建了自动驾驶系统试验平台.然后以车身速度和两侧履带速度为状态变量,考虑履带滑移率建立了航向预估控制模型,进而提出了一种基于航向预估模型的路径跟踪控制方法.最后在沥青路面分别以低速(1 km/h)、中速(5 km/h)和高速(9 km/h)进行了直线路径跟踪试验,结果显示,在不同作业速度条件下,路径跟踪误差无明显差异,最大跟踪误差为-2.00 cm,标准差为0.93 cm.进行了田间曲线路径跟踪试验,结果显示,当拖拉机以6 km/h速度跟踪曲线路径时,跟踪误差优于10 cm,在滑移区域无明显误差增大现象.试验表明,提出的航向预估控制方法对作业速度有较好的适应性,一定程度上克服了滑移现象对控制精度的影响,可满足履带拖拉机耕整地作业精度要求.

家禽在线自动称量分级技术是实现家禽屠宰加工自动化的关键技术之一.主要介绍了基于PIC18F25K80单片机的称量分级控制系统的工作原理、系统组成和实现方法,并对质量传感器的选择,电路设计和单片机软件设计等方面进行了阐述.

At present, the combine harvester is developing towards the direction of large scale and high speed. It is more and more difficult to recognize the harvest boundary only by people's eyesight, which is to ensure the consistency of cutting when combine harvester operates. When the harvester works in the field, it usually works in full cutting conditions, which requires the driver has high driving skills, and the whole tracking operations keep for a long time; the labour intensity of the driver and the dust of field work make it difficult to rely on the naked eye to obtain accurate boundary. Combine harvester yield monitoring system is according to the need of harvest cutting and actual speed for real-time calculation of harvest area at home and abroad, and combine harvester yield measuring system mainly relies on the operator's input of harvesting information manually, but the actual harvest is difficult to ensure the full harvest. In the detection of combine harvester's feeding quantity, cut width, density, and speed of operation are needed to measure for the straw obtaining. The automatic driving system of combine harvester can automatically track driving according to the harvest boundary. Therefore, the on-line detection of harvest boundary is important for intelligent monitoring system of combine harvester. Aiming at the problem of on-line recognition of harvesting boundary of combine harvester, an on-line recognition system for harvesting boundary of combine harvester was developed by laser non-destructive detection technology. Firstly, the composition of the system, and the selection and working principle of the laser sensor were introduced, and the polar coordinate of the sensor output data is converted to the right angle coordinate. The laser sensor is a 1D (one-dimensional) scanning laser rangefinder supplied by the SICK company. It uses an infra-red laser beam and it works on the principle of light propagation time measurement. A short luminous impulse is emitted. The luminous ray is then deviated by a revolving mirror and thus covers a half-plane. When the ray meets an obstacle, retro diffused light is collected by the detector. The distance from the sensor to the object is then calculated from the time interval between the emission and the reception of the impulse. As the harvest process will produce a lot of dust, it will have a laser detection distance and signal reflection. The laser ray from the sensor is retro diffused when it meets suspended particles of dust. In consequence, the measured distance is shorter than that separating the top of the vegetation from the rangefinder. Through the comparison with crop characteristic threshold, the error data affected by dust are effectively identified and eliminated. Using moving average digital filtering algorithm, system measurement noise is eliminated. Through the signal step change pattern recognition algorithm, the on-line detection of the harvest boundary is realized, and the cutting amplitude of the combine harvester is calculated accurately. The test results show that the system can realize on-line monitoring, and the measurement error is not more than 12 cm, which can provide reference for the practical application of intelligent monitoring system of combine harvester.

For the problem of low positioning accuracy about agricultural goods and AGV (automated guided vehicle) in the warehousing environment, a system model applicable to agricultural warehouse is developed, which uses IR-UWB (impulse radio - ultra wide band) technology. This system adopts SDS-TWR (symmetric double sided - two-way ranging) scheme to set up positioning system model, and uses TOA (time of arrival) location algorithm to locate the mobile station. First of all, this system measures the distance between base station and mobile station, and then gets the optimal solution of error function using nonlinear least squares method. The optimal solution is the coordinate of the mobile station. This system also takes into account the factors of carrier frequency deviation, researches the source of error and tries to find out the method to reduce it. The analysis shows the main source of positioning error in 2 aspects, namely the method of distance measurement, and the method of calculating the position of the target nodes by the distance value. And we put forward the corresponding countermeasures for these 2 aspects, such as the selection of ranging algorithm, and the arrangement of the base stations. IR-UWB has many advantages such as strong anti-interference ability, high range accuracy, low power consumption, fast transmission speed and good security. Because of the convenience to layout network nodes and no strict requirements in the field environment, it is suitable for the field of agricultural warehousing. Finally, we design the positioning system of mobile and base station nodes based on DW1000 RF (radio frequency) chip, and respectively carry out static ranging experiments, static positioning experiments and dynamic positioning experiments. To increase contrast, this article also adds comparative trial, which uses TWR ranging algorithm. The static ranging experiments of this system adopt 6 distances of 10, 15, 20, 30, 40 and 50 m, and collects around 3000 sets of data respectively. The experiments show that the error of the mean value between the actual distance and the ranging distance is less than 50 mm, and the root mean-square error is less than 41 mm using SDS-TWR ranging algorithm. However, the former is more than 110 mm and the latter is more than 60 mm using TWR ranging algorithm in the same experiment condition. In the static positioning experiments we conduct some experiments to measure coordinates of some spots under the fixed coordinate system. The result shows the positioning error is less than 50 mm and the root mean-square error is less than 69 mm with SDS-TWR ranging algorithm. And the positioning error is more than 90 mm and the root mean-square error is more than 115 mm with TWR ranging algorithm in the same experiment condition. In the dynamic positioning experiments, according to the actual situation of agricultural material warehouse, we move the mobile station along 5 produce aisles and 9 routes, and obtain the distance between the gathered data and the actual path. The experiments show that the positioning accuracy is 85 mm, which can meet requirement of 150 mm positioning accuracy. Comprehensive experiments show the system set up by this paper can satisfy the requirements of practical application in indoor agricultural material storage.

小麦粗蛋白含量是其品质评价的重要指标,为探讨基于选择的短波近红外光谱变量定量判别小麦籽粒粗蛋白的可能性,采集了52份小麦籽粒样本,用湿化学方法分析其粗蛋白含量,获取其900～1 700 nm波段的光谱,进而利用该光谱进行预处理方法的优化研究及小麦籽粒蛋白敏感变量的优选研究,以偏最小二乘的方法建立了基于短波近红外光谱的小麦籽粒蛋白定量模型.结果表明:多元散射校正和小波变换结合是短波近红外光谱定量判别小麦籽粒粗蛋白含量较优的预处理方法;利用200次竞争性自适应重加权变量优选的统计结果,优选出1 028、1 158、1 199、1 367、1 407、1 445、1 478、1 494、1 550、1 584、1 661、1 686 nm 12个变量为小麦籽粒蛋白敏感变量,占全谱的2％,该方法可稳定、高效地优选光谱变量,降低水分对模型的影响;结合预处理优化及变量优选建立偏最小二乘模型,模型预测决定系数和预测均方差分别为0.961和0.369.可见优选的短波近红外光谱变量可用于定量判别小麦籽粒粗蛋白含量.

This study investigated the feasibility of developing a multispectral imaging method using key wavelengths from hyperspectral images for modeling and simultaneously predicting total volatile basic nitrogen (TVB-N), thiobarbituric acid reactive substances (TBARS) and K value in grass carp fillet during chemical spoilage. The established least-squares support vector machine (LS-SVM) and multiple linear regression (MLR) models using five successive projection algorithm (SPA)-selected and six genetic algorithm (GA)-selected wavelengths showed excellent performances for predicting TVB-N and K value with R(2)p > 0.900 and RPD > 3.000, and poor results for TBARS value prediction. The LS-SVM model using six GA-selected wavelengths showed good reliability and was considered the best for simultaneous determination of TVB-N, TBARS and K value. The distribution maps of chemical spoilage changes were generated using image processing algorithms. The results demonstrated the feasibility of developing a rapid and on-line multispectral imaging system using the feature wavelengths and chemometrics analysis. (C) 2016 Elsevier Ltd. All rights reserved.

在声光可调滤光器(AOTF)近红外光谱仪系统中,稳定的射频驱动信号源是AOTF实现分光功能的关键.本文根据近红外光谱分析所需的波长范围,利用直接数字频率合成(DDS)技术,设计了驱动AOTF所需的射频信号源.系统包括单片机控制模块、DDS信号发生模块、滤波电路、功率放大模块四部分.经实验测试,系统可在80～150 MHz频段内任意频点输出稳定的正弦信号波形,相对误差在0.45％以内,功率为1W,满足AOTF的工作要求(对应输出波长范围900～1 700 nm).

Agricultural Mini-UAV is playing a greater role in precision agriculture and is becoming the new focus of the agricultural machinery industry.The reliability of the agricultural Mini-UAV data link is the key factor that affects its stability.The redundancy switching system of radio modem and the quality testing system of radio communication are de-signed in this paper to improve and test the reliability of the agricultural Mini-UAV data link.The two systems are based on PX4 project on data link platform constructed by 3DR and Q Ground Control.This paper mainly introduced the struc-ture and principle of the two systems.The redundancy switching system of radio modem has passed tests to improve the reliability of the agricultural Mini-UAV data link.

In order to solve the problem of seedlings leakage planting caused by the lack of plug seedlings in trays during with the all automatic transplanter working, an automatic recognition and control device for vegetable plug seedling transplanter was designed. In contrast to the home and abroad information, the design considered the domestic user price affordability and practicality. In this device, a stepping motor was used for driving the seedling disc horizontal motion and another was used for driving the seedling disc vertical motion. Four photo-electric sensors as the seedling recognition device were fitted on both sides of the seedling gripper and seedling disc, which can detect of seedling shortage from four directions. A programmable logic controller (PLC) was used to control clamping seedling gripper motion and the seedling disc transport along horizontal and vertical direction. A touch screen as man-machine interface can set the parameter to fit different size seedling disc, which can also choose manually or automatic control method. In this paper, a comparative experiment f was carried out at different transplanting velocity under the condition of automatic recognition system or one month old pepper seedling. The results showed that it could work effectively. The movement of the seedling trigger photoelectric sensor produced a signal, so that the seedlings were stopped and clamped away when moving to the position of the seedling gripper. On the contrary, the lack of seedling lattice cannot trigger photoelectric sensor, on arrival will not stop the seedling fetching device position, but quickly through, directly send the next seedling to stop on position of the seedling gripper, waiting for the clamping seedling gripper. This improves the seedling grab efficiency, reducing the leakage rate of transplanting machine plant. Automatic recognition and control device had low misjudge rate when transplanting speed was form 60 to 120 seedling per minutes. When transplanting speed was from 60 to 90 seedling per minute, the transplanter had well leakage plantings rate and throwing seedlings rate that can effectively work at transplanting seedling. The response speed of automatic recognition and control device can work at high speed transplanter. The leakage transplanting rate with automatically seedlings picking recognition control system was 12% lower than without the automatically seedlings picking control system. The device achieved the objective of reducing leakage rate purpose. Relative to not used for automatic identification of transplanting seedling mode, it is reduced about 12%that the leakage rate of the whole system of plant in 60-90 seedling/min transplanting speed. Leakage rate of transplanting machine plant is close to the practical level of planting. It achieved the purpose of reducing the leakage rate of planting. <br> Through observation and analysis during experiments, if the conveying of seedling disk was changed to horizontal, it can maximize the seedling erect and reducing sensor misjudgment caused by seedling roll or tilt, the identification rate was improved. The system was only carried out in the laboratory test, but field test has not yet performed. In field operation, some factor may affect the automatic recognition and seedling fetching device reliability and stability, such as vibration of machines, interference of the sensor by sunshine outdoor. These need further test and examination. More field experiment needs to be tested for the influences of machine vibration and sunlight to the photoelectric sensor.

In order to satisfy the agricultural demond of rice transplanting, the transplanter work should guarantee the escapement even in straight line transplanting that is convenient for field management and harvest later. Because of variable soil conditions and hard work environment, the driver driving level and the long boring driving to follow line is a big influence to the accuracy of the rice transplant. It is easy to produce overline, leak line and cause losses in yield. Aimed at above problem, this text introduces an automatic navigation system developed on PZ60 rice transplanter based on global navigation satellite system (GNSS). The steering, transmission and transplanting control system of the rice transplanter were modified from manual control system to electro-hydraulic control system using electro-hydraulic proportional valve. According to the position information of the rice transplanter acquired from GNSS receiver and vehicle sensors, the close-loop feedback control system of steering was builded. The system can accurately control rice transplanter to follow row navigating and turn around at the end of field by the self-adaptive fuzzy control method. The road and field experiment results indicated that the lateral tracking error could be kept within 100 mm when the speed of the rice transplant is not greater than 1.0m/s. The control system can satisfy the requirment of rice transplanting.

为了进一步提高农业用水效率,设计了一种喷灌机自动作业控制系统.该系统通过GPRS通讯从服务器获得处方图,再通过ZigBee无线传输模块实时监测喷灌水源的压力,并以此对处方图进行补偿,再结合GPS坐标信息,即可对变频器在当前作业点的工作频率进行调整,进而控制喷灌机的行走速度,从而实现按处方图调整喷灌量的目的.该设计同时具有手动功能,可以控制喷灌机行走电机的启动、停止和调整变频器的工作频率.

A FPGA?based data acquisition system working in synchronous mode was designed for CMOS linear array image sensor G9214?512S produced by Hamamatsu Company. VHDL language is used to describe the system. The Isim software within Xilinx ISE Design Suit software is used to simulate the design time sequence. The system is configured with FPGA and verified by Spartan 3 XC3S200A?VQ100 produced by Xilinx Company. The simulation result indicates that the time sequence of the data acquisition system is right. The design of the data acquisition system has laid the foundation for further image information pro?cessing in the NIR?spectroscopy.

To improve the ride comfort of Agricultural machinery , based on GB/T 18697-2002 and ISO 2631 , A vibra-tion and noise testing measurement system for corn harvester was established .By using MATLAB , the processing to ac-celeration time-history series were completed .vibration dose and A-meter-weight noise were chosen to evaluate the ride comfort , vibration dose is used to evaluate the vibration and A-meter-weight noise is used to evaluate the noise .Depen-ding on the two indexes , A final evaluation on the corn harvester was made and some improve guidance were given .

In order to improve the accuracy and robustness of detecting tomato seedlings nitrogen content based on near-infrared spectroscopy (NIR), 4 kinds of characteristic spectrum selecting methods were studied in the present paper, i. e. competitive adaptive reweighted sampling (CARS), Monte Carlo uninformative variables elimination (MCUVE), backward interval partial least squares (BiPLS) and synergy interval partial least squares (SiPLS). There were totally 60 tomato seedlings cultivated at 10 different nitrogen-treatment levels (urea concentration from 0 to 120 mg . L-1), with 6 samples at each nitrogen-treatment level. They are in different degrees of over nitrogen, moderate nitrogen, lack of nitrogen and no nitrogen status. Each sample leaves were collected to scan near-infrared spectroscopy from 12 500 to 3 600 cm-1. The quantitative models based on the above 4 methods were established. According to the experimental result, the calibration model based on CARS and MCUVE selecting methods show better performance than those based on BiPLS and SiPLS selecting methods, but their prediction ability is much lower than that of the latter. Among them, the model built by BiPLS has the best prediction performance. The correlation coefficient (r), root mean square error of prediction (RMSEP) and ratio of performance to standard derivate (RPD) is 0. 952 7, 0. 118 3 and 3. 291, respectively. Therefore, NIR technology combined with characteristic spectrum selecting methods can improve the model performance. But the characteristic spectrum selecting methods are not universal. For the built model based or single wavelength variables selection is more sensitive, it is more suitable for the uniform object. While the anti-interference ability of the model built based on wavelength interval selection is much stronger, it is more suitable for the uneven and poor reproducibility object. Therefore, the characteristic spectrum selection will only play a better role in building model, combined with the consideration of sample state and the model indexes.