【作者】 杨丽；

【导师】 张铁中； 张宾；

【作者基本信息】 中国农业大学 ， 农业机械化工程， 2005， 博士

【摘要】 在工厂化组培苗生产过程中,为减少病菌污染,生产上多采用细口培养瓶育苗。由于培养瓶口径小,瓶内空间狭小,造成机械化操作难度大,因此,目前尚没有相应的自动化分割移植装置。组培苗的分割移植生产还处于手工作业阶段。但手工作业工作效率低、劳动强度大、人力成本高,同时由于人的介入造成组培苗易被病菌感染,使苗的成活率降低、品质变差,已经严重制约了组培产业的发展。本文针对这一问题,对条状组培苗自动化生产方式和装置进行了较深入的研究,并取得了重要进展。主要研究成果如下: 1、首次对培养瓶育苗方式条状组培苗的自动化生产技术进行了研究。针对瓶式育苗的特点,提出采用机器视觉技术对瓶中每株苗进行识别定位,采用机器人将苗从瓶中分别取出,再进行切割、移植的思想。据此规划了简易可行的组培苗分割移植自动化生产总体实现方案。 2、在国内首次研制开发了能自动进行组培苗抓取和切割作业的5自由度关节式组培苗切割移植机器人。该机器人结构小巧、紧凑,设计新颖,具有创新性。 3、对组培苗分割移植机器人进行了运动学分析,依据D—H方法建立了机器人的运动学方程,实现了运动学方程的求解;针对运苗、取苗插苗作业过程要求,分别在关节空间和直角坐标空间对机器人的运动轨迹进行了规划,以满足机器人在不同作业区段对运动路径的要求;给出了机器人运动轨迹的实时生产方法。该研究为精确地实现机器人的运动控制提供了数学基础和理论依据。 4、建立了基于TMS320C6711 DSP的嵌入式机器人视觉系统。对组培苗的识别算法进行了研究,并基于TMS320C6711实现了识别算法的软件开发。实验结果表明,培养瓶中每株苗位置识别算法的准确率为92%,单株苗节点位置识别算法的准确率为91%,均达到了设计要求;开发的视觉系统软件具有良好的稳定性和较强的实时性,苗位置的平均识别时间为0.266s,,单株苗节点位置的识别时间为0.562s,能很好地满足机器人控制系统对视觉系统的实时性要求。 5、自行研制开发了适于本机器人系统的嵌入式控制器,包括基于ARM嵌入式微处理器的上位机控制器和基于单片机AT89C52的关节控制器的全部硬件设计与软件开发,并在上位机控制器硬件电路板上成功地实现了uCOS-Ⅱ实时嵌入式操作系统的移植,对整个控制系统进行了切实可行的软硬件抗干扰设计。该嵌入式控制器的成功开发,对机器人控制系统的小型化、专用化、高稳定性、低成本奠定了基础,对相关的嵌入式应用具有参考价值。 6、对机器人系统的整机工作性能进行了测试,测试结果表明:机器人达到运动定位精度范围的准确率为75%,机器人从培养瓶中取苗的成功率为70.5%,剪苗成功率为71.8%。整机工作性能不太理想,机器人的运动定位精度有待提高。更多还原

【Abstract】 In the process of tissue culture plantlet production, the plantlet bottle with small entrance is used for reducing bacteria pollution. Since the diameter of the plantlet bottle’s mouth and the space inside the bottle are both in the small size, at present, there hasn’t been the relevant automatic transport system yet. The transplant of tissue culture plantlet is still done by hand. However, because of the low efficiency, labor compression and high cost of the transplant by hand, and especially the infection of bacterium from hands in transplantation, the ratio of plants’ livability and the quality of the transplant becomes bad, which highly limit the development of tissue culture plantlet. This article aims to resolve those problems basing on a deeper study on tissue culture plantlet transplant robot and some important progresses have been made. The followings are the main research findings:1. Automatic product technique of tubular tissue culture plantlet in plantlet bottles is studied for the first time. Aiming at the characteristic of the plantlets in bottle, the idea of taking out one plantlet from bottle at one time to cut and transplant is put forward. According to this, the entire plan on automatic production of tissue culture plantlet is programmed.2. Design and make an articulated robot with 5 degrees of freedom, which can pick up and cut tissue culture plantlet automatically. On the basis of plentiful experiences, the length of the robot’s arm and the maximum range of every joint’s movement are confirmed on the request for working position, and the main technological parameters are given.3. Kinematical analysis is processed on the cut and transplant robot for tissue culture plantlet, the model of kinematical equation is made, and the positive solution and the negative solution of the kinematical equation are completed. According to the request of transportation, picking up, inserting plantlets, the motion track of the robot is programmed in the space of joints and rectangular in order to meet the need of route and velocity in different kinds of working place. The real-time calculating method for motion track of the robot is given. This research offers mathematic base and theoretical reference for the motion control of the robot.4. Found the embedded robot vision system based on TMS320C6711 DSP. The research on identification algorithm of the position of every plantlet in bottle and the node of single plantlet is done, and software of the identification algorithm is completed based on TMS320C6711 DSP. The results of the experiences showed that the accuracy rate of identification algorithm of the position of every plantlet in bottle is 92%, and the accuracy rate of identification algorithm of the position of the node of single plantlet is 91%. Both of them have reached the requirements of design. The software of vision system is of good stability and strong characteristic of real-time. The average time of identification of the position of every plantlet is 0.266s, and the average time of identification of the position of the node of single plantlet is 0.562s, which can perfectly meet the requirements of the real-time of robot control system.5. According to the request of function, the entire control plan is put forward, and aembedded controller suitable for this robot system is developed and explored. A hardware circuit board of the main controller based on ARM embedded MPU is produced, the start-up program bootloader of the system is complied, and the transplant of uCOS- II real-time embedded operation system is transplanted successfully on this board and the multi-assignment application program is explored on the base of uCOS- Ⅱ. The hardware circuit board of the joint controller based on MCU AT89C52 is produced, the control program of the position of robot joints is complied. Anti-jamming design of software and hardware to the entire control system is pressingly and reliably completed, which enhance the reliability of the control system.6. The test on the complete machine is done on the robot system, w更多还原