SIA OpenIR  > 空间自动化技术研究室
巡检机器人自主控制方法研究
其他题名Research on Autonomous Control Algorothm for an Inspection Robot
李贞辉1,2
导师王越超 ; 王洪光
分类号TP242
关键词巡检机器人 自主越障 障碍物识别 路径规划 质心调整 自主抓线
索取号TP242/L36/2014
页数145页
学位专业模式识别与智能系统
学位名称博士
2014-05-29
学位授予单位中国科学院沈阳自动化研究所
学位授予地点沈阳
作者部门空间自动化技术研究室
摘要目前,架空输电线路的巡检主要依靠巡检工人携带各种检测设备沿线路徒步或驱车进行检测,效率低、劳动强度大、危险性高。输电线路巡检机器人是一种能够工作在野外架空输电线路环境下协助或代替巡检工人进行线路巡检的特种机器人,能够降低巡检工人的劳动强度和危险性。本文以“863”计划支持项目“超高压输电线路巡检机器人技术与系统”及国家电网支持项目“山区智能巡检装置”和“智能输电线路巡检机器人”为依托,针对巡检机器人在架空输电线路环境下的自主运动控制开展相关研究工作。介绍了巡检机器人的国内外研究现状,并对研究热点问题进行了总结,重点对巡检机器人自主控制技术进行了较为详细的综述。根据架空输电线路的环境特点和巡检机器人的任务需求,介绍了本文研制的巡检机器人的机械系统和控制系统设计,建立了机器人的运动学模型,并对其逆运动学进行了分析。以此为基础,针对巡检机器人的自主障碍物检测和自主越障开展以下四个方面的研究:(1)输电线路障碍物的自主定位与识别;(2)巡检机器人越障路径的自主规划研究;(3)巡检机器人越障中的自主质心调整策略和自主抓线控制方法;(4)巡检机器人实验研究。 (1) 针对巡检机器人自主运动中各阶段对障碍物检测的精度和实时性要求,提出了一种用于线路障碍物定位与识别的多传感器集成结构,并介绍了各阶段的障碍物定位与识别算法。基于单目测距原理对机器人前方障碍物进行了粗定位,并提取了障碍物相对高度、物理宽度、长宽比和图像占空比特征;利用接触传感器、编码器和电机电流传感器的融合对障碍物进行精定位;在确定位姿下提取障碍物图像边缘小波不变矩特征,针对障碍物环境和提取特征的不确定性,研究利用巡检机器人不同阶段提取的障碍物特征对障碍物进行模糊识别的方法。 (2) 根据障碍物定位与识别结果构建障碍物环境,对巡检机器人越障路径规划开展了相关研究。巡检机器人越障路径的自主规划能力是实现机器人自主越障的基础,即在已知或未知的障碍物环境中得到连续、光滑和最短的越障路径。针对本文所设计的输电线巡检机器人结构和期望的评价指标(距离、执行难度),提出了一种基于可视图法的越障路径规划方法。针对越障过程中,巡检机器人手臂连杆易于障碍物侧面发生碰撞的问题,定义了碰撞危险系数指标,提出了一种基于模糊逻辑的手臂避障算法。设计了机器人跨越防振锤和悬垂线夹的仿真和实验,验证了所提越障路径规划方法和手臂避障算法的有效性。 (3) 对巡检机器人越障中的局部自主行为——质心的动态稳定控制和基于视觉的自主抓线控制,开展了相关研究。针对轮臂复合式巡检机器人质心的动态稳定控制,提出了一种质心调整策略,以时间最短为目标规划了机器人关节的运动轨迹,并根据机器人动力学模型设计了一种基于状态反馈的质心调整控制器,解决了以往方法只能保证机器人质心的静态稳定问题,使机器人在运动过程中能够保持机器人质心的动态稳定。针对巡检机器人自主越障时的抓线问题,基于输电导线纹理特征和积分投影方法估计导线的位姿(偏距、偏角),根据位姿偏差设计了自主抓线的仿人智能控制器,利用偏角、偏距和线宽的估计值并结合机器人的倾角信息对机器人进行自主抓线控制,解决了以往算法在导线和背景灰度差异小时无法可靠抓线的问题。 (4) 对所设计的双臂巡检机器人进行了实验室模拟线路环境和现场实际线路环境下的实验研究。在实验室搭建的模拟线路环境下,进行了最大滚动行走角度实验、防振锤自主跨越实验、悬垂线夹自主跨越实验、过耐张线夹和引流线实验等,以验证机器人在模拟线路上的爬坡和越障性能;在实际现场线路环境下,进行了跨越防振锤和悬垂线夹实验等,以验证机器人在实际线路环境中的越障性能。
其他摘要This inspection task for power transmission lines is carried out commonly by foot patrol inspection that a team of workers drive or walk from tower to tower to inspect the power transmission lines with the help of binoculars and infrared cameras on the ground. Foot patrol inspection has many disadvantages, e.g. inefficient, high working intensity and risk for workers. The mobile robots applied to inspect power transmission lines are autonomous or semi-autonomous, which replace people or assist people in inspecting power transmission lines. The research of this paper is supported by the National High Technology Research and Development Program of China (‘863’Program) item ‘inspection robot technology and system for ultra-high power transmission lines’ and by the State Grid item ‘inspection robot for inspecting power transmission lines in mountainous regions’ and ’ autonomous inspection robot of power transmission lines’. For the purpose of improving automation of the inspection robot, the researches on inspection robot system and their research hotspots are introduced, especially the automation technologies of inspection robot are proposed in detail. According to the enviroment of overhead transmission lines and the mission requirements of inspection robot, the design of mechanical systems and control system are proposed. The kinematics model is developed and its inverse kinematics is analyzed then. Aiming at the obstacle detection and the obstacle crossing, four aspects of research work are engaged as follows: (1)location and recognition for the obstacles of power transmission lines; (2) path planning in obstacle crossing; (3) autonomous Centroid adjustment strategy and Line-grasping Control; (4) experiments. (1) Aiming at the obstacle location and recognition when the inspection robot on high voltage transmission line is running, the accuracy and real-time performance requirements for obstacle location are analyzed when the inspection robot is running, and an integrated structure of multi-sensor system for obstacle location and recognition is present. The obstacles are detected by the Monocular measurement principle, located by fusing the sensors of contact sensor, encoder and current sensor. The obstacle recognition method is proposed by fusing the wavelet feature and shape characteristics of images from different cameras. (2) Based on the obstacle environment built by the result of the obstacle location and recognition, the research on path planning of obstacle crossing is carried on. For the structure of the inspection robot and expected evaluation(distance and execution difficulty) , a method for path planning of obstacle crossing is proposed based on a visual graph algorithm. Aiming at the collision problem between robot arm and obstacles, a collision risk factor is defined and a obstacle avoidance algorithm based on fuzzy logic is proposed. Some experiments of crossing the damper and suspension clamp are carried out to show that the effectiveness of the aforementioned functionalities. (3) Aiming at the dynamic stability for a dual arm inspection robot when it is hung by single arm, this paper presents a centroid adjustment strategy. Analyzing the obstacle negotiating process of the robot, the kinetic model when the robot is hung by single arm is built and a state-feedback controller based on the robot model is presented. This paper discusses the planning method for joints based on best time, achieves the dynamically stable movement by regulating joint variables and adjusting the centroid of the robot. Aiming at the line-grasping control when the inspection robot crosses obstacles automatically, this paper presents an automatic line-grasping control method. A pose estimation method for transmission line is designed based on its textural features and integral projection algorithm. The line-grasping algorithm is a human simulating intelligent controller with the offset distance, declination, and width of the line, and the obliquity of the robot. (4) Experiments in laboratory and in field enviroment are carried out. In laboratory, the robot is tested by the large-angle rolling experiment, autonomous counterweight crossing and suspension clamp crossing experiment, tension clamp crossing experiment. These experiments validate the performance paramenters of the robot in laboratory enviroment. In field enviroment, the robot is tested by counterweight crossing and suspension clamp crossing experiment to validate the ability of obstacle avoidance and HRI in field enviroment.
语种中文
产权排序1
文献类型学位论文
条目标识符http://ir.sia.cn/handle/173321/14818
专题空间自动化技术研究室
作者单位1.中国科学院沈阳自动化研究所
2.中国科学院大学
推荐引用方式
GB/T 7714
李贞辉. 巡检机器人自主控制方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2014.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
巡检机器人自主控制方法研究.pdf(5589KB) 开放获取CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[李贞辉]的文章
百度学术
百度学术中相似的文章
[李贞辉]的文章
必应学术
必应学术中相似的文章
[李贞辉]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。