基于单目视觉的水下机械手自主作业方法研究
文献类型：学位论文
| 作者 | 霍良青
|
| 学位类别 | 硕士 |
| 答辩日期 | 2012-05-28 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 张竺英 |
| 关键词 | 水下机器人 机械手 自主作业 单目视觉 运动规划 |
| 其他题名 | Research on Independent Operation Method of Underwater Manipulator Based on Monocular Vision |
| 学位专业 | 机械电子工程 |
| 中文摘要 | 随着海洋产业的迅速发展，对水下作业技术与水下作业工具提出了越来越高的要求，机械手是水下作业不可或缺的一部分，机械手的性能很大程度上决定了水下机器人作业能力。水下作业机械手按驱动方式可以分为电动机械手和液压机械手，目前液压机械手是水下主流作业工具，现有的液压机械手采用的控制方式大多数是遥控主从伺服式和开关式，目前主从伺服控制技术已经成熟，被广泛的应用于现有的机械手控制。但是主从式机械手需要操作者娴熟的操作技能和丰富的操作经验，操作起来费时费力，并且操作精度不高。可自主作业的水下机械手是未来发展的必然趋势，尤其是基于视觉系统的自主作业液压机械手更是以后智能作业水下机械手发展的方向。由于水下环境的特殊性，基于视觉的机械手自主作业技术面临诸多问题。 首先，自主作业机械手研究的关键技术之一是机械手的运动规划问题，本论文研究的对象是一款六自由度液压机械手，不同的机械手关节构成形式具有不同的运动规划方法，本文的运动规划算法主要讨论了一种结合几何法与反变换法思想的运动学逆解方法、基于登山法原理的关节状态快速搜索法、可达作业区域判定方法，机械手自干涉检测方法。 其次，在自主作业技术中对外界环境的感知是另一项关键技术，在水下作业中，无论是自主还是半自主作业，视觉都是不可或缺的外界环境感知方式，然而受水下环境的非结构化、光强度与距离成指数衰减、浮游生物对光的吸收和散射等不确定因素影响，水下图像纹理特征的提取、图像匹配都很困难，这样导致机器视觉在水下的应用受到限制，相比之下二值图像更稳定、抗干扰性强，但是二值图像含有的信息有限，如何从二值图像提取有效信息进而获得目标的三维信息是非常有必要进行研究的，本文基于单目视觉和手眼协调原理，提出了一种对具有几何形状约束特征的水下目标进行三维信息重建的方法。 最后，通过将单目视觉获取的目标三维信息作为运动规划的依据，进而完成机械手运动规划，最终实现自主作业。 |
| 索取号 | TP241/H97/2012 |
| 英文摘要 | With the rapid development of Marine industry, Human put forward more and more high demand for the underwater operations technology and underwater operations tools, manipulator is an integral part of underwater operations, the performance of the manipulator largely determines the underwater robot operation capacity．According to the driving way underwater operations manipulator can be divided into electric manipulator and hydraulic manipulator, at present the hydraulic manipulator is mainstream underwater operations tools, most of the existing hydraulic manipulator are master-slave servo type and switch type, at present the master-slave servo control technology is mature, and is widely used in the existing robot control but master-slave control robots need skilled operator and the rich operation experience, operate need time and manpower, what is more the operation accuracy is not enough. Autonomous underwater manipulator is an inevitable trend in the development of the future, especially the autonomous underwater hydraulic manipulator based on visual is the development direction of intelligent operations in the future. But the technology of autonomous underwater manipulator based on vision faces many problems. First of all, one of the key technologies of the autonomous manipulator is manipulator motion planning, this paper research a 6-dof underwater hydraulic manipulator, Different robots joint need different motion planning method, this paper mainly discussed the motion planning algorithm, Which mainly include inverse kinematic solution method of the manipulator、operational areas determinant method and self interference detection methods. Secondly, the technology of environmental perception of autonomous underwater manipulator is another key technology, in underwater operations, whether the Autonomous or half Autonomous operation, vision is an indispensable way to obtain external environment. In order to realize the Autonomous operation based on vision, the position and posture information of the target must be obtained. The target space pose information first to get the target space pose information, in the water environment information for close to target the most commonly used is in machine vision, however it is difficult to extract and match texture feature from an image underwater , Because of the underwater environment is unstructured、distance will introduce an exponential decay of light intensity, the plankton absorption and scattering light and some uncertain factors . This has caused in the application of machine vision underwater is restricted, by contrast binary image more stable and robust, but the binary image contains information is limited, it is necessary to study that how to extract information from binary image effectively and get the three-dimensional information of objective. Based on the monocular vision and eye-hand coordination principle, this paper proposes a method to reconstruct the three-dimensional information of objective underwater that with geometric shape constraint. Finally, through the monocular vision reconstruct three-dimensional information of objective, and transmit target information to motion control computer, the latter compute trajectory planning of manipulator, at last achieve the Autonomous operations. |
| 语种 | 中文 |
| 公开日期 | 2012-07-27 |
| 产权排序 | 1 |
| 分类号 | TP241 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/9270]  |
| 专题 | 沈阳自动化研究所_水下机器人研究室 |
| 推荐引用方式 GB/T 7714 | 霍良青. 基于单目视觉的水下机械手自主作业方法研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2012. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
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。