| 题名 | 基于视觉的移动机械臂控制研究 |
| 作者 | 艾坤 |
| 学位类别 | 工程硕士 |
| 答辩日期 | 2015-05-22 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 曹志强 |
| 关键词 | 移动机械臂 轮廓提取 双目视觉 视觉避障 目标趋近与抓取 mobile manipulator contour extraction binocular vision vision-basedobstacle avoidance object approaching and grasping |
| 其他题名 | Research on Vision-Based Mobile Manipulator Control |
| 学位专业 | 控制工程 |
| 中文摘要 | 移动机械臂通过结合移动平台的移动能力和机械臂的操作能力，实现系统整体工作能力的提升，在军事、水下作业、危险品处理、星球探测等方面具有广阔的应用前景。本文针对基于视觉的移动机械臂控制开展研究，论文的主要内容如下： 首先，介绍了移动机械臂的研究背景和研究意义，对机器人视觉系统和移动机器人路径规划进行了综述，阐述了移动机械臂系统的研究现状和控制方法，并对论文内容和结构做了介绍。 其次，开展了基于单目视觉的移动机器人避障研究。针对移动机器人避障需要获取障碍物边界位置的问题，将边缘提取算法和图像形态学处理方法结合起来，实现图像纹理边缘和轮廓边缘的有效区分，再通过摄像机成像几何模型计算出障碍物轮廓相对于机器人的位置关系，从而达到定位障碍物的目的。在此基础上，给出了一种移动机器人实时避障策略。未知环境下的避障实验表明了所提方法的有效性。 第三，利用光轴平行的双目视觉成像原理，通过求取双目视觉视差图的方式实现了场景中物点三维坐标的获取。基于双目摄像机标定的内外参数，对双目摄像机获取的图像对进行校正，而后利用局部图像匹配的方式求取校正后图像对的视差图，从而实现双目视觉的三维重建，通过实验进行了验证；最后通过该立体视觉方法实现了对所设计的红色柱状目标的识别及定位。 第四，面向视觉框架下的移动机械臂目标操作，分别提出了基于模糊控制和基于行为的目标趋近与抓取方法。基于模糊控制的目标趋近将双目视觉处理得到的目标相对于机器人的角度作为模糊控制器的输入，而后通过模糊控制器实现移动平台对目标的趋近，并利用逆向运动学实现对目标的抓取。考虑环境中障碍物的约束，设计了基于行为的目标趋近与抓取方法。该方法包含目标搜索、目标趋近、避障行走和目标抓取四种行为。移动机械臂先通过双目视觉进行目标识别和定位，搜索到目标后，激活目标趋近行为；当移动机械臂在趋近目标过程中检测到障碍物阻碍运动时，切换为避障行走行为，并在完成避障后转回对目标的搜索或趋近；移动机械臂进入目标可操作范围时，激活目标抓取行为，控制机械臂完成对目标的操作。实验验证了所提方法的有效性。 最后，对本文工作进行了总结，并指出了需要进一步开展的研究工作。 |
| 英文摘要 | Mobile manipulator achieves better ability by combining mobility of mobile platform with the manipulator's operation capability. It has many potential applications in military, underwater operation, hazardous materials handling and planetary exploration. This thesis focuses on the vision-based mobile manipulator control. The contents are as follows: Firstly, the research background and its significance of mobile manipulator are given. The robotic vision system and path planning are reviewed. The research development and control methods of the mobile manipulator system are then described. The contents and structure of this thesis are also introduced. Secondly, the research on monocular vision-based obstacle avoidance of mobile robot is carried out. Aiming at the boundary locations of obstacles, the edge detection algorithms and morphological method are combined to distinguish textured edges from contour edges. Then the relative positions of obstacles’ boundries to the robot can be calculated by the camera imaging model. On this basis, a real-time obstacle avoidance strategy is given. The experiments of obstacle avoidance in unknown environments demonstrate the effectiveness of the proposed approach. Thirdly, according to the principle of binocular stereo vision, the 3D coordinates of object points are obtained by disparity map. The image pairs acquired by binocular camera are first rectified by camera calibration parameters, and then disparity map can be generated from them by local stereo matching, which is testified by experiments. At last, the recognition and localization of a red cylindrical object are realized. Fourthly, aiming at object operation by a mobile manipulator under the visual framework, fuzzy-based and behavior-based object approaching and grasping methods are proposed, respectively. The fuzzy-based method realizes object approaching by a fuzzy controller which takes the angle of object relative to the robot obtained by the binocular vision system as an input. And object grasping is realized by inverse kinematics. Consider the constraint of obstacles, a method based on behaviors is designed. It includes object searching, object approaching, obstacle avoidance and object grasping behaviors. The mobile manipulator first recognizes and localizes the object by binocular vision. After the object is found, the object approaching behavior is activated. When the obstacles affect the motion of the robot, it switches to the obstacle avoidance be... |
| 语种 | 中文 |
| 其他标识符 | 2012E8014661082 |
| 内容类型 | 学位论文 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/7751]  |
| 专题 | 毕业生_硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 艾坤. 基于视觉的移动机械臂控制研究[D]. 中国科学院自动化研究所. 中国科学院大学. 2015. |
| 个性服务 |
| 查看访问统计 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
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。
修改评论