基于光栅投射立体视觉的暗环境中移动机器人视觉导航方法研究

项目名称： 基于光栅投射立体视觉的暗环境中移动机器人视觉导航方法研究

项目编号： No.61305123

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 张晓玲

作者单位： 江苏理工学院

项目金额： 23万元

中文摘要： 在黑暗环境中且携带能源有限无法采用外来光源照明的情况下，移动机器人如何进行自主导航是亟待解决的难题。采用面结构光栅投射和立体视觉相结合的新型视觉模型，本项目提出基于光栅投射立体视觉的暗环境中自主移动机器人视觉导航方法，本方法既规避立体视觉的立体匹配难题，又避开光栅投射的相位展开难题，进行暗环境中机器人导航的障碍物检测、SLAM和视觉里程计研究。首先,研究光栅投射立体视觉传感理论，建立光栅投射立体视觉几何及数学模型；第二,研究机器人视场内空间障碍物的三维坐标计算方法，精确定位障碍物；第三,分析图像序列中特征目标的3D坐标信息，研究特征目标识别和跟踪算法；最后,建立机器人运动参数与特征目标在图像序列中的位置变化关系，研究机器人姿态、朝向和运动距离等运动参数的获得方法，实现运动机器人精确定位。本研究有利于目前黑暗环境中机器人无法自主导航难题的突破，为暗环境中无GPS支持的机器人导航提供基础探索。

中文关键词： 光栅投射立体视觉传感器；暗环境；移动机器人导航；车体定位；动态特征跟踪

英文摘要： In dark environment without enough energy and extra light source, autonomous navigation of mobile robot is an urgent problem in the robot field. A novel visual navigation method based on grating projection stereo vision for mobile robot in dark environment is proposed. This method is combining with grating projection profilometry of plane structured light and stereo vision technology. It can be employed to realize obstacle detection, SLAM(Simultaneous Localization and Mapping) and vision odometry for mobile robot navigation in dark environment without the image match in stereo vision technology and without phase unwrapping in the grating projection profilometry. Firstly, we research the new vision sensor theoretical, and build geometric and mathematical model of the grating projection stereo vision system. Secondly, the computational method of 3D coordinates of space obstacle in the robot's visual field is studied, and then the obstacles in the field is located accurately. Thirdly, the 3D coordinates of feature objects in the image sequences of two CCDs are analyzed, and the Recognition and tracking algorithm of feature objects is researched and established. Finally, the motion parameters of mobile robot including attitude, orientation and movement distance are obtained by building the relationship between motio

英文关键词： Grating projection stereo vision sensor；Dark environment；Mobile robot navigation；Robot localization；Dynamic tracking algorithm