摘要
移动机器人立体视觉系统不仅提供三维地形图用于障碍规避和路径规划,其结果还可以用于视觉导航。以移动机器人立体视觉系统为基础,研究了基于前后两个位置上立体图对的视觉测量算法用于移动机器人的连续导航,讨论了影响导航精度的因素和改进方法;研究了基于局部和全局三维地形图的地形匹配算法用于定期校正位置误差,算法实现简便,定位精度取决于地形图精度。实验结果证明了两种方法的有效性,可以兼顾近距离和中远距离导航任务。
Mobile robot stereo vision system not only provides 3D terrain map to be used in obstacle avoiding and path planning,but also lays the base for visual navigation. The visual odometry algorithm based on two stereo pairs is studied for application in continuous navi-gation. The factors related to navigation precision are discussed and improvement measures are presented. A terrain matching algorithm based on local and global terrain map is proposed for application in positioning error correction. The algorithm is simple and convenient and its precision depends mainly on the precision of terrain map. Experiments with analog lunar terrain indicate that these two algorithms are effective and can be used in close range and long range navigation.
出处
《计算机技术与发展》
2014年第3期106-109,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(61171189)
辽宁省教育科学研究一般项目(L2012400
L2012396
L2012397)
博士后基金项目(2012M520158)
关键词
立体视觉
视觉测量
地形匹配
stereo vision
visual odometry
terrain matching