一种基于单目视觉的自主机器人导航控制方法

首先对单目视觉获得的原始图像进行图像校正处理，利用边界判据函数分割获得可行区域，得到障碍物位置信息，然后在可行区域中根据障碍物位置分段给出机器人导航控制边界，最后采用拟合曲线模型得到前进路径曲线。试验表明，该算法有效稳定，可实时获取可行区域并实现移动机器人自主避障功能。

A Method of Autonomous Robot Navigation and Control based on Monocular-vision

This paper takes correction measures to the original image obtained by single camera firstly, and then segments the accessible region for the robot from image via boundary criterion function, obstacles will be positioned by segment result, and the navigation and control boundaries will extracted from the distribution of the obstacles in the accessible region, at last, path curves matched quadratic model are derived by curve fitting method. Experiments results proves the stability of the algorithm, autonomous robot can segment the accessible region and avoid obstacles with acceptable time-consuming.