机器人的姿态是机器人自主爬楼的前提.为了获得机器人与台阶棱线的夹角,对视觉传感器拍摄的数字图像进行处理.首先对图像一阶微分后按“录取线”取阈值二值化,凸现台阶棱线的像;其次采用聚类分析的办法,提取出像上台阶斜率,该方法能滤除噪声且逼近真实值;最后根据几何成像原理,列出了像上斜率与真实斜率之间关系的表达式并对式中参数进行了标定.经实验验证得出误差不超过5°.
The pose of a mobile robot is the precondition of stairs climbing automatically. A digital image from visual sensor is processed in order to get the angle formed by a robot and a stair edge. Firstly the method of one order differential is used to show stairs edges obviously in an image according to an appropriate threshold acting as "the lowest mark of matriculates". Secondly the clustering analysis is used to pick up the stair slope in the image. Finally the expression of the stair slope in an image and its real slope is listed on the basis of imaging principle and two parameters are calibrated. Errors are no more than five degrees by experiments.