基于计算机视觉的自主导航机器人门检测算法

摘要

Door detection problem in Autonomous navigation robot was studied. For robot autonomous navigation area, no - visual sensor is not suitable for closed door detection, so the major work is how to effectively improve the indoor door detection's location. According to indoor door shape characteristic, this paper put forward a computer vision based autonomous navigation robot door detection algorithm. Algorithm only needs monocular vision image collection. According to the height, width and the characteristics of the shape of the door, the door detection can be realized. In the detecting process of, of the door features, the improved linear detection algorithm was used with high detection speed and high efficiency. The experimental results show that this method can be applied in not only a single background of doors, but also the more complex background. The door detection is effective and more robust. Therefore, it has great application value for robot autonomous navigation of home intelligence service.%研究了自主导航机器人中如何有效提高室内房门检测定位的问题.针对导航中非视觉传感器通过探测距离来判断门的位置,而关闭状态的门和周边的墙几乎处于同一平面无法定位,导致检测不准.可根据室内房门的形状特点,提出了一种计算机视觉的自主导航机器人门检测算法,能在单且视觉下进行图像采集,并根据房门的高度、宽度比以及门的形状特征,进而实现图像中门的检测.由于在检测门特征过程中使用了改进了的直线检测算法,因此具有检测速度快、效率高的特点.实验结果表明,与传统非视觉距离探测方法相比,改进方案不仅适用于单一背景下开状态的门检测,更对关闭状态门的检测具有有效性,完成导航平均处理时间约为2.2s,速度较高,对于家庭智能服务机器人的自主导航具有很大的应用价值.