研究了移动机器人的视觉定位和目标的运动估计。采用单目视觉系统,借助人工标识物,由小孔成像模型及空间几何关系,推导出视觉测距模型,并实现了移动机器人的自定位和目标的定位。通过序列图像,应用基于特征的运动分析方法估计球体的运动参数,推导出移动机器人对运动目标的跟踪模型。球体定位实验结果表明:该方法的定位精度较高。
Visual localization and motion estimation for a mobile robot are investigated. First a monocular vision system is presented to deduce a visual depth measurement model by means of landmarks, pinhole imaging model and spatial geometrical relationship, thus localizing the mobile robot and a moving object. Then, a feature-based motion analysis method is proposed to estimate motion parameters of the object from sequential images. A motion model for the robot to track the moving object is deduced. Experimental results show that the object is localized accurately.