Automatic Calibration for a Robot Navigationsystem

This paper attacks the problem of automatic calibration of a mobile robot in an indoor environment. The problem is that the robot must locate itself accurately with respect to given landmarks, before it starts to navigate. In this paper, we examine the problem in a special case, where the ground plane is assumed to be horizontal and there are two locally parallel sidelines available. This assumption holds in many indoor environments, such as hallways, where the system's success has been demonstrated. The algorithm uses geometric features such as vanishing points and line orientations. Both theoretical analysis and experimental results show that this algorithm works very robustly and accurately.