抄録
The odometric dead reckoning is a fundamental method to estimate the self-location of a mobile robot. However, an environment to which the method is able to be applied is only a flat surface. Therefore, we plan to extend the environment to a curved surface by taking into account a spatial relation between the surface and robot's wheels. Our extension requires information of wheel-terrain contact points in addition to rotation angles of wheels. In previous work, we have proposed two types of contact point detectors, an image sensor type and a mechanical probe type. However, precisions of the image sensor type and a mechanical probe type are influenced by environmental light and elastic deformation of the probe tip, respectively. In this paper, we propose a new detector that finds a wheel-terrain contact point by scanning the terrain with an infrared proximity sensor.
