A good extension method of free space in an uncertain 2D workspace by using an ultrasonic sensor

In a robot workspace expressed explicitly by a quadtree representation that controls free regions and other regions, good algorithms have been presented which generates a fast collision-free path and a sequence of collision-free motions between the given start and goal for a mobile robot with complicated shape. Thus it is important to represent explicitly the workspace quadtree by using some sensor. If the workspace shapes can be acquired enough from an image obtained through a camera on the workspace ceiling, the presented algorithms fulfil their functions in the workspace. Otherwise, an another sensor must be used to construct the workspace quadtree. The authors propose an extension method of free space in an unknown two-dimensional workspace by using an ultrasonic sensor. Combining this extension method with the previous planning algorithms, the mobile robot can navigate to the target while avoiding obstacles even if it does not know their shapes and locations at all