Visual and laser guided robot relocalization using lines and Hough Transformation

Abstract

Knowledge about where is a mobile robot in a specific moment is crucial for decision-making when it comes to be explored an area known or unknown. Map building and relocalization are two fundamental problems by mobile robots and humanoids, and they are very closely connected. The use of geometric entities provided by the various input devices (laser rangefinder, ultrasound, etc.) helps us for the easy implementation of these maps in 2D. In this paper the authors present a new technique for the construction of maps using a laser rangefinder and a stereo camera, where the goal is to build a 3D geometric map. Using the conformal geometric algebra the authors will have a compact representation of the elements of the 3D environment. The relocalization of the mobile robot in the environment when this is already known, either built or previously proposed artificially is another problem which is attacking in this paper. Using Hough transform on the line segments already recorded on the map will make a matching in the Hough domain to get the current position of the robot in the environment.