A cost effective probabilistic approach to localization and mapping

Localization and mapping in robotics are preliminary but challenging problems. A learning approach must be followed by a robot to understand its environment and perform data association before it accomplishes any other tasks. In this paper, we describe a novel combination of techniques to map the environmental boundaries traced by the robot and localize it inside the bounded region. This is an effort established using only an iRobot educational package and no expensive high-end external sensor. This method may be treated as a solution for mapping and localization in a static environment with a few low cost IR sensors. In the proposed approach, we trace the robot´s movement in an arbitrary shaped bounded region and map the same using coastal rule wall following technique and the method of least squares. A full traversal of robot maps the boundary and the robot is localized in the environment using particle filter approach and computational geometry. Also, we studied the effect of localizing a kidnapped robot once the map is known.