Localization and navigation of a mobile robot in an office-like environment

This article focuses on the localization and navigation of a mobile differential robot in an indoor office-like environment. These are fundamental issues to service robotics, which is a branch with a strong market growth. The work implements a vision tracking system, environment mapping, route planning and navigation for an autonomous robot application inside services buildings. One goal of the methodology is its application with low cost equipment. The test bed chosen was a Pioneer P3-DX robot [16] in a service building, with an attached USB webcam, pointed at the ceiling to take advantage of the position of the light fixtures as natural landmarks. The robot location is estimated through two distinct probabilistic methods: a particle filter, when there is no information about the starting location of the robot, and the Kalman filter, given the convergence of the particle filter. Both methods use the detection of light fixtures together with the robot kinematics as information to estimate the pose. The mapping of the environment and its obstacles is obtained from the localization estimates and the information gathered by ultrasound sensors, representing the entire navigation space discretized in the form of an occupation grid. Planning the navigation path is determined by a simple search algorithm, namely the Wavefront algorithm, based on the information contained in the occupancy grid. For a given path, navigation is performed with obstacle avoidance using the virtual forces method. Replanning is used to recover from local minima situations.