Montecarlo localization for a guide mobile robot in a crowded environment based on omnivision

Publication Type:

Conference Paper

Source:

Towards Autonomous Robotics Systems (TAROS 2008), Edimburgh (UK), p.1-8 (2008)

ISBN:

978 1 906849 00 9

Keywords:

autonomousrobots; robotica; C.Gamallo; P.Quintía; C.V.Regueiro; M.Mucientes

Abstract:

This work presents a localization system for a robot guide on a crowded environment based on omnidirectional vision and a map of ceiling landmarks. The developed approach uses a Monte Carlo particle filter to manage uncertainty, both on observations and control, in order to track the position of the robot. We describe how landmarks are detected on each image and how the problem of landmark association is posed. To demonstrate the robustness and reliability of our system, we present experiments carried out in a real environment, the Domus Museum in A Coruña (Spain). Results show that the proposed localization system can run on real time and along middle-long trajectories.