Vision-Based Nao Robot Localization Using Gaussian Approach

In this paper a vision-based self-localization method for Nao humanoid robot is proposed. We achieve a 2D position of robot using a monocular vision with likelihood estimation of Gaussian model. In general, due to noisy environment and sensor errors most recognized features are not accurate and these errors lead to inaccuracy in localizing robot. In this research, Gaussian distribution is used to deal with these uncertainties. The results of this research can also be used in probabilistic approaches for localization which are among the most acceptable approach of localization today. This project is implemented in Webots simulation environment.

Keywords:

Robot Localization , Gaussian Distribution , Geometrical Trilateration , Computer Vision