Indoor Slam Using an Omnidirectional Camera

Abstract

This paper presents a Simultaneous Localization and Mapping (SLAM) algorithm for an indoor robot using bearing-only observations. An omnidirectional camera is used to observe indoor scene from which vertical lines are extracted to obtain bearing measurements. To track vertical lines through sequence of omnidirectional images, a matching algorithm based on histogram of oriented gradients technique is proposed. The Extended Kalman Filter (EKF) is used to estimate the 3-DoF motion of the robot along with two-dimensional positions of vertical lines in the environment. In order to overcome bearing-only initialization, the Unscented Transform is used to estimate the probability distribution function (PDF) of an initialized vertical line. Simulations have been carried out to validate the proposed algorithm.