Monocular-SLAM Using Floor Lines

Abstract

This work proposes a SLAM technique based on Extended Kalman Filter (EKF) to navigate a robot in an indoor environment using odometry and pre-existing lines on the floor as landmarks. The lines are identified by using the Hough transform. The prediction phase of the EKF is implemented using the odometry model of the robot. The update phase directly uses the parameters of the lines detected by the Hough transform without additional intermediate calculations. Experiments with real data are presented.