Real-time single camera SLAM using fiducial markers

Abstract

In this paper, a real-time single camera simultaneous localization and mapping (SLAM), that uses artificial landmarks is proposed. Proposed method uses the extended Kalman filter (EKF) to estimate robot pose and landmarks position. The core of the approach is the online creation of a map of fiducial markers in the environment within a probabilistic framework. Our key contributions include a development of measurement model of fiducial markers, solutions for global registration of fiducial markers, and the development of calibration-free indoor localization method. We present a detailed method to estimate the 3D location of fiducial markers from an image and how the robot is positioned. Simulation and experimental results for a self-developed mobile robot are both presented.