Simultaneous Localization and Mapping (SLAM) has been one of the active research areas in robotic community for the past decade of years. SLAM addresses the problem of a robot navigating and building a map of an unknown environment, without an initial map or an absolute localization means. This paper attempts to provide a comprehensive overview of the SLAM problem. Successful SLAM implementations using laser, sonar and radar can be found in the literature. However, recent extensions to the general SLAM problem has looked into the possibility of using 3-dimensional features and the use of vision sensors. We will focus on these two approaches to the SLAM problem using vision: one with single or monocular camera and another with stereovision. Current applications and future challenges will also be discussed.
