A Review of Robotic SLAM

Abstract

Since its inception in the late 1980s, the process of simultaneous localisation and map building (SLAM) has become a key subject of discourse amongst the robotics community. Many consider it to be instrumental to autonomous mobile robot navigation in an a priori unknown environment, especially when planning efficient and purposeful trajectories. But due to data association uncertainty, navigation error (e.g., odometric drift) and sensor noise, SLAM has proven to be a complex problem. This paper first describes the SLAM problem and then reviews the current state of the art in solving it with regard to real-world operation.