Relative-Absolute information for simultaneous localization and mapping

Abstract

In this paper, a new algorithm, Relative-Absolute SLAM (RASLAM), is proposed to resolve the simultaneous localization and mapping problem. RASLAM utilizes relative and absolute information to estimate the robot pose and map in uncertain environment. Moreover, by combining different kinds of metric SLAM techniques, such as scanning matching, occupancy grid map, and feature-based SLAM, RASLAM can be applied to various unstructured surroundings. The experimental results show that RASLAM can achieve excellent performance even without odometry information in both outdoor and indoor environments.