2A1-O06 Indoor SLAM using wall as prior information(Localization and Mapping)

Abstract

This paper assumes the wall to be prior information and describes the SLAM technique using proposed Monte Carlo Localization (MCL) which is improved for our study. Because the wall is not structurally changed easily, robot can use the wall information as "wall map" for SLAM. Vertical LRF which is vertically set up on ground is used to extract the wall. We also propose MCL using the measurement value obtained at several times. The particle filter used with MCL has some particles and each particle has a particle filter used for tracking itself. In experiments, it is understood that the method is effective for building 3D map though we did not obtain accuracy like Fast SLAM.