Efficient Mapping Technique under Various Spatial Changes for SLAM-based AR Services

Abstract

Recently, many attempts have been made to apply real-time simultaneous localization and mapping (SLAM) technology to augmented reality (AR) applications. Such AR systems based on SLAM technology are generally implemented by augmenting virtual objects onto a diorama or three-dimensional sculpture. However, a new SLAM map needs to be generated if the space or lighting where the diorama is installed changes. This leads to the problem of updating the coordinate system each time a new SLAM map is generated. Updates to the coordinate system signify that the positions of the virtual objects placed in the AR space change as well. Therefore, we proposed a SLAM map regeneration technique in which the existing coordinate system is maintained even if a new map is generated.