Map generation in unknown environments by AUKF-SLAM using line segment-Type and point-Type landmarks

Recently, autonomous mobile robots that collect information at disaster sites are being developed. Since it is difficult to obtain maps in advance in disaster sites, the robots being capable of autonomous movement under unknown environments are required. For this objective, the robots have to build maps, as well as the estimation of self-location. This is called a SLAM problem. In particular, AUKF-SLAM which uses corners in the environment as point-Type landmarks has been developed as a solution method so far. However, when the robots move in an environment like a corridor consisting of few point-Type features, the accuracy of self-location estimated by the landmark is decreased and it causes some distortions in the map. In this research, we propose AUKF-SLAM which uses walls in the environment as a line segment-Type landmark. We demonstrate that the robot can generate maps in unknown environment by AUKF-SLAM, using line segment-Type and point-Type landmarks.