Using 3D Visual Landmarks to Solve the Correspondence Problem in Simultaneous Localisation and Mapping 1Michael

We present an approach which uses 3D visual landmarks for solving the correspondence problem in Simultaneous Localisation and Mapping (SLAM). The 3D landmarks are computed from camera views of the robot’s local space. Using multiple 2D views, identified landmarks are projected, with their correct location and orientation into 3D world space by scene reconstruction. As the robot moves around the local space, extracted landmarks are integrated into a scene representation for the local space which comprises the 3D landmarks. The landmarks for a local space’s scene are compared against the landmarks for previously constructed scenes to determine when the robot is revisiting a place it has been to before.