Abstract
In this paper, we propose a 3D extension method to planar map information in 2D SLAM from visual information using RT middleware. Extension for 2D SLAM software is realized by cubic shaped object detection with stereo vision to detect 3D obstacle. 3D objects are described as primitive cube to apply another task such as manipulation. Experimental results shows 3D extension can detect obstacles that is impossible to detect by a 2D range sensor.