Three dimensional structure estimation and planning with vision and range

A 3D structure estimation approach is proposed. The uniqueness of the approach lies in the fusion of vision and 2D range, which is a common sensor combination for mobile robots. 2D range information can be used for hypothesizing 3D structures, feature association, and improving the estimate accuracy. These problems are difficult to solve if using only vision. The proposed approach will be beneficial to 3D map building in mobile robots. Experimental results validate the effectiveness of the approach.