Extraction of Geometrical Features in 3D Environments for Service Robotic Applications

Modeling environments with 3D feature based representations is a challenging issue in current mobile robotics. Fast and robust algorithms are required for applicability to navigation. We present an original and effective segmentation method that uses computer vision techniques and the residuals from plane fitting as measurements to generate a range image from 3D data acquired by a laser scanner. The extracted points of each region are converted into plane patches, spheres and cylinders by means of least-squares fitting.