Three-Dimensional Robotic Mapping

This paper presents an approach to the generation of three dimensional maps of complex environments that exploits improvements in vehicle location estimation provided by the Simultaneous Localisation and Mapping (SLAM) algorithm. It is shown that the use of the SLAM algorithm is able to improve the estimated vehicle path, resulting in a higher fidelity model of the environment. The map complexity and the effects of noise in the sensing and localisation processes are reduced by fitting planes to regions representing flat surfaces. Results are shown for the approach applied to a mobile robot equipped with a pair of scanning laser range finders operating in an indoor environment.