RGB-D sensor based mobile robot SLAM in indoor environment

Abstract

A simultaneous localization and mapping (SLAM) approach based on RGB-D sensor in indoor environment is presented in this paper. The S-NARF feature of point cloud was extracted, and then used for point cloud registration. In particular, a semi-random search strategy is proposed for the loop closures detection. The poses estimation was optimized under the g2o framework. The experiments demonstrate that the approach can effectively solve SLAM problem in indoor environment and the poses estimation have met the demand for precision.