Research On Environment Modeling For Mobile Service Robot

The mobile service robot is the hottest spot in the world's robotics. The environment modeling technology of mobile robot is an important research direction in the field of robot navigation. A map built by environment modeling has guidance significance to the robot localization and path planning. Binocular stereo vision has been more and more used in the robot environment modeling because of its low cost and rich information. A number of key technologies of mobile service robot environment model such as three-dimensional camera calibration, binocular visual match and the construction of environment map were researched. And the main work is as following:Firstly, the robotics Research about binocular stereo vision and environment model algorithm in domestic and foreign are summarized. Camera calibration technique of the binocular stereo vision is researched. An improved harris corner detection algorithm is proposed and is better than ordinary harris algorithm in accuracy through experiment. The inside and outside parameters of the camera are obtained by Plane-based camera calibration method of grid points. It has high accuracy with verification by experiment.Secondly, the matching technology of the binocular stereo vision is researched. The four steps of the SIFT (Scale Invariant Feature Transform) algorithm to extract feature points is analyzed. The feature points obtained by SIFT (Scale Invariant Feature Transform) algorithm and the harris algorithm is compared by experiments. The feature point matching algorithm is researched and SIFT feature points is matched by experiments.Finally, the programs that using SIFT feature point to build natural landmarks are designed. The SIFT feature point's difference and three-dimensional visual information are obtained through the binocular stereo vision matching experiments and used to construct a feature map, grid map and topology map. The three maps were compared and good experimental results are obtained.