Three-dimensional mapping based on SIFT and RANSAC for mobile robot

In recent years, mobile robots are playing an increasingly important role in the rescue work. Relative research has been widely attention. Mapping of the environment and navigation research become more important when robots need to move in a complex environment. Two-dimensional(2D) mapping can not be remarkable in robot navigation due to the limitation of information access, while laser-based 3D mapping always owns high cost. Besides, the accuracy of ordinary 3D mapping cannot meet requirement. In this paper, a method is presented to realize the low cost 3D-environment-mapping. We combine the two-dimensional navigation algorithm based on extended Kalman filtering with the three-dimensional mapping based on SIFT algorithm. The 2D laser ranger cooperates with Kinect sensor. Experimental result shows that the method can guide the robot to explore independently in the unknown environment with good performance for mapping and navigation.