SLAM for Planar Mobile Robot

Abstract

For faster process speed and higher accuracy of RGB-D SLAM of planar mobile robot, the new method improved the traditional method of ORB feature detection and match based on the image segmentation. Our method considering the features of planar robots, simplify the feature match process and PnP pose transform matrix of neighbor fames. It evaluates the motion planar of robot while SLAM system running, which is used for the optimization and simplification later. Using color images and cooperating depth images, it can easily get the height of each extracted feature points and divide those points in several groups to reduce wrong match and accelerate match speed.