The Localization And Navigation Research Of Mobile Robot Based On RGB-D

Robot has quickly gone into our daily life. People hope that use the research of positioning and navigation to improve the ability of mobile robot autonomous to walk, so as to further enhance the robot application value. Currently, using a laser sensor for autonomous mobile robot localization and navigation has been realized. But, the simple characteristics of this method cause a lot of positioning error on localization and navigation. At the same time, the sensor is expensive and bulky.This paper describes a SLAM system based on RGB-D sensor. The biggest advantage of this system is use RGB-D sensor to acquire depth information for 3D reconstruction. In view of the problems of bigger position error, using the extended kalman filter algorithm improve the accuracy of estimation and positioning. Firstly, the SURF algorithm is applied to process feature point extraction and matching for two-dimensional image gathered by the sensor. We put the 3D data that is the corresponding two-dimensional image matching point registration and complete the 3D reconstruction of environment around the robot. Then we analyze the classic filtering estimation algorithm, and use it to optimal estimation for mobile robot localization. Finally, we carry on a test in the indoor environment map. When a given target site location, the robot realize the autonomous positioning and navigation by using the method.