RGB-D sensor-based visual SLAM for localization and navigation of indoor mobile robot

Abstract

To increase the localization accuracy of the moving service robot in restaurant, A localization method and a landmark recognition method based on RGB and depth (RGB-D) vision sensor is presented. Based on the landmarks, we utilized an Extend Kalman Filter(EKF) method to calculate robot position and pose. The path planning approach based on a differential-flatness point-to-point trajectory planning is proposed for the robot. There are kinds of control manners which improved the enjoyment of human-machine interface, such as mobile client and voice control. Finally, the proposed methods have been verified to be effective and feasible by experiment in indoor environments.