Virtual Experiments on Mobile Robot Localization with External Smart RGB-D Camera Using ROS

Abstract

Precise robot localization is important for all mobile robots, which has to deal with accumulating odometry errors, onboard sensory noise, harsh environmental conditions, unstable or missing GPS signal, and absence or uncertainties of a global map. Yet, localization is considered in Smart Environments applying dynamic connectivity with external local sensors within the Internet of Things (IoT) paradigm. This paper presents experimental results of robot indoor localization using a single external smart RGB-D camera. The virtual experiments were performed in the ROS Gazebo simulator with Turtlebot3 Waffle Pi mobile robot model. Three types of robot motion within a virtual office environment were considered: static state, linear motion, and three different cases of curvilinear locomotion. In all cases, external RGB-D camera usage allowed to obtain a reasonably accurate location of the robot.