AR Tags Based Absolute Positioning System

Abstract

The presented work describes the implementation of a system for positioning a mobile robot in a workspace using an AR (Augmented Reality) tag system. This was realized by means of a stereovision camera placed on top of the robot, which, with the aid of a gimbal (a two-joint camera reorientation system), had the ability to dynamically rotate in order to control the entire area around the robot. Markers with predetermined identification numbers were placed at specific locations in the robot’s work area. A topological map of the workspace was created. The robot determined its position based on this map. A simulation and an actual model of a 6-wheeled robotic platform were used to test the stereovision system. The experiment results came from the simulation performed in Gazebo software designed for such analyses and adapted to work with ROS (Robot Operating System) ecosystem. The results of the algorithm’s performance for both the simulation and the actual robotic platform were compared with each other. To verify the algorithm’s performance, tools available in both ROS and Gazebo (including rviz) were used.