Robot navigation is a contemporary field of research in which the optimal path a robot could take has to be computed with decreased computational costs. The robot should then move as enduringly and efficiently as possible by taking the area’s different costs into consideration.
The present bachelor thesis analyzes the performance of Move Base Flex and the ROS Navigation Stack into which the Grid Map library was integrated. Grid Map is a library for 2.5D map representation while Move Base Flex and its plugins make use of these maps for robot navigation and controlling. The Navigation Stack is a bundle of libraries in ROS which is the Robot Operating System providing several tools for robot navigation \cite{ros}.
The current plugins written for the ROS nav_core API use the Costmap2D library for their map representations and its functionality has to be replaced sufficiently. Therefore, Grid Maps difference in functionality and behavior compared to the older Costmap2D and how these could be overcome is shown. Moreover plugins implementing the nav_core plugin interfaces were adapted by this project in a way that they are using the Grid Map library instead of Costmap2D. A Move Base Flex Navigation Server Node, Plug-in Executions and the Move Base Flex’ plugin interfaces were to be implemented to provide usage with Move Base Flex for the Grid Map plugins. Furthermore ROS’ map server does not support Grid Map. As of this fact a Gridmap Handler was developed which initializes the static map information in a global map and continually updates the dynamic information of moving obstacles in a local map. It also periodically processes the maps data in terms of data merging and obstacle inflation by the robot’s inscribed radius. Finally, the advancements that have been made are concluded by what is still to come.