Procedural Generation of Underground Environments for Gazebo

Abstract

One of the challenges in the field of autonomous robotics is the thorough testing of the navigation capabilities of the different methods in a wide variety of situations. Towards this end, simulation plays an ever increasingly important role as the improvements in the technology keep closing the gap between reality and simulation. This work presents a scalable, efficient and easy-to-use framework for generating random underground environments. The system is based on the procedural placement of discrete modules in such a way that the end result are sprawling and complex networks of tunnels that can be used for testing robotic applications for underground environments.