Procedural content generation

Abstract :

Creating large artificial environments for testing robots in simulation can take a lot of time when done completely manually. Procedural content generation however can automate such a task when it is not the goal to match a real existing environment exactly but to have an artificial environment that has a natural structure. Such environmants can be parameterized, so that they can easily be adapted to certain needs just by changing a few parameters. It is also possible to base such a content generation on base data from a real environment like a map of a city or an elevation map of a moon crater and then procedurally add details that are not present in the original data, like details on buildings or little irregulatities on the surface of a crater.