I am willing to train a visual navigation algorithm with Sphinx + Parrot Anafi. To this end, I would like to build a training set that has multiple situations so that the algorithm doesn’t overfit to the ones I manually build.
Therefore, an automated way of doing this is by generating random parametrized use cases i.e. a world with an office with two doors and two pedestrians walking around. Is there any tool or functionality to bootstrap this? Do you know any example already done in Sphinx/Gazebo similar to the maze generation algorithm?