Developing autonomous navigation algorithms without using extern positioning signals is the key to allow Multi-rotor UAVs to fly in complex environments where the GNSS signals are week or invalid. However, it is an expensive and time consuming process to develop and test autonomous navigation algorithms with real UAV platforms in the real world. It becomes even more difficult when developing autonomous navigation algorithms based on deep learning techniques, since it requires to collect a large amount of annotated training data. To address such problem, we developed a simulation platform based on Unreal Engine, providing physically and visually realistic simulations, to validate and test navigation algorithms for Multi-rotor UAVs.