An integrated virtual environment for feasibility studies and implementation of aerial MonoSLAM

This work presents a complete framework of an integrated aerial virtual environment (IAVE), which could effectively help implementing MonoSLAM (single-camera simultaneous localization and mapping) on an aerial vehicle. The developed system allows investigating different flight conditions without using any preloaded maps or predefined features. A 3D graphical engine integrated with a full 6 DOF aircraft dynamic simulator together with its trajectory generator completes the package. The 3D engine generates and accumulates real-time images of a general camera installed on the aerial vehicle. We effectively exploit C++ to develop the 3D graphics engine (3DGE) and all its associated visual effects, including different types of lighting, climate conditions, and moving objects. The existing 3DGE exploits the so-called Frenet Adapted Frames (FAF) with constrained angular velocities that is very effective in motion modeling of both ground and aerial moving objects. An in-house-developed MATLAB GUI puts into service the offline MonoSLAM system, which is very user friendly. The current version of IAVE effectively employs the so-called Inverse Depth Parameterization notions for features' depth estimation in monocular SLAM, where different case studies show its dependable results for low-cost aerial navigation of a general aviation low-speed aircraft.