Comparison of 3D Vision Mapping Models

Merits

1. The algorithm constructed in the research paper uses live and real-time data which makes it useful to implement in the UAV.

2. The global 3D map built has information that could be useful for navigation of the UAV.

3. The model is primarily mathematical and hence the only input is the camera feed as well as the INS and GPS coordinates.

Flaws

1. The algorithm was primarily developed for implementation in ALVs and hence it may not be best implemented in a UAV.

2. The local 3D map generated contains very little information on distant points.

3. It requires two cameras built into the UAV along with GPS and INS.

Takeaways

1. No algorithm would be perfectly transferrable onto the UAV. This research provides a good mathematical model to generate a 3D map for navigation.

2. This algorithm, if implemented, isn’t going to be efficient for flights at large altitudes. However, for near-ground flights with obstacles, it should be useful for mapping and navigating the environment.