What are open problems in 3D reconstruction, SfM, Visual SLAM

Hi everyone, I will be working for my master thesis on this area of computer vision. I am reading papers and trying to understand where the state-of-the-art is in the field. Like what specific problems are scientists trying to address that limit the performance of the current algorithms?

From what I see, non-static scenes and pushing real-time inference seems to be big challenges worth tackling. What other obstacles are currently limiting in your opinion? What problems do you believe are …

3d reconstruction computervision slam visual slam