Distance (or BBox3D) estimation using RGBD camera with GSAM or similar

[maciejmajek]

Is your feature request related to a problem? Please describe. RAI agent could benefit from better spatial reasoning.

Describe the solution you'd like Image -> 2D segmentation mask (open set) -> Masked depth data -> Distance estimation

Relevant links:

• https://github.com/ycheng517/tabletop-handybot