Pinhole camera factor for localization only

[jlblancoc]

Feature

Having a factor where the pose of the camera is the only variable, with fixed landmarks and calibration.

Motivation

Vision-based localization / tracking / etc.

Additional context

I have already written such a factor by modifying the existing GenericProjectionFactor... would it be useful to integrate it into GTSAM? If yes: any preference on its name?

[dellaert]

There is also one in unstable, and @lucacarlone is creating another one. We could probably benefit from coordinating :-)

[jlblancoc]

👍

In gtsam_unstable, I found ProjectionFactorPPP and ProjectionFactorPPPC, which estimate more variables (body-camera pose), but landmarks are still an unknown too. I'll let Luca to decide whether his proposal in #696 makes this one #697 unnecessary.