selective keys in the LM optimizer

[Anusna Chakraborty]

Hello All,

I am working with GTSAM to implement a very simple landmark-based localization problem (similar to the localizationExample.cpp file but instead of custom GPS measurements, I want to estimate the 2D robot pose using range/bearing measurements from landmarks). I have used the RangeFactor to add measurements. In order to use the RangeFactor I had to create two different symbols (x for the robot and l for landmarks). Unlike SLAM, I know the exact position of the landmarks and so I do not want their locations to be optimized. 

Is there any way I can only optimize the robot and not the landmarks and still use the RangeFactor?

I would greatly appreciate any help or suggestions you might have on the matter. 

Thanks and regards,

Anusna

[Jay Elrod]

Hi Anusna,

You can use gtsam::NonlinearEqualityFactor on landmarks instead of gtsam::PriorFactor from the examples, and it should provide much less (or zero) fluctuation on the landmarks' poses during optimization steps.

Jay

[Anusna Chakraborty]

Hi Jay,

Thanks for your help. It works with the NonlinearEquality factor. 

Thanks and regards,

Anusna