Any robust optimization technique available?

[Ilinca Burdulea]

Hello,

I am interested to know whether any robust optimization techniques (switchable constraints, dynamic covariance scaling, maximum mixture) have been implemented in gtsam. I couldn't find any mention about this topic in the google group but I find it unlikely that these are niche topics in SLAM...Does someone have experience with any?

Looking forward to your reply, thank you!

Ilinca

[Matías Mattamala]

They are all defined in the LossFunctions file: https://github.com/borglab/gtsam/blob/2af2c966d112efd5dc8cbf1b1252040f9d010d59/gtsam/linear/LossFunctions.cpp#L34 
This blog entry shows how to use them: https://gtsam.org/2019/09/20/robust-noise-model.html
You also can find an example for Graduated Non-Convexity here: https://github.com/borglab/gtsam/blob/2af2c966d112efd5dc8cbf1b1252040f9d010d59/examples/GNCExample.cpp