Clarification on ISMC

[Manouselis]

Hey @evan-palmer . I am reading through your Angler article and at some point you note:

assume that the initial error state dynamics exist on the manifold describing the desired error dynamics (i.e., s(0) = 0).

This makes sense as you want to stay on the sliding surface from the get-go in order to guarantee robustness from t=0. However, based on how you defined the sliding variable you would need the proportional gains K_p to be equal to identity in order to get s(0) = 0. Looking through your code (i.e., control_integration/config/bluerov2_heavy_controllers.yaml), I see that the K_p is defined as [10.0, 10.0, 6.0, 3.0, 6.0, 10.0] which will in turn make s(0) not equal to 0.

Am I missing something?