Closed sara9915 closed 1 month ago
Hello, @sara9915. Are you referring to the integration of the model dynamics?
Yes, exactly, but during the predictions along the horizon, not after the optimization
Ok, the library is handling things differently between linear and non-linear MPC. Which one are you interested in?
Non-linear MPC
If you already have a discrete-time model, the library is simply propagating the state x(k+1) = f(x(k),u(k)). Otherwise, if you provide a continuous time model, the integration is performed using central difference with the integration step of the sample time provided. In this case, the integration exploits the prediction horizon to implement the integration scheme.
I got it, thank you!
Hello @nicolapiccinelli, I'd like to use your library, but I need to employ a custom numerical integrator to make the MPC predictions along the prediction horizon. Can you give me some info about the routine now used to do this? I see there is an RK4 class, but I didn't find anywhere its usage in the provided examples.