What does 'res_stat' field mean in residuals and why after the first SQP_RTI iteration it is always the same for different problem configurations

Hi all,

I have OCP for path tracking task. I use SQP_RTI for this and I noticed that I am getting the same res_stat residuals after the first iteration no matter which constraints and config I set.

I have attached two screenshots with different configurations of my problem. As you can see I have the same value 9.88000000e+01 for them after the first iteration.

Honestly, speaking I don’t understand deeply what ‘res_stat’ field mean, so I cannot conclude it is ok or not.

Thanks in advance!
Alex

Hi Alex,

res_stat is the infinity norm of the residual of the stationarity condition within the KKT conditions.
This residual is the some of the gradient of the objective function plus the weights of the multipliers times the gradients of the constraints.
The multipliers are initialized with zeros by default in acados.
Thus, what you report is expected behavior.

Cheers,
Jonathan

1 Like