I’m having an issue where “PARTIAL_CONDENSING_HPIPM” and “FULL_CONDENSING_HPIPM” work nicely, but “FULL_CONDENSING_QPOASES” fails specifically in the presence of soft constraints. I’ve reduced the OCP to something very simple - linear dynamics, one state and one input variable, no nonlinear constraints, linear LS cost.
I’ve uploaded this example here:
Specifically, when constraints are softened (soft_constraints = True) and the number of time steps is large enough (N >= 12), FULL_CONDENSING_QPOASES does not converge and eventually (usually on the first or second iteration) throws Error 37 ‘RET_INIT_FAILED_INFEASIBILITY’. Everything works fine with hard constraints, so the OCP shouldn’t be infeasible.
Any idea what the problem might be? Thanks for your help!