“…Moreover the profile h(w) will be assumed to be polynomial in w: hence the class of signal considered is no more restricted to a simple purely linear case, but will take into account a polynomial combination of constant and sinusoidal signals with unknown frequencies, amplitudes and phases. All those assumptions allow us to cast the tracking problem as a problem of output regulation, (see (Byrnes et al, 1997b), (Gentili and van der Schaft, 2003)) complicated by the lack of knowledge of the matrix S (see (Serrani et al, 2001), (Bonivento et al, 2004b)), and suggests to look for a controller which embeds an internal model of the exogenous signals, augmented by an adaptive part in order to estimate the characteristic frequencies. In order to show how the port-Hamiltonian formalism could be really helpful to describe the problem and to find an elegant solution, it is now possible to define a change of coordinates and introduce a new error system again fitting in the port-Hamiltonian framework: we will point out that the tracking problem is now cast as a regulation problem complicated by the presence of exogenous "virtual" disturbance signals.…”