Hi Francesco,
Yes we have done a preliminary analysis on the model of the drive system. We built a test-stand using one full actuator (motor and pulley-belt)
and generated random motions and used a simple controller to track them. Here is one set of the motion:
then we tried to fit a model based on inertia, viscous and Coulomb friction and here is the plot of the contribution of each term to the torque required for realising this motion
and here is the parameters based on a model that has the form 𝜏=𝐽𝜃¨+𝑏𝜃˙+𝑓1arctan𝜃˙ (we used arctan as an approximation of the sign function), the inertia parameter is equal to J=0.00000632 kg.m2, Viscous friction is equal to b=0.00000011 kg·m−1·s−1 and Coulomb friction parameter is f1 = 0.00447564 N.m. Following plot shows that this simple model can we ll approximate the torque required for this motion