Pole vector has to be on the plane of the bend. If the prefered position i positive, you can jsut simply position the pole vector at the knee maybe add one fo the vectors from bend judiciously so its not really on top of it.
Can v write the steps to find pole vector ctrl's position as a MEL procedure?
Sure bbut i would need to know what your naming conevntiona and bone structure/bind position is to quantify.