Might be interesting

[h106frp]

Stumbled on this which might be interesting;
http://www.lista.it/atnet/casmtt.htm

and this; interesting note about low speed riderless stability weave, wobble and capsize- i wonder which model GPB uses 
http://www.lista.it/atnet/casperen.htm

[HornetMaX]

and this; interesting note about low speed riderless stability weave, wobble and capsize- i wonder which model GPB uses 
http://www.lista.it/atnet/casperen.htm

Which note are you referring to exactly ?

[h106frp]

The last paragraph of  http://www.lista.it/atnet/casperen.htm, it seems that an established model claims that a bike with no rider is stable much like we see in GPB, but their analysis found that it should fall. Maybe PB is just correctly using the established model and depicting the result correctly  I guess they are referencing this author http://www2.ee.ic.ac.uk/publications/p1408.pdf

[HornetMaX]

As far as I can see, GPB uses the same approach described on the italian site.

But the wording is weird. This is the english sentence:

Sharp shows that capsize is well damped at low speed ; on the contrary WM3D shows that the mode always is unstable especially at low speed (the motorbike falls, as in reality).

And this is the corresponding in italian (their own wording, I'm not translating):

In particolare si è dimostrato che a basse velocità al contrario di quanto ottenuto da Sharp il motoveicolo è instabile e cade lateralmente in un tempo molto basso.

Notice however that the "Sharp" results quoted in the article are very old (1971 ans 1976): they involved an explicit expression of the bike dynamics, with equations obtained by hand !!
At any rate, if the "Sharp" model had an almost stable capsize at very low speed, it was just wrong (or it included some approximations that made it not valid at very low speeds).