On computing flat outputs through Goursat normal form | IEEE Conference Publication | IEEE Xplore