I can only conclude that knuckle walking is a derived trait that has been lost in humans and retained in related species. Otherwise we are postulating that gorillas first separated, then chimpanzees (and then chimpanzees separated into two species) from the lineage that lead to mankind. Additionally we are postulating that after all of these separations, all of them developed identical structures that lock the wrist. After that, the lineage that led to humans lost this mechanism. We know that "Lucy" was a biped. Why on earth would that species have developed this wrist walking mechanism along with the three other species, only to lose it again. Is it not more logical to understand that the wrist-locking mechanism was vestigial in "Lucy"?
There are less complicated and tortorous routes which would acheive the same thing. The common ancestor of humans, chimps and gorillas would have been a knuckle walker. Whilst the ancestors of gorillas retained this feature, the ancestors of chimps and humans began to develop bipedality. Some of these bipeds, however, which would include the ancestors of chimps, began to get back down on all fours as their ecology demanded, and whilst they still retained enough of a knuckle-walker's anatomy to not make this prohibitvely unlikely. They'd still have the protusion on the radius, as demonstrated by the fact that it's retained in bipedal Lucy, so there's no need for this to be evolved again.
I don't know how realistic this scenario would be based around the anatomy of the fossils we know, but I don't think humans and chimps sharing a bipedal ancestor requires the wrist-locking mechanism to evolve three times independently - it still only needs to arise once. Knuckle-walking as a whole could have evolved once, been lost once, then been subsequently regained once, by an ape still in possession of some of the features that made their ancestors successful knuckle-walkers.