From “For Early Hominins in Africa, Many Ways To Take a Walk” (Science, 4 May 2012), we learn
“Several new studies of incredibly rare fossils of feet and partial skeletons reported at the meeting reveal the complexity of early bipedalism. In a talk, a paleoanthropologist showed how the primitive foot of a still-unnamed species of Australopithecus shared features such as an opposable big toe with the 4.4-million-year-old Ardipithecus ramidus, which suggests that both hominins still spent considerable time in trees. The foot also shared a key trait with Au. africanus, which lived about 2 million years ago in South Africa. Neither of those features is found in Au. afarensis, suggesting that Lucy’s species cannot be the direct ancestor of Au. africanus. That means that a second hominin lineage must have led to Au. africanus, the researcher notes. (Paywall)
It certainly doesn’t sound like we know very much about early bipedalism or lineages. No “Ascent of Bipedalism,” step by step.
It’s not clear that there’s much to learn here if we want to know why humans walk exclusively upright, which is different in principle from sometimes using an upright stance. It means foregoing for good any advantage that running on all fours would provide.