I remember Eric Yundt and I talking about this style of walking a long, long time ago. I'm glad to see someone has finally implemented it. I wish I had had the time and resources to have done it. I'm betting this is the way all walking robots mechanics will be someday.

My opinion is that the future will be a combination of active walking like Asimo, Qrio,... and passive walking like the bipeds of MIT, Delft and Cornell. The big advantage of passive walkers is that they are very energy-efficient, the disadvantage is that they are not able to start and stop walking and the step velocity cannot be changed. So performing tasks is untill now impossible. The adavantage of active walkers is that they can change velocity and are able to start and stop, but they consume a lot of energy. Both sides will grow to each other: the passive walkers will be added with active control, while the active walkers will need passive elements (for example swinging the leg). But then you need compliant joints instead of the stiff joints created by using electrical drives with the necesarry gearboxes. Compliant joints are created using pneumatic artificial muscles, series elastic actuators or the AMASC (An Actuator with Mechanically Adjustable Series Compliance)
It is well understood that leg compliance and the use of passive dynamics plays an important role in human walking and running.