There's a massive opportunity for a startup to move towards the personal/home robot market and make a killing, without any government help nor without some fabled visionary type person.
Google sought what were essentially military robots from BD. Those machines, if ever turned into a product, would have been tens of millions of dollar each. They make no sense and google cannot solve the problem of loud servos/actuators and the relatively low power density of battery technology. Heck, some BD demos I've seen run on gas engines. When the DoD is your customer, the last thing you want to sell them is a tool that'll give away their soldiers' positions.
I've seen ROS projects on commodity hardware that's 75% of the way to a decent home robot. I think this is doable within the next 10 years. Having a home robot will be like having a smartphone today. We'll wonder how we got along without them. Hell, I'm half-tempted to get into this field myself.
Everything about robots from these big companies has been "big picture" bullshit; big spending, big PR releases, big promises, untried tech, and sadly lots of questionable patent filing sprees. None of these big companies are interested in making a real affordable product it seems, or believe it to be too difficult or unprofitable. Its like they're trying to sell mainframes in 1980 and the PC Jr, Commodore64, Apple //, etc are soon going to eat their lunch. The big iron approach failed in computing the same way its failing in robotics.