Was just discussing this with an engineer friend of mine yesterday.
He was commenting on how many menial jobs will soon be replaced with automation. You know, fully automated restaurants etc.
But it's not just manual labour - think of the auto-generated news aggregators etc. that are becoming better and better because of access to Big Data. Administrative and even engineering jobs will be increasingly replaced by AI and automation.
My job will likely be replaced by a script.
As far as the robotics world goes, I think the biggest issues are currently the same as are faced by the automotive and computing industries - power and heat management. And then there's our increasing reliance upon sensors, and filtering out the abundance of electromagnetic (as well as optical) interference in order to preserve signal integrity.
My angle on the discussion was that it was too cost-prohibitive to employ these concepts usefully in the broader consumer markets - we simply won't pay the premium for a machine that can reliably do what we can do ourselves. If I'm going to have to invest more time getting an automated vacuum working properly than it takes me to actually vacuum the house myself, why would I spend the money (other than the "cool" factor, that is!)
Not to mention that the device would require maintenance and repair.
The higher-end, more reliable products are likely to be purchased by enterprises with deeper pockets, such as military or large industrial companies. Where their usefulness is realized best is in hazardous environments etc.
But it's interesting to see how improvements are being made in bottlenecked areas.