Actually, the barrier is price. An "AutoBurger5000" could be made with current tech, but there would be costs associated with R&D in order to implement it. If the resulting device costs half a million with say 10% or 50k annual operating costs, it will never pay for itself by replacing one shift slot at 7.5/hr (7.5 x 20hours/day x 365days/year =54k/year). Well, at those numbers, the break even cost is 125 years (500k/(54k-50k)), which is waaay too long of a timeframe to contend the implicit assumptions are valid. Double that wage to 15/hr, and the break even point is 9 years. Still too long, but within the realm of reality. If you could deliver the AB5k for say 250k, even with operating cost of 50k/year, the break even is under 5 years at 15, whereas it's still over 30 years at 7.50. At five years, there would be crazy demand.
Of course, I'm simplifying: the costs of a worker slot are more than just hourly wage. There's taxes, health insurance, vacation, insurance, etc. And I suspect I may be low-balling the per unit cost of the necessary device as well. But the basic math is the same, and going to 15/hr makes the development of such devices much closer to economically feasible.
Now, one place we are seeing replacement of workers is in order placement/checkout. That's a great deal easier, and just became a great deal more viable.