As most of us here are in the technological sector we are often blinded by the idea of unending linear progress, but we really have to respect how hard robotics actually is. The current manual labor tasks that have to be automated are optimized are actually not that 'repetitive' in the scientific sense, they actually require a lot of elaborate limb movements, split-second decision makings, and handling of countless edge cases for even just the simple task of moving objects from point A to B. It's easy to manage this when you're in a warehouse that has been purposefully designed for robots to do things easily (with entirely flat floors and standardized box sizes). But as soon these robots meet environments where humans live, come and go, with all the irregularities and edge cases it has, it will fail miserably. Too much of our surrounding infrastructure are designed for humans and not for robots, that's why there is immense value in creating robotic systems mimicking natural systems. But robotics researchers are still having an incredibly hard time replicating the delicate movements of human limbs and hands (both in terms of hardware and software), and rightly so; the human body is an incredibly elaborate system of muscles, bones, and neural pathways that almost seems like a miracle of nature (how the fuck did evolution really create this insane complicated system that can do all sorts of weird movements???) And even if we know how the whole system works, it's not a easy task to replicate this in mechanical terms. Note that we haven't begin talking about mass production of these complicated robots, there's always a question of whether it will scale (and would it be more competitive to humans, especially in the developing world?). I think robotics will still have quite some progress in the 21st century but a bit pessimistic that it will seriously begin to replace traditional manual labor in my lifetime.