It depends on how broadly you are willing to count as having made one. Super crude examples have already been made at home and can lift things. Power consumption and size are big issues but the largest one is that without really good software the person wearing it will have to fight against the exo-'suit' so much that normal movement will take more energy. One of the big improvements touted in the article is just that, the prosthetic got better software so that it predicted how the user wanted to move better.
What's the hard part? Aligning to your leg? Powerful enough motors? Software?