That was interesting, thanks for posting. What was most impressive was not so much what the robot could do, as the intellectual effort that was required to enable it to do it. Even when we think about planning an action, we don’t have to think about including steps like “after picking up the cup, make sure that the cup has been picked up” although we must be doing this unconsciously.
Also, I’m bemused by the prospect of the robot having to go on to the internet to ask other robots how to make tea. Wouldn’t it be better if it could ask the person who wanted the tea, how to make the tea?