Deep, Deep in the uncanny valley lives this Japanese humanoid robot, aptly called an 'actroid'. She can function autonomously, talking and gesturing while interacting with people. While her appearance may not be as hyper-realistic as her cousin Geminoid F, her interaction is autonomous instead of tele-operated. This makes her at the same time more effective and a bit creepier.
As you can see in the video, the actroid is not very convincing as a real human, yet as an interactive wayfinder, she does quite well. It is impressive that she can do speech recognition in a crowd of school children, direct gestures to the person speaking, and can even handle a quip. Child: "You are wise!" Actroid: "No way, you are much wiser than me".
I believe that there lies the future of humanoids: The goal should be to have them function as well as possible, in a task that is actually useful. An example is the robotic school teacher above, which is only tele-operated, but has a worthwile and important task. When a humanoid has a clear reason to be there, their presence is a bit less intimidating.
By changing the raison d'être of humanoid robots, will the 'uncanny valley' shift? Or should we change the appearance of autonomously interacting robots to avoid anthropomorphobia altogether? Funny enough, as the horrifying example of the Telenoid shows, the answer does not necessarily lie in making the robots look less human. Perhaps humanoids should leap right over the uncanny valley, and we should start seeing them as real people?