Why does ISS have two humanoid robots (rather impractical for microgravity) instead of something not human shaped?

A robot with "RCS" composed of a set of propellers could efficiently navigate the inside of ISS, aiding with "fetch" tasks, inspections and experiments (which could be performed remotely from Earth through the robot instead of involving the overworked crew) and essentially be of broad use as a mobile unit, unlike the robots on the station which are working in fixed locations.

A small remote-controlled drone with some simple rocket engines (hydrazine, ion, even compressed gas), reaction wheels, camera, and a set of tools/manipulators could be stationed permanently on the outside of ISS and save weeks of cumbersome EVA time by performing inspections, simple repairs, assisting astronauts during EVA, or even flying a couple kilometers away for photoshoot opportunities, examining (or capture) of objects like the loose antenna cover - or even serve as a makeshift EMU unit to pull astronauts around.

Dedicated manipulators with tools would be more useful than fake arms, never mind the completely useless legs. When I read Chris Hadfield's AMA, I couldn't help an impression he was underwhelmed by KIROBO and the Robonaut. OTOH, he was quite enthusiastic about Canadarm 2, which is a genuine help for the astronauts.

So, why the decision to go with totally impractical "humanoid" option?

  • 1
    $\begingroup$ Jets are problematic for moving around in proximity to the ISS, parts are super fragile and contamination sensitive with keep-out-zones all around it. Some kind of spider that could clamber around might be better. Don't discount the paranoia factor though. Have you ever seen the speed at which Dextre operates? Or the planning that takes place to ensure nothing is bumped? Letting an autonomous robot loose on the ISS would take a major cultural change in ops. $\endgroup$ Commented Dec 11, 2015 at 16:39

1 Answer 1


There are other robotic tools similar to what you describe being developed for assisting astronauts such as the SPHERES project at MIT.

The primary reason Robonaut is humanoid is that it can essentially use tools and grab onto components exactly how an astronaut would -- meaning no special considerations are required when designing components for servicing (if it can be serviced by an astronaut then it can be serviced by Robonaut). To quote NASA more elegantly:

The value of a humanoid over other designs is the ability to use the same workspace and tools - not only does this improve efficiency in the types of tools, but also removes the need for specialized robotic connectors.

Another reason is out of some concern that astronauts need a robot that is less intimidating and easier to work alongside. Quoting this research paper by the creators:

Having a head on Robonaut gives human co-workers a feel that Robonaut is more human-like, allowing it to more readily become one of the team.

In my opinion this is not the way to go, but NASA insists that there is a major psychological barrier when it comes to astronauts trusting and working with robots. As far as I have seen, Chris Hadfield's similar reaction is common among the astronaut corps. Canadarm, Canadarm 2, and especially Dextre are entirely invaluable for operations on the ISS and supporting EVA servicing -- the astronauts work with them fine and they aren't humanoid in the least bit.

Who would you trust?...

enter image description here

enter image description here

enter image description here

  • 3
    $\begingroup$ I have a feeling nobody at NASA read up on "Uncanny Valley", if the result is astronauts like a 17-meter crane better than the humanoid robot :) $\endgroup$
    – SF.
    Commented Dec 11, 2015 at 16:07
  • 4
    $\begingroup$ Robonaut is not humanoid enough to trigger the Uncanny Valley effect. $\endgroup$
    – Hobbes
    Commented Dec 11, 2015 at 16:09
  • 10
    $\begingroup$ The creepy factor increases a lot when the legs are attached. spaceflightnow.com/falcon9/009/140416robonaut/r2_400533.jpg $\endgroup$ Commented Dec 11, 2015 at 16:34
  • $\begingroup$ Interestingly, NASA arguments for humanoid robots were predicated by Asimov and other SF authors in the 1970s and before, but when real industrial robots became a thing, they weren't humanoids. It seems than building a hand able to grab a screwdriver is way more complex than building a screwdriver with a robot dedicated handle. $\endgroup$
    – Pere
    Commented May 25, 2020 at 12:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.