The December 2022 issue of IEEE Spectrum is here!

Close bar

I Am Still Not Convinced That We Need Androids

Robots that play active roles in our lives don’t need to look like us

3 min read
A series of six pictures showing the head of an android baby making different facial expressions

Nikola is an android under development at Riken, a scientific research institute in Japan. Nikola’s job is to help researchers validate the accuracy of facial expressions, which it does by doing its best to make a bunch of facial expressions that, hopefully, can be identified by humans. It’s just a head at the moment, but it will eventually have body parts, and it’s generally modeled on a male human child “to promote natural interactions with both adults and children,” always a dicey proposition with robots.

This is part of a larger project from Riken, called the Guardian Robot Project, the goal of which is “to develop an autonomous robot that can be close to people and make people feel the ‘heart.’ Once such a robot is realized, it will be accepted by people and will play an active role in every aspect of our homes and society!” But if it’s going to do that, does it really need to rely on an android form factor for those “natural interactions” that Riken is looking for? I’m not so sure.

To be clear, Nikola here is absolutely not the worst android I have ever met. And it’s a research platform, too, so we shouldn’t judge it too harshly. That’s especially true because experiments show that it’s effective at what it’s supposed to do; the paper accompanying this video reveals that people were able to accurately interpret the facial expressions that the robot was making. The reason that this research was necessary is because androids can be tricky to read at times, especially when making expressions associated with negative emotions, which are more difficult to distinguish. And that’s one of the reasons why I’m so skeptical that androids are the best answer for human-robot interaction.

And if I were the Riken researchers, I’d also back off a little bit on the somewhat aspirational claim about robots making people “feel the heart.” Even more dubious is the assertion that they will play an active role in every aspect of our homes and society. Making people feel things is not particularly difficult for robots. There’s a bit of a leap, though, from feeling things to playing an active role in everything. Even now, robots that make us feel things can play very small, targeted roles in our homes and society. This is not to say that such roles aren’t important and effective and valuable, just that (and I’m sorry to have to keep saying this) we really need to make sure that we keep expectations realistic and grounded.

Grounded expectations are especially relevant with robots that are intended to connect with us or otherwise influence our emotions, mainly because using other humans as a benchmark will doom your robot to failure. We’ve seen over and over again that simple robots can be just as effective as complex robots at emotional engagement, without wandering into the Uncanny Valley. That said, check out this graphic from the Guardian Robot website itself:

A three panel cartoon showing a little wheeled robot smiling while helping an elderly man safely stand up

Look at that lil’ guy! How cute he is! And not just cute, but expressive too, with a bare minimum of degrees of freedom. A couple of basic eyes, a minimally adjustable mouth, and there you go. I love him already, and there’s no reason why you couldn’t translate what he’s got going on into hardware, with the help of an artist or animator.

The aims of Riken’s Guardian Robot Project extend far beyond the Nikola android, and yet the project appears to be in its very early stages. So, there are plenty of directions where things can still go. I appreciate that they’re doing the basic research first and exploring different form factors; we’ll see what they ultimately come up with.

The Conversation (4)
Ian Thompson21 Mar, 2022

This is one of my biggest pet peeves in consumer / healthcare robotics. The original idea was sound, octogenarians aren't going to react well to a robot, so make it look like a human so they are more comfortable. And while I have to acknowledge the studies done based on childcare and the value human like expressions / stature might have, I do not think a humanoid / android is a must or even a recommendation. In 2022, I don't think we are worried about that "human-heart" feeling. Humans are limited in what we can do, but we can do it as well as we can give our bodily dynamics. Why limit ourselves? Robots should do the things we cannot do, or the things that are difficult to do with dynamics that are not limited by our bi-pedal motion. 3 legs, 4 arms, 1 leg, whatever it might be, let the robots be designed better than us for the applications that will benefit them and us.

David Stocker15 Mar, 2022

The "uncanny valley" is real, and apparently quite difficult to bridge. Personally, I would be much more comfortable interacting with a Rumba vacuum than a robot with faux human appearance. (And I have spent nearly my entire engineering career working in machine automation and robotics.)

Isabel Bowman10 Mar, 2022

Give me a Wall-E, a TARS, a Gladys, even, anything but more godforsaken humans.

The Transistor at 75

The past, present, and future of the modern world’s most important invention

2 min read
A photo of a birthday cake with 75 written on it.
Lisa Sheehan

Seventy-five years is a long time. It’s so long that most of us don’t remember a time before the transistor, and long enough for many engineers to have devoted entire careers to its use and development. In honor of this most important of technological achievements, this issue’s package of articles explores the transistor’s historical journey and potential future.

Keep Reading ↓Show less