The Wild (emotional) Robot

[Spoiler Alert]

A robot head, after Roz with only two circular eyes.
Robot head image by CS4FN

If you are in the business of designing robots for the coming, personal robot dominated world, who would you employ? Robotics engineers naturally, AI experts, computer scientists … and of course artists, animators, actors and film makers! Why the latter? Well, they have been designing robots far longer than either robotics engineering or computer science has even existed as a career. What they might add is illustrated (literally) in the animated film ‘The Wild Robot.’

The Wild Robot tells the story of a newly dispatched personal robot, ROZZUM Unit 7134 or “Roz”, that is shipwrecked in a storm on a remote wilderness island. Looking for a task to do to justify its existence, it unexpectedly finds itself having to learn how to be a mother. This is not part of its training. However, it is adaptable. It starts the film very robotic but gradually, as it teaches itself more about its new role, it develops much more human-like emotions. After all a first requirement of any parent is to show love. The film is as much about Roz’s journey learning and reprogramming itself to be emotional as the actual story.

So, suppose you are designing a robot to show human emotions, where would you start…with a programmable mouth probably. After all, happiness, sadness and lots more are all there in the mouth. Then you would want eyes that can change shape, eyebrows, and then, to really get it right, something in place of all those muscles in our faces that give us the ability to show micro-emotions.

So what did the film makers give Roz? Two big circular eyes that are cameras and … nothing much else. No face, no nose, no eyebrows, no muscles to move in the face at all. Just some lines of LEDs that can change colour. And yet, it is very clear throughout the film that Roz is becoming gradually more emotional, and exactly what emotions they are feeling. This was something director Chris Sanders was explicitly after. He wanted something much more than a 2-dimensional robot switching from being robotic to being human-like and that is what the film delivers. It has even been praised by critics for its “emotional depths”.

Emotions of course are not the same as facial expressions. We can understand the emotions of others through expression but through lots of other means too, including gestures (like slumping and arm movements) and tone of voice. It is easy to program a robotic face with mouth and eyes etc to show different expressions but that of course does not mean the robot shows emotions, any more than carving a pumpkin to smile means it is happy! However, humans are easily fooled and naturally attach emotions to things that show expressions we recognize. We also use it to help us understand others, so if robots are to live with us it probably needs to fake emotions somehow (assuming it doesn’t actually feel anything). If a robotics company wants us to feel comfortable around robots, and live with them happily, they need to get this right. As the film shows, that doesn’t just mean making robots that are impossible to distinguish from humans (and that approach has risks as unless you make it really, really accurate you hit what is called the “uncanny valley” where what we see are zombie like monsters rather than pleasant companions).

A lot of Roz’s emotions come from the brilliant voice acting of Lupita Nyong’o, who plays Roz. She gradually changes the way she voices Roz through the film. It isn’t a sudden switch but a gradual change you hardly notice. The artists and animators also put an enormous amount of time into developing Roz too. In the story, Roz, gradually reprograms herself based on her ability to mimic others. In terms of animation, the robot uses more and more human-like gestures but also gradually uses those LEDS more and more to show different emotional states, both in the way they flash and in changing colour between red and blue. That was something the animators improvised , using what they had, rather than the plan from the start. This mirrors ideas in Ian M. Banks Culture novels where robot AIs show emotional states through wide use of coloured lights.

We could move towards a world of everyday, everywhere robots that look robotic and emotion-less, or one where they mimic humans with exact copies of the way humans show emotions, Blade Runner style, but much more interesting would be one where, as with Roz, they are clearly something else: where they have their own language to express emotions, that helps us understand them. As the film-makers have shown, wide emotional range can be shown in many subtle but different ways. Time to recruit some artists, film-makers and actors to your robotics start-up if your robots are going to stand out from the crowd?

Paul Curzon, Queen Mary University of London

More on …

Subscribe to be notified whenever we publish a new post to the CS4FN blog.


Emotional glasses

It’s fun to add emoticons to messages, and they help ensure people understand our feelings. They are helping some people understand feelings face-to-face too, with a bit of help from an Artificial Intelligence.

Reading faces

We take it for granted that we can look at someone’s face and tell whether they are happy or sad, angry or surprised. Autistic children, however, often struggle to understand people’s expressions. When anxious we also all tend to avoid eye contact. Some autistic children do that all the time. They are then even less likely to see the clues in people’s faces, and so start to understand emotions. This can make it harder to make friends.

From robots to glasses

Many hi-tech ways have been tried to help autistic children learn about emotions. One, for example, involves letting them play with robot ‘friends’ as some find the cartoon-like expressions on a robot face more comfortable and easier to follow. A different approach is based on wearable technology. Researchers at Stanford University have created a program for autistic children that works out a person’s expression and displays an emoticon of it in a pair of smart glasses.

An AI reading faces for you

A camera in the glasses records what the wearer sees and the Artificial Intelligence (AI) program detects any faces. This kind of technology is also used in smartphones to detect faces in your photo collection. It uses ‘machine learning’: the program learns what a face is by being shown lots of images, some with and some without faces. The program uses all that data to work out the patterns in an image that mean there is a face. It then uses that pattern to spot new faces.

In a similar way it can be trained on faces with different expressions. A training set of faces are used that are labelled with the emotion in that image. This allows the program to spot what pattern in a face makes a happy face, what makes a sad face, and so on. Having recognised an expression, the glasses finally act as a screen and show an emoticon, such as a smiley, corresponding to that expression. Superimposing digital images on the real world like this is called augmented reality. It makes looking at faces like a game and means that the child can use the emoticon to understand what the person in front of them is feeling. It also means they can start to learn for themselves – almost like the AI! The AI is labelling the faces for them, just as people had done for it. With the glasses, autistic children can be sure what each face is actually saying rather than having to guess. Eventually they might then form their own rules and so do it on their own.

Making a difference

The Stanford system was trialled with autistic children in their own homes. They used the system for several months and their parents found it made a clear difference. By the end many of the children were engaging much more with their family including making a lot more eye contact.

Emoticons are making a real difference to their lives.

Paul Curzon, Queen Mary University of London


More on…


Related Magazine …


Subscribe to be notified whenever we publish a new post to the CS4FN blog.


Can you trust a smile?

Yellow smiles image by Alexa from Pixabay

How can you tell if someone looks trustworthy? Could it have anything to do with their facial expression? Some new research suggests that people are less likely to trust someone if their smile looks fake. Of course, that seems like common sense – you’d never think to yourself ‘wow, what a phoney’ and then decide to trust someone anyway. But we’re talking about very subtle clues here. The kind of thing that might only produce a bit of a gut feeling, or you might never be conscious of at all.

To do this experiment, researchers at Cardiff University told volunteers to pick someone to play a trust game with. The scientists told the volunteers to make their choice based on a short video of each person smiling – but they didn’t know the scientists could control certain aspects of each smile, and could make some smiles look more genuine than others.

Continue reading “Can you trust a smile?” →