The eyes have it: how robot vision learnt from Walt Disney

When people struggle to relate to robots without a humanoid form, the answer is to use the great animator’s trick of sticking on a pair of googly eyes

How to solve one of our trickiest UX problems? Put googly eyes on it. While we’ve all seen faces on highly sophisticated state-of-the-art humanoid robots, they could be the answer for a much wider range of semi-autonomous robot helpers.

A team at Futurice designed and built a social robot to help guide visitors at Helsinki’s new central library, Oodi. With 10,000 visitors a day, the librarians were spending much of their time helping people to locate shelves and return books – both tasks which could be taken up by a robot helper, freeing up their time to issue more specialised advice.

During tests, they found that people were either not using or misusing the robot, struggling to relate to the robot as a social object. The robot’s form – essentially a box on wheels – was too abstract for people to make sense of how they were supposed to interact with it.

Exaggeration works

The researchers used one of Walt Disney’s 12 principles of animation – Exaggeration – to give the robot large googly eyes. This was accompanied by programming to change the way it appeared to behave using sound, eye movements and lights. For example, if it was bored, it would try to attract attention by looking around, dancing and flashing lights. This transformed people’s interactions with the robot; they started to follow it even when it was guiding someone else, and several even wanted to pet it.

This is not the first example of googly eyes being used to shape our interactions with autonomous machines. At the beginning of 2019, the US supermarket chain Giant Food Stores announced that it would introduce customer-assisting robots in East Coast locations. Similar to the Oodi robot, these robots – collectively dubbed Marty – are fitted out with an outsize pair of googly eyes. While this is likely to raise a chuckle in the aisles, it’s based on academic research showing that adding eyes to inanimate objects puts people at ease, in addition to encouraging them to be more generous and pro-social. It’s called the ‘watching-eye paradigm’, exploiting the deep-seated human trait of needing to be valued within our social circles.

Minimal suggestions

In the course of human evolution, our ability to cooperate with others has been critical to our survival – this proved so important that we’ve become sensitive to even minimal suggestions that something might be looking at us and this can influence our behaviours.

A study at the University of Virginia found that donation boxes with a photograph of human eyes above it generated a statistically significant increase in donations compared to photographs of a chair or mouth. In a previous research study, Dr Anisha Vaish found that putting a picture of a pair of eyes near the communal milk supply significantly reduced the rate at which people would help themselves.

As it turns out, people see faces – and assign agency as a result – in almost anything. When something appears to move on its own, people tend to process it as alive. It turns out that adding googly eyes to a roving robot can elicit the same social responses from people even when we know that it is not alive.

Limits to the effect

Before anyone starts plastering their office in eyes, there are limits to this effect. Milk consumption went back up once people had become accustomed to the watching gaze. In addition, the Uncanny Valley exists for robot eyes as well: this is the phenomenon whereby a computer-generated figure or humanoid robot that looks nearly-identical to a human being generates a sense of unease or revulsion.

A recent study from Princeton University  showed a mixture of photographs and stylised, abstract depictions of macaque faces to real macaques. The monkeys looked at the stylised and photographic images, but they avoided images that were very close to being real, but not quite there; this is interpreted by the brain as negative signal. This potential lack of trust in a realistic-looking robot eye might mean that, while the underlying technology becomes more sophisticated, the googly eye is here to stay.

You can read Imogen’s Privett’s full design round-up in WORKTECH Academy’s Innovation Zone here. Access to the Innovation Zone is for Global Partners and Corporate Members only. Joining details here