Greeting Machine Explores Extreme Minimalism in Social Robots

A simple robot that does almost nothing

Greeting Machine, a social robot from IDC Herzliya
Image: IDC Herzliya
Advertisement

Robots in general, and social robots in particular, tend to be very focused on functionality. When we see one, we want to know what it does, how well it does it, and whether it does those things better or worse than other robots or systems. The necessity for robots to be “useful” drives their design, but what happens when you instead design a robot that only has to do one very simple thing? The Greeting Machine is an abstract robot developed at the Media Innovation Lab (miLAB) at the IDC Herzliya, in Israel, with collaborators from Cornell University. It was conceptualized to greet people with a gesture, and do nothing else—a challenge that’s more complicated than it sounds.

Almost every other robot in existence is designed to evoke something that we (the humans interacting with it) will recognize, be it another human, an animal, or an object that we’re familiar with. This is deliberate, because our recognition of the appearance of a robot immediately informs our expectations for that robot’s functionality. With the Greeting Machine, the question is whether it’s possible to instead use a completely abstract robot with no human features to communicate in an intuitive way: Will users be able to recognize shapes and motions that they have no prior context for as a positive greeting?

The Greeting Machine consists of two spheres. The larger sphere contains an arm with a magnet on the end that can trace all the way around the interior shell of the sphere. The smaller sphere also contains a magnet, and sticks to the arm inside the larger sphere, so that moving the arm inside moves the smaller sphere along with it, pulling it across the sphere’s surface. This motion is the extent of the Greeting Machine’s capabilities, so it was important to make the most of it, and the researchers consulted with psychologists, choreographers, puppeteers, and animators to develop the deceptively simple set of approach and avoid gestures. 

It’s remarkable how much study participants varied in their reactions to The Greeting Machine, and the different kinds of emotions and intents that they attributed to the robot. Here are some excerpts from the paper:

Participants often perceived the Greeting Machine’s intent as loaded with judgment about them. When an Approach gesture was triggered, most participants perceived it as signaling that they were acceptable for social interaction (i.e. a positive opening encounter cue): “when he moves he was like ’I just want to interact with you’.” Participants described varying levels of acceptance from the opening encounter: ”[...] he would be like ’oh I’m super excited,’ or he would be like ’I just need to know who she is’[...] when he was moving a little bit more maybe he was excited to see me.” One participant diverged from this trend and perceived the Approach gesture as an aggressive cue signaling that he was not acceptable for social interaction: ”I felt it was just looking at me, staring at me, the same way you pass by the street and there is an aggressive dog at the gate and it looks at you in an unpleasant way.” 

Participants also described their own response to the Greeting Machine’s willingness or unwillingness for social interaction. When they felt that the Greeting Machine found them acceptable for interaction, responses were positive: ”When it stared at me, I was happier than when it wasn’t. When it turned to me, it made me smile a little.” ”When it looked straight at you then you knew the situation was fine, because it was greeting you and everything was okay.” When they felt that the Greeting Machine found them unacceptable for interaction, they felt rejected ”I was a little annoyed: Why is he not looking at me again?” Some of them even became angry: ”when he was facing the wall, it was really not nice. I was standing right there looking at him. I was thinking: Why are you doing that? It’s not nice, not polite.” Participants were also intrigued by the fact that this abstract object could make them feel emotional: ”When I would walk in and it would face away from me, it was like ”I don’t want to talk to you”. It’s weird, because it’s an object and it shouldn’t make me feel anything, but it did. It’s the same as if a person wouldn’t want to talk to you.”

While it’s not necessarily easy to imagine what kinds of practical commercial applications the Greeting Machine might be good for, study participants suggested that it might be nice to have by the entry to your home, welcoming you as you come in the door—especially if you live by yourself. But, the Greeting Machine project is more about figuring out how effective minimalist robots can be at interacting with humans in a useful way.

For more details, we spoke with the researchers from miLAB and Cornell who created the Greeting Machine via email.

IEEE Spectrum: Why did you make the Greeting Machine? What were you hoping it would be able to do, both in terms of research and potential applications?

Hadas Erel, miLAB Faculty: We made the Greeting Machine to explore the possibility that gestures of an abstract robotic object, that has no association with human appearance, can be perceived as a social experience.

Lucy Anderson-Bashan, miLAB Student: We wanted to extend our lab’s previous work on Kip and Vyo, and design a robotic object that was not only non-humanoid but also abstract. Regarding movement, we wanted to test if a minimal movement can be meaningful, inspired by human or animal greeting, which can be minimal yet very meaningful.

Oren Zuckerman, miLAB Director: Greeting is such an important social feature, we hoped that our research will show that social objects like the Greeting Machine may improve well-being, by adding short, positive social-encounters to everyday life.

Iddo Wald, miLAB Faculty: We decided to study if minimal gestures can create meaningful social interactions. We strive to define the "body language" of robotic objects, and create guidelines that will hopefully empower robot designers when working on social gestures. Minimal gestures can lead to low-cost and low-complexity robots. Minimal gestures can also be added to IoT devices and home assistants, adding the "social" aspects to existing smart objects.

It sounds like the specific design of the Greeting Machine is based on a variety of research. Can you describe how you decided on the shapes and motions that comprise the robot?

Benny Megidish, miLAB Student: We wanted to avoid immediate associations to humans or creatures, so we created a variety of low-fidelity shapes inspired by Gestalt psychology (cone, pyramid, sphere). We decided to design it as a sphere due to its symmetry around all axis, and also because curvilinear shapes are considered to be symbols of warmth and happiness. We were also greatly inspired by character animation, and added the small ball according to generate expressive movement. The result was a unique language formed by the relationship between the two spherical forms. From a technical perspective, we leveraged the geometrical properties of the large sphere to create random secondary motion of the small ball. The combination of professionally animated movement and random secondary motion lead to fluent natural movement.

Lucy Anderson-Bashan: In order to understand how to animate the gestures, we met with movement experts (choreographers, puppeteers, and dancers) and asked them to brainstorm on human movement and social interpretation. We created a low-fidelity version of the Greeting Machine for the experts to show us the movement they imagine, using a simple stick that move the small ball along the surface of the sphere. They brainstormed and explored movement for four hours, and at the end showed us the gestures they thought are the most relevant: Approach and Avoid gestures with variance in movement path and end-point visibility.

What do you think is the minimum amount of functionality that a robot can have to still be useful? Are there arguments against adding additional functionality?

Oren Zuckerman: The minimum amount of useful functionality a social robot can have is to react to the presence of a human at the right time and context. The reaction can be very minimal, humans are very sensitive to movement, and very good at interpreting responsive movement as social cue. Adding more functionality to a well-designed human-robot social experience is like adding more functionality to a meaningful picture framed at one’s home. By adding you compromise the existing experience.

What surprised you about how study participants perceived or interacted with the robot?

Hadas Erel: I did not expect the richness of the participants’ social interpretation. An interaction that lasted a few seconds with an abstract robot performing minimal movement led to rich descriptions of opening encounters both negative and positive. People attributed intent and emotions to the robot’s gestures, and the social interpretations were extremely consistent between participants.

Are there specific ways in which you think the Greeting Machine should influence the design of future social robots?

Andrey Grishko, miLAB Faculty: Designers can be much more open and creative when thinking of shapes and movements of social robots, and do not need to restrict themselves to humanoid features or to the repetitive non-humanoid designs we saw at last year’s CES. Social cues can be implemented into any object or device as movement, regardless of the device’s appearance.

Guy Hoffman, Cornell: A lot of what affects our interaction with other agents, including robots, are the cultural priors and expectations that people have going into the interaction. When we encounter robots, we are influenced by science fiction, our experiences with humans and animals, and so forth. The design of a robot always relates to these expectations and plays into them. German designer Dieter Rahms has a criterion for good design he calls “honest.” In his words, honest design “does not make a product more innovative, powerful or valuable than it really is. It does not attempt to manipulate the consumer with promises that cannot be kept.” In that sense, humanoid robots are not “honest” designs. With designs like the Greeting Machine, we are trying to exemplify “honest” robot design, while also arguing that this minimal design can still enable a deep emotional, social, and psychological effect.

“The Greeting Machine: An Abstract Robotic Object for Opening Encounters,” by Lucy Anderson-Bashan, Benny Megidish, Hadas Erel, Iddo Wald, Andrey Grishko, Guy Hoffman, and Oren Zuckerman from miLAB and Cornell, was published at RO-MAN 2018.

[ The Greeting Machine ]

Robotics News

Biweekly newsletter on advances and news in robotics, automation, control systems, interviews with leading roboticists, and more.

About the Automaton blog

IEEE Spectrum’s award-winning robotics blog, featuring news, articles, and videos on robots, humanoids, automation, artificial intelligence, and more.