A new MIT study demonstrates how robots can interact socially with one another and understand the differences between those interactions.
Eventually, MIT researchers hope that the model will be applicable to both robot and human interactions. Researchers believe that quantifying social interactions will benefit not only robotics, but also the automotive industry, healthcare, and other industries.
When we think of robots, we think of cold machines with little understanding of human nature, but that may be changing soon.
A new study published by a group of Massachusetts Institute of Technology researchers looks at how robots can become more social and how we define social interactions in general.
The study’s findings will pave the way for a future in which robots are more helpful and understand humans, which will be critical as robots become more prevalent in our daily lives.
“Robots will increasingly become a part of our lives, and even though they are robots, they must understand our language,” said Boris Katz, principal research scientist and head of the InfoLab Group in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), and a member of the Center for Brains, Minds, and Machines (CBMM).
“More importantly, they will need to understand how humans interact with one another.”
What the Study Discovered The study, titled “Social Interactions as Recursive MDPS,” grew out of the authors’ interest in quantifying social interactions.
According to Andrei Barbu, a research scientist at CSAIL and CBMM and co-author of the study, almost no datasets and models in computer science look at social interactions.
“The categories for social interactions are unknown; the degree to which a social interaction occurs or does not occur is unknown,” he said during a video call. “And so we really thought this is the type of problem that might be amenable to more modern machine learning.”
READ ALSO: Are Robots Coming For Models’ Jobs in Fashion Shows?
The researchers created three distinct types of robots with distinct physical and social objectives and had them interact with one another. Barbu stated that a level zero robot had only a physical goal in mind, whereas a level one robot had physical and social goals to assist other robots but assumed that all other robots only had physical goals. Finally, a level two robot assumed that all robots had both social and physical objectives.
The model was validated by placing robots in a simple environment and allowing them to interact with one another based on their levels. Human test subjects were then shown video clips of these robot interactions in order to determine their physical and social goals.
The findings revealed that, in most cases, the study’s model agreed with humans on whether or not social interactions occurred in different clips. This means that technology for detecting social interactions is improving and could be applied to robots and a variety of other applications.
A Technological Future That’s More Social.
Barbu stated that this research would be expanded to test not only robot-to-robot social interactions, but also how robots can interact with humans on a social level, which is desperately needed in robotics.
“One aspect of the future is robots that understand us better,” he said. “For the time being, most robots are not particularly friendly. In many cases, they are not particularly safe to be around because they can easily do something dangerous or unpredictable to us. So having a robot that can actually assist you in doing something is critical.”
Consider having a conversation with Alexa or Siri and having these assistants accurately assist you rather than constantly misunderstanding you. The authors of the study also published a follow-up research paper in which they expanded the framework for richer social interactions between robots such as cooperation, conflict, coercion, competition, and exchange.
While a world in which robots can better understand us will be beneficial, Barbu believes that social skills for machines will be useful in a variety of situations.
READ ALSO: “I Think I Need A Robot, As I No Longer Trust People” – Tonto Dikeh Says
“For example, we collaborate with the Toyota Research Institute, and autonomous cars actually require a certain level of social skills when they reach an intersection.”
Barbu elaborated. “In that scenario, it’s often not just about who has the [right-of-about way]—it’s the social interaction between the two cars.”
However, Barbu stated that the ability to quantify social interactions with this model would allow for the monitoring of social interactions for diseases and disorders such as autism, depression, Alzheimer’s, and others.
“This type of thing is really important in cognitive science because social interactions are understudied—they’re kind of a big black box,” he explained. “Being able to quantify them also makes a significant difference.”