FOREST: A Project Building Human-Robot Relationships Through Music and Dance
There is no reliable way to get a person to trust a robot. Part of the problem is that robots generally just do what they were programmed to do, and people usually don't feel that a robot has any interest in forming some kind of non-functional connection.
From the robot's point of view, people are fragile, mobile beings that must not be touched and that help with tasks when needed, and nothing more.
The impact of robots on human life is evident everywhere, from service sectors such as healthcare and retail to industrial enterprises such as car manufacturing. But using robots as helpers is one thing, and interacting with them emotionally, which can lead to greater trust, is another.
People begin to trust other people by forming an emotional bond with them, which is known to be hard for robots. An emotional bond obviously does not have to mean love or anything similar, but it does mean a certain level of mutual understanding, communication and predictability. It is the feeling that the other person sees you not simply as an object (and vice versa).
For robots, which are objects, this is a real problem. That is why, with financial support from the National Science Foundation, the Georgia Tech Center for Music Technology teamed up with the dance faculty of Kennesaw State University on the FOREST project, which consists of robot musicians and robot dancers that interact with people to explore creative collaboration and build trusting relationships between humans and robots.
“It was interesting to watch choreographers and dancers, who are not surrounded by technology every day, learn to trust and gradually build relationships with our robot performers. When you combine such interesting, diverse work and different disciplines, there is hope that it will inspire others to new achievements in dance, music and robotics”
According to the researchers, the FOREST dancing robots and the accompanying musical robots are not rigid imitators of human melodies and movements. Rather, they demonstrate a remarkable level of emotional expression and fluidity of human gestures, which the researchers call "emotional prosody and gesture" to project emotions and strengthen trust.
If you are curious what "prosody" means, you can read Wikipedia for a long time, but the term broadly refers to aspects of speech that are not determined by the actual words. For example, you can say "robots are smart" and give it many meanings depending on whether you say it ironically, sarcastically, as a question, or sadly. That is prosody. You can imagine how this concept might extend to movements and gestures, but this would require considering how robots interact effectively with people.
Many of the robots in this show are already well known, including Shimon, one of Gil Weinberg's most creative performers. The video provides some additional background on how the performance came together.
“The main thing we really wanted to focus on is capturing natural emotions. One of my main goals was for the robot to look as if it moves like a human”
Rogel developed a new platform for controlling robot movements driven by emotions, based on human gestures. Wearing a motion-capture suit and recording body movements, he mapped each movement to a different degree of freedom. He and other researchers soon learned that when one part of the body moves, other parts make small adjustments.
Another part of the FOREST project was getting robots to express emotions through sound. Weinberg explained that the team developed a new machine learning architecture to generate sound based on a recently created dataset of emotionally labeled phrases. These emotional-musical phrases (EMP) were built into the robots to increase trust with people. The researchers then evaluated the robots' nonverbal emotional responses using three robotic platforms.
FOREST choreographer Ivan Pulinkala noted how quickly the dancers socialized and got used to the robots.
“We worked through a number of functional modalities with dancers physically manipulating robotic arms. The dancers became more social and developed an interpersonal connection with the robotic arms”
Pulinkala considers FOREST a vivid example of "disruptive collaboration", a term he coined for new artistic techniques that challenge the traditional choreographic process. Conventional choreography creates a movement, and then co-authors develop that movement in different ways.
What some find a bit strange about all this is the idea of trust, since in a sense robots seem like they should be entirely trustworthy, because they can (in an ideal world) be fully predictable.
For example, if a robot is programmed to perform actions X, Y and Z in that sequence, you don't need to believe that the robot will do Y after X the way you would have to trust a human to perform those actions. Because, strictly speaking, the robot has no choice.
However, as robots become more complex and people increasingly expect them to interact socially, the gap between what is technically predictable (or perhaps predictable only after the fact) and what is predictable to the end user can become very, very wide, so a more abstract kind of trust becomes increasingly important. Music and dance may not be the way to do this for every robot, but it has certainly been a start.
Ultimately, the researchers at Georgia Tech and Kennesaw State hope that improving human-robot interaction can inspire new forms of artistic expression while also increasing trust between people and robotic platforms.
