News — After years of existing only in fiction, social robots are finally being designed that can more closely emulate how people express themselves, interact and learn 鈥� and doing so while performing jobs like teaching social behavior to children with autism or helping stroke patients with their physical rehabilitation exercises.
Recently, The Kavli Foundation brought together three pioneers in Human-Robot Interactions to discuss these advancements, as well as the upcoming technological hurdles. What they say is that, while there are many challenges ahead, the biggest remains getting the robots to match the needs and expectations of the human mind. 鈥淗ow we interact with embodied machines is different than how we interact with a computer, cell phone or other intelligent devices,鈥� says Professor Maja Matari膰, University of Southern California. 鈥淲e need to understand those differences so we can leverage what is important.鈥�
A director of USC鈥檚 Center for Robotics and Embedded Systems, Matari膰 has developed social robots for use in a variety of therapeutic roles. According to Matari膰, one of the keys for a successfully designed social robot is considering not only how it communications verbally, but physically through facial expressions and body language. Also important: embedding the right personality. 鈥淲e found that when we matched the personality of the robot to that of the user, people performed their rehab exercises longer and reported enjoying them more.鈥�
Another key is matching a robot鈥檚 appearance to our perception of its abilities. Ayse Saygin is an assistant professor at the University of California San Diego and faculty member of the Kavli Institute of Brain and Mind. Last year, Saygin and her colleagues set out to discover if what they call the 鈥渁ction perception system鈥� in the human brain is tuned more to human appearance or human motion. By using brain scans, they found that as people observed highly humanlike robots compared to less humanlike robots, the brain detected the mismatch and didn鈥檛 respond as well. 鈥淢aking robots more humanlike might seem intuitively like that鈥檚 the way to go, but we find it doesn鈥檛 work unless the humanlike appearance is equally matched with humanlike actions.鈥�
A social robot also needs the ability to learn socially. Andrea Thomaz is an assistant professor at the Georgia Institute of Technology and director of its Social Intelligent Machines Laboratory. At her lab, they have built a robot designed to learn from humans the way a person would -- along with speech, through observation, demonstration and social interaction. "In my lab, we see human social intelligence as being comprised of four key components 鈥� the ability to learn from other people, the ability to collaborate with other people, the ability to apply emotional intelligence, and the ability to perceive and respond to another person鈥檚 intentions. We try to build this social intelligence in our robots."
Read the complete roundtable discussion at:
RSS