Caring, empathetic robots the goal of OU professor's research
Robots may one day learn to care for and nurture one another, according to research by an OU professor.
Computer science professor Dean Hougen studies robotics in the OU Robotic Intelligence and Machine Learning Lab, founded by him in 2001.
Hougen’s most recent project investigates whether robots can learn to care for one another and, eventually, humans.
“I am hoping one of the things we will evolve is the capacity for empathy, to look at another individual and say, ‘I see what this person needs. I am going to respond to it,’” Hougen said. “There are people who try to write code to make a robot empathetic, but it’s hard to sit down and say, ‘OK, this is what it means.’”
Hougen said he realized most organisms are born with instincts that tell them how to survive, but if an organism is in a rapidly changing environment, these skills may not be applicable, and it will have to learn new skills.
From there, the idea of having a nurturer seemed most logical.
“What if you had someone taking care of you, protecting you from predators, finding food for you, keeping you warm, all these things you don’t know how to do for yourself?” Hougen asked. “Maybe if you have nurturing first and you’re in a changing environment, then you would evolve learning mechanisms as opposed to going extinct or someone with instincts not appropriate.”
Birth of an idea
Hougen said he hashed out a lot of these ideas through conversations with one of his graduate students, Mark Woehrer.
The two found this was not an unusual idea, and researchers in the OU zoology department supported the concept that nurturing leads to learning.
To instruct the robots to learn, the lab designed a simple experiment with a computer-simulated robot in the middle of a circle with a light switch and a light bulb, Hougen said.
The goal was to manipulate the robot to turn on the switch, then sit under the bulb to charge itself.
“Over time, they got better and better, and then they would turn and head right for the switch and then head straight back to the light,” Hougen said. “Then we said, ‘OK, what if we wanted a parent nurturing a child.’”
Parent and child
Hougen and a French exchange student working in the lab this summer tweaked the experiment to have two robots in the same circle, one acting as a parent and one as a child, Hougen said.
The robots had similar algorithms that made them related and had the ability to turn on the switch. However, only the child would benefit from going under the light this time, he said.
“Over time, the parent learned to go to the light switch while the child went to the light,” Hougen said. “Now the parent is taking care of the child and doing what it can within its environment to provide for its offspring.”
These results show robots, as parents, were indeed learning to be nurturers and children were also learning to be nurtured, Hougen said.
Computer science senior Bryan Hoke will take over work on the project to continue expanding the ability of robots to nurture one another.
Hoke is working on the project through the OU Honors College’s Honors Research Assistant Program.
“The prospect of working on problems that have not yet been solved is very exciting to me because I will be able to fully and freely use my intellect in the attempt to develop solutions to these new problems,” Hoke said.
Even with robots, the research could take nearly 20 years, but Hoke said he is still excited about contributing what he can during his 10-week work with the project.
Hoke said he has always been interested in Darwin and his theories of evolution and wants to apply these to robots.
“Even now — after having three years of college education in computer engineering and computer science, physics and math — I find Darwin’s theory of evolution by natural selection to be one of the greatest, most robust theories of all time,” Hoke said.
A lot of Hougen’s personal interest in this project came from his work as a graduate student with robots, he said.
Hougen also worked with teaching robots to perform simple tasks but realized even simple robots required maintenance and a human to take care of it, such as untangling wires.
If a robot could take care of another robot, it would reduce the amount of required human attention.
The experiment could eventually expand to include other social sciences, such as psychology and anthropology, to see what robots are capable of learning, Hougen said.
Hougen said he sees a strong potential for having robots learn to teach one another.