Why Robots Should Be More Like Babies

Why Robots Should Be More Like Babies

Many people do not realize that an infants most important tool for learning is through imitation of other human actions. By observing how the people around them behave, infants rapidly gain skills needed to progress through development stages.

However, little humans do not simply parrot what others are doing and saying. Around about 18 months of age, they begin to understand the intent behind actions and behaviors. Being able to connect intention with observable action is important. Critical thinking starts to come into play when they can begin to come up with other ways to complete a task or goal.

“Humans are the most imitative creature on the planet and young kids seamlessly intertwine imitation and innovation,” states Andrew Meltzoff, psychology professor at University of Washington and co-director of the Institute for Learning & Brain Sciences at the university. “They pick up essential skills, mannerisms, customs, and ways of being from watching others, and then combine these building blocks in novel ways to invent new solutions.”

Is it possible for robots to learn in a similar manner? To explore this question, Meltzoff teamed up with roboticists and machine-learning experts. Their research and findings were reported in the journal PLOS ONE last month.

“The secret sauce of babies is that they are born immature with a great gift to learn flexibly from observation and imitation. They see another person and register that the person is ‘Like Me.’ They devote great attention to the ‘Like Me’ entities in the world,” Meltzoff explained. “Roboticists have a lot to learn from babies.”

Using precise algorithms, the team programmed the robots to calculate how different actions might result in different outcomes. The robots relied on a probabilistic model to predict what the researcher wanted it to accomplish. The researchers also programmed the robots to ask for help when they were not sure what the outcome was supposed to be.

Two experiments were carried out by the team to test the program. For one experiment, a robot would learn to follow a human’s gaze. The second experiment consisted of having the robot imitate someone arranging fake food around a tabletop.

During the first experiment, the robot would learn the movements of its own head, and conclude that the human’s head worked in the same way. The robot would observe the motion of the human’s head, such as which direction that person was looking and where their gaze was falling, and imitate the movements. During the second experiment, the robot practiced moving food-shaped objects around on a table. Interestingly, the robot took things further by not only repeating what the human did with the toys, such as sweeping them off the table top, but occasionally using means different from the human to achieve the same goal.

This kind of robot adaptation was a huge finding. The robots we use today in assembly lines and for other repetitive tasks do a great job at imitating and repeating a human task. “They are not so good at inferring the intention behind a human action and achieving the same goal using different means,” said Rajesh Rao, the director of the Center for Sensorimotor Neural Engineering, and one of the lead researchers in these experiments.

However, robots are still one up on humans when it comes to the ability to access and distill huge troves of data and information, as well as sharing information with other machines quickly. “Eventually, robots might be able to learn complicated tasks more quickly than babies if they are provided with more powerful sensors, more versatile actuators, and sufficient computational power to implement human-inspired learning strategies,” Rao claimed.

By modeling human development, Rao and his colleagues predict that robots will be able to learn increasingly more sophisticated skills just by observing other humans and robots.

“We are convinced that bringing together the roboticists and developmental psychologists may allow us to combine the best of human learning and the best of machine learning to the benefit of both,” Meltzoff said. “I’m trying to teach the roboticists to think like a baby. And I mean that in a good way.”