“We evaluate the resulting model in simulation and with a user study on the real system,” Disney Research says. “We show that our method enables simple autonomous human-robot interactions that are comparable to the expert-operator baseline, and that users can recognize the different robot moods as generated by our model.”
According to the researchers, the AI uses a complex system to predict two things, the first being smooth and continuous movements like waving, and the second being specific actions like saying hello. After conducting a trial run on a computer, they introduced the robot to real people. To their surprise, the robot was an instant success, engaging and interacting with individuals almost as seamlessly as it did under the expert’s control.
“Finally, we demonstrate a zero-shot transfer of our trained model onto a different robotic platform with the same operator interface,” the researchers conclude.