In a groundbreaking study published in the journal Science Robotics, researchers at Columbia University’s School of Engineering have unveiled Emo, an AI robot that can accurately predict and mimic human facial expressions. Emo, which has 26 actuators in its head-like robot face and is covered with soft silicone skin, was trained using two AI models that analyze subtle changes in facial expressions and generate motor commands for corresponding expressions.
Through self-modeling, Emo was able to learn the relationship between facial expressions and motor commands by observing videos of human facial expressions. By analyzing subtle changes in the opposite face before a person starts to smile, Emo can accurately predict human facial expressions. This breakthrough in human-robot interaction has the potential to improve the quality of interactions and build trust between humans and robots. In the future, interacting with robots may involve them observing and interpreting facial expressions in real-time, similar to how humans interact with each other.
