Building robots with human-like faces and the ability to mimic human expressions has long been a fascination in the field of robotics. However, most of the best robots, ones that can walk, run, climb steps, and perform complex movements, do not have faces. There may be a good reason for this. If these advanced robots were to have faces, it is likely that humans would be so captivated by their expressions that they would be unable to react quickly enough to the robot’s movements.
Columbia Engineering has recently developed a new research robot named Emo, which focuses on “Human-robot Facial Co-Expression.” This development is impressive and important work towards creating more expressive robots. In a recent scientific paper and accompanying YouTube video, researchers describe their work on Emo’s ability to make eye contact, imitate, and replicate human expressions in real-time.
Upon observing the human-like expressions demonstrated by Emo, it is evident that they are rather eerie. While its head shape, eyes, and silicon skin mimic a human face, they do not quite achieve a realistic appearance, falling into what is known as the uncanny valley. This refers to the discomfort or unease humans feel when robots or other artificial beings resemble humans but fall short of appearing convincingly alive.
However, the purpose of Emo is not to create a robot with a realistic appearance for everyday use. Rather, Emo serves as a platform for programming, testing, and learning. It is a stepping stone towards eventually having expressive robots in homes and daily life.
Emo is equipped with two high-resolution cameras in its eyes, enabling it to make “eye contact” with humans. Through sophisticated algorithms, it can analyze human facial expressions and predict the emotions being displayed. Drawing on the concept of human modeling, in which individuals unconsciously imitate the movements and expressions of those they interact with, Emo uses this insight to mimic the predicted facial expressions.
By observing subtle changes in a human face, Emo can predict an approaching smile 839 milliseconds before the human actually smiles and adjust its own facial expression to smile simultaneously. In the video demonstration, Emo’s expressions change rapidly, closely reflecting the researcher’s facial expressions. While its smile may not appear fully natural and its sadness or surprise expressions may be unsettling, the 26 under-the-skin actuators of Emo bring it closer to delivering recognizable human expressions.
The researchers behind Emo’s development believe that predicting human facial expressions is a significant advancement in the field of human-robot interaction. Traditionally, robots have not been designed to consider and respond to human emotions and expressions. However, with Emo’s capabilities, it opens up new possibilities for more intuitive and seamless interactions between humans and robots.
Intriguingly, Emo’s learning process for understanding human expressions is fascinating. Researchers placed Emo in front of a camera and allowed it to make various facial expressions. By correlating its motor movements with the resulting facial expressions, Emo learned how its own face and motors work together. Additionally, the AI within Emo was trained on real human expressions, combining these methods to achieve nearly instantaneous human-like expression capabilities.
Looking ahead, the researchers envision Emo serving as a front end for an AI or Artificial General Intelligence. This concept refers to a thinking AI that can understand and interact with humans in a more comprehensive and natural way. Emo’s development aligns with the broader goal of bringing robots and AI closer to human-like capabilities.
Emo’s debut coincides with the recent unveiling of Figure AI’s Figure 01 robot, which incorporates OpenAI technology to understand and respond to human conversation. Interestingly, Figure 01 does not possess a human-like face. Imagining a hybrid robot that combines Emo’s expressive face with Figure 01’s conversational abilities sparks curiosity and anticipation for how such a creation could revolutionize human-robot interactions.
In conclusion, the development of Emo represents a significant advancement in the field of robotics, specifically in human-robot interaction. While there may be challenges to overcome in achieving a fully realistic and natural appearance, Emo showcases the potential for robots to emulate human expressions. By studying human modeling and employing advanced algorithms, Emo can predict and mirror facial expressions in real-time. As technology progresses, the integration of expressive faces in robots may become more commonplace, leading to richer and more seamless interactions between humans and machines. The continued research and development of emotion-expression robots like Emo pave the way for a future where robots are not only capable of performing complex physical tasks but also engaging with humans in a deeply human-like manner.
Source link
