A team of researchers from Osaka University has developed a new system for replicating facial expressions more realistic in humanoid robots. The study, published in the Journal of Robotics and Mechatronics, proposes an innovative approach based on “waveforms” to generate more fluid and natural facial movements. The video you can find it here.
The method, called “dynamic arousal expression”, classifies various facial gestures such as yawning, blinking and breathing as individual waveforms. These are then linked to the amplitude of movements such as opening and closing the lips, moving the eyebrows or tilting the head. A control parameter based on a spectrum of moods, from “sleepy” to “excited,” determines how these waves propagate and overlap to adjust the physical characteristics of the robotic face.
According to the researchers, this approach eliminates the need to program individual facial movements for every emotional state, as happens with current methods. “The automatic generation of dynamic facial expressions to convey a robot’s internal states, such as mood, is crucial for communicative robots,” say the study authors.
For example, a “sleepy” state generates certain results in the robot’s breathing, yawning, and blinking parameters. These combine with each otherfurther amplifying or reducing facial movements such as mouth size, eye blinking, and head tilt. Once calculated, the resulting physical movements are reproduced almost instantly.
Professor Koichi Osuka, senior author of the study, said: “Advancing this research into synthesizing dynamic facial expressions will allow robots capable of complex facial movements to exhibit more vivid expressions and convey mood changes in response to surrounding circumstances, including interactions with humans.”.
Although it represents a step forward towards more realistic robots, the system still shows some limitations. Facial movements appear smoother and more natural than many other contemporary machines, but the robot’s eyes remain distinctly artificial. Furthermore, the fact that the eyes often seem to move rapidly from side to side and lose focus does not help reduce the sensation of uncanny valley.
The study, published in the Journal of Robotics and Mechatronics, proposes an innovative approach based on “waveforms”
Despite this, treating facial features as an interaction between waves of different intensities seems to offer more realistic results than considering them as pre-programmed one-to-one reactions. The next step could be to test the system on an adult robot to further reduce the feeling of strangeness.
This research opens up new possibilities for developing humanoid robots with more natural and compelling facial expressions, potentially improving their ability to interact more effectively with humans in various contexts, from assistive robotics to human-machine interaction research .
Source: www.tomshw.it
