New Algorithm Enables Humanoid Robots to Mimic Human Facial Expressions

Due to the sharp increase in the need for smart machines in everything from homes and medical procedures to warehouses, robot development is currently one of the fastest-expanding industries.

Businesses worldwide are vying to create the most appropriate robot that can mimic human characteristics.

Chinese scientists have now created robots with realistic facial expressions that resemble people.

The research team led by Liu Xiaofeng, a professor at Hohai University in Jiangsu Province, east China, is creating a humanoid robot with highly expressive facial features.

Also Read: Top 7 Best Robotics Courses & Certificates Online [2024]

The research team created a new algorithm for humanoid robot face expression generation to produce this robot.

Liu asserted that the lack of complex and genuine facial emotions typical of humans in humanoid robots causes problems for seamless user interaction.

To overcome this obstacle, Liu and his colleagues presented a thorough two-phase approach that will enable our self-governing emotional robot to display a wide range of realistic facial expressions.

According to Liu, their approach initially creates complex robot facial expression images that are led by artificial units (AUs). In the second stage, they create an emotive robot that has several degrees of flexibility for facial expressions, which allows it to represent artificially generated fine-grained facial expressions, according to Xinhua.

The paper, which was published in the IEEE Transactions on Robotics journal, describes a novel approach to facial expression disentangled synthesis driven by Action Units (AUs). This method makes it possible to generate robot facial expression images that are subtle and guided by Action Units.

Also Read: Deepak Pathak Net Worth: Co-Founder & CEO at Skild AI Robotics Company

Through the use of facial AUs in a poorly supervised learning framework, the researchers can overcome the lack of paired training data (pictures of source and target face expressions).

Researchers in the paper stated, “We use a latent facial attribute space to separate expression-related and expression-unrelated cues, using only the former for expression synthesis, to maintain the integrity of AUs while reducing identity interference.”

“In the second stage, we implement an emotional robot equipped with multiple facial movement degrees of freedom, enabling the expression of the artificially generated fine-grained facial expressions.”

The researchers developed a specific motor command mapping network to act as a bridge between the robot’s realistic facial reactions and the generated expression images.

Also Read: Discover the AI-Powered Robots Big Tech Believes Will Solve the Global Labor Shortage

By using the physical motor positions as limitations, Chinese researchers improved the prediction of exact motor orders from the robot’s generated facial expressions.

The study claims that this refinement procedure guarantees that the robot’s facial motions accurately and naturally portray expressions.

Lastly, the efficacy of the suggested generation approach is confirmed by qualitative and quantitative assessments of the benchmark emotional dataset.

The researchers stated that their approach “achieves a promising generation of specific facial expressions with given AUs, significantly enhancing the affective human-robot interaction” based on the affective robot they had constructed themselves.

Also Read: West Japan Railway’s New Humanoid Robot Revolutionizes Network Maintenance