Controlling Robots by Frequency! ... UNIST Develops Sound-Sensitive Artificial Skin Technology

[Asia Economy Yeongnam Reporting Headquarters Reporter Hwang Dooyul] An interface technology that controls robots using artificial skin capable of hearing sounds has been developed.

The robot can distinguish the texture of materials through artificial skin, recognize sounds, and execute commands. It can also mimic human movements exactly.

This technology is perfectly suited for the metaverse and avatar robot era and can be applied in various fields.

The research team led by Professors Ko Hyun-hyeop and Kim Jae-jun at UNIST developed a human-machine interface that recognizes and transmits human movements, textures, and sounds to machines.

The team stated, “Instead of pressing buttons or keyboards, this is a human-machine interface that intuitively transmits information to machines,” adding, “The sensor is thin and attachable, making it usable in various virtual reality (VR), augmented reality (AR), and Internet of Things (IoT) technologies.”

The interface is based on an artificial skin sensor that mimics the cochlear structure of the ear.

It applies the principle of the cochlear basilar membrane, which varies in thickness, width, and stiffness by region, allowing it to distinguish sounds by frequency.

Thanks to these characteristics of the sensor, it can transmit to machines not only low-frequency signals that repeat slowly like human movements but also high-frequency signals such as rapidly vibrating sounds and textures with a low signal-to-noise ratio.

Characteristics and Application Areas of Control Technology Based on Sound-Detecting Artificial Skin Sensors.

The research team demonstrated application technologies such as avatar robot hand control using the sensor and smart haptic gloves.

In a demonstration controlling an avatar robot hand by sound, changing the frequency allowed control of the robot hand’s movements.

When a user wore the smart haptic gloves and moved, the avatar robot hand mimicked the user’s hand movements exactly and recognized the textures of eight different materials including glass, paper, and silk with 93% accuracy.

The developed sensor consists of multiple unit triboelectric sensors with varying thickness, porosity, and area, arranged continuously like the cochlear basilar membrane.

The sensor’s internal structure was specially designed, improving pressure sensitivity up to eight times compared to conventional flat sensors.

The recognition frequency bandwidth ranges from 45 to 9000 Hz, enabling detection of various biological signals such as electrocardiogram signals (0.5?300 Hz), electromyogram signals (50?3000 Hz), heart sounds (20?20000 Hz), and voice (100?400 Hz).

Even in noisy environments, machine learning allows recognition of human voices with 95% accuracy, enabling use as a microphone equipped with noise cancellation features.

The research was published on March 25 local time in ‘Science Advances,’ a sister journal of the world-renowned journal Science, published by the American Association for the Advancement of Science (AAAS).

The study was supported by the Ministry of Science and ICT’s Mid-career Researcher Support Program, EMTEK, and POSCO Science Fellowship.

This content was produced with the assistance of AI translation services.