Insights
- NC State researchers merge 3D embroidery and machine learning to craft a fabric-based touch sensor.
- The triboelectric device, powered by friction, integrates into clothing, enabling touch control of electronic devices.
- Machine learning aids gesture recognition, enhancing accuracy.
- While in early stages, the sensor showcases potential in wearable electronics.
A new project from NC State University combines three-dimensional embroidery techniques with machine learning to develop a fabric-based sensor capable of controlling electronic devices via touch.
As wearable electronics gains popularity and additional features are added to garments, an embroidery-based sensor or button capable of controlling those capabilities will become increasingly crucial. The sensor, which is integrated into the fabric of an item of clothing, may activate and control electrical devices such as smartphone apps entirely via touch.
The device consists of two parts: the embroidered pressure sensor and a microcontroller that processes and delivers the data acquired by the sensor. The sensor is triboelectric, which means it runs on the electric charge generated by friction between its several layers. According to the study, it is manufactured from yarns containing two triboelectric elements, one with a positive electric charge and the other with a negative charge, which were incorporated into traditional textile fabrics using embroidery machines.
According to Rong Yin, the study’s corresponding author, the sensor’s three-dimensional structure was critical to getting correct. “Because the pressure sensor is triboelectric, it required two layers with a space in between. Because we were utilizing embroidery, which is typically two-dimensional, this gap was one of the most difficult portions of the procedure. It is a cloth decoration method. That method makes it difficult to create a three-dimensional structure. We were able to regulate the spacing between the two layers by adding a spacer, which allowed us to manipulate the sensor’s output.”
Data from the pressure sensor is subsequently transmitted to the microprocessor, which is in charge of converting the raw input into particular instructions for any linked devices. According to Yin, machine learning techniques are critical to ensuring that this goes properly. The device must also be able to distinguish between movements allocated to different purposes and ignore any unintended inputs that may arise from the cloth’s regular movement.
“Sometimes the data that the sensor acquires is not very accurate, and this can happen for all kinds of reasons,” Yin explained. “
Environmental factors such as temperature or humidity can occasionally alter the data, or the sensor may accidentally touch something. We can educate the device to detect such objects using machine learning. Machine learning enables this little device to do diverse tasks by recognizing various input types.
The researchers showed this input detection by creating a simple music-playing mobile app that linked to the sensor via Bluetooth. The app’s six functionalities are play/pause, next song, last song, volume up, volume down, and mute, each controlled by a distinct sensor motion. The device might also be used to set and enter passwords, as well as control video games, according to the researchers.