The automated processing of textiles and thus robot-controlled production are among the great visions of Industry 4.0 in the clothing industry. So far, however, the idea of being able to grip and even process such flexible, light and different materials as textiles has failed because of the robots’ abilities.
Researchers at the Robotics Institute of the US Carnegie Mellon University in Pittsburgh have now come one step closer to the goal and have developed a robot that can use machine learning and the latest sensors to grip individual layers of fabric. The robot uses a tactile sensor called “ReSkin” and a simple machine learning algorithm known as a classifier. Most attempts to teach robots how to handle fabrics are based on the use of cameras that only collect visual data. However, visual data provide too little information about the type of substance and how to handle it. Another advantage of the tactile sensor, which is integrated in the gripper tongs, is that disruptive factors such as light and patterns can be eliminated.
Carnegie Mellon University and the social media group Meta AI have jointly researched ReSkin, with use in the textile industry being just one of many possible areas of application. Capable of collecting touch data, ReSkin can improve the sense of touch in overall robotics, wearables, smart clothing and artificial intelligence.