Zurich ETH Invents Low-Cost Tactile Sensor

Using machine learning, researchers at the Federal Institute of Technology (ETH) in Zurich have developed a novel low-cost tactile sensor. The assembly can measure the force distribution with high resolution and high precision and allows the robot arm to grasp fragile objects. Humans can easily grab fragile or slippery objects. Tactile sensation allows us to judge whether we are firmly grasping the object, or whether the object will slide out of the finger, thereby adjusting the grasping force accordingly. For robotic arms that need to grab objects that are fragile, slippery, or complex, they also need such feedback.

Robot researchers at the Federal Institute of Technology in Zurich have developed a tactile sensor in response to the situation and believe it is an important step towards "robot skin". The engineers pointed out that the design of this sensor is very simple, so the manufacturing cost is low. Essentially, it consists of an elastic silicone “skin” with colored plastic beads and a normal camera attached to the underside.

The measurement method using the pure optical input

The sensor is based on visual operation: when it comes into contact with the object, an indentation will appear on the silicone skin and change the pattern of the microbead, and the fisheye lens on the underside of the sensor will record. According to the change of the pattern, the system can calculate the force distribution of the sensor.

Carlo Sferrazza, a Ph.D. student in dynamic systems and control at the Federal Institute of Technology in Zurich, said: "Traditional sensors only register the applied force at one point. In contrast, our robot skin allows us to distinguish several forces acting on the sensor surface and calculate them with high resolution and accuracy. We can even determine the direction of a force. "In other words, researchers can not only determine the vertical pressure applied to the sensor but also determine the lateral shear force.