Researchers from ETH Zurich in Switzerland have created a low-cost tactile sensor capable of measuring force distribution at high resolution to enable robots to clutch fragile objects, bringing robotic skin one step closer to reality.

Mimicking a human’s ability to firmly grip such objects without dropping them, the tactile sensor behaves like a so-called robotic skin, composed of elastic silicone, colorful microbeads and a camera built into the sensor’s underside.

When the vision-based sensor makes contact with an object, an indentation forms in the sensor’s skin. According to researchers, differences in those indentation patterns could be used to determine the sensor’s force distribution. Unlike traditional sensors, which only register applied force at just one point along a sensor, the robotic skin sensor can distinguish among many different forces applied to the sensor’s surface along with the direction of the force.

Using machine learning, the ETH Zurich team experimented with the sensor, controlling and varying the contact location, objects used and the force distribution to create thousands of different contact scenarios, matching them with alterations in the microbead patterns.

As the ETH Zurich team works on developing thinner sensors — ideally, 0.5 cm thick — they envision that the sensors will pave the way for robotic arm grippers capable of clutching fragile objects and readjusting their grip when the robotic arm senses an object is slipping out of its grasp.

Additionally, the team also envisions the sensors being used to determine a material’s hardness or incorporated into wearables.

The research appears in the journal Sensors.