Thanks to our sense of touch, we can figure out the best way to grab keys in our pocket without ever looking at them. And now, thanks to engineers at the University of California — San Diego, robots could soon have the same skill.
Usually robots need to be trained or have a camera in order to see and grip an object. However, that can be challenging in low-light scenarios or when encountering new objects. This new soft gripper can decipher the shape of unknown objects just by touching it.
With the three pneumatic fingers — covered in a sensing skin — electrical signals transmit from the fingers to create a 3D model of the object, the gripper can then grasp and manipulate the object. In tests, the gripper was able to screw in a lightbulb in addition to picking up other items.
Researchers say that by combining the system with artificial intelligence the technology could be improved to eventually identify objects just by what they feel like.