According to an article published in Azonano, flexible touch sensors could improve the operation of robots. It is a great step towards the creation of robots capable of identifying and manipulating objects in unstructured environments.
According to Chang Liu, an Illinois professor of electrical and computer engineering, researchers are developing artificial touch sensors that will mimic the functionality and efficiency of biological structures like human fingers. Using low-cost sensor arrays, it is possible to analyze and identify the texture of different surfaces.
Biological sensors provide a wealth of information about an object's shape, hardness, and texture. However, robots that normally have a single pressure sensor when grasping something are unable to determine whether the object is hard or soft, or whether it is exerting pressure on the object or not. This causes problems for them when handling delicate objects like eggs for example. As another professor, Douglas Jones, explains in the article, the distributed sensors that humans have in our hands allow us to take an egg with enough force so that it does not fall, but without breaking it. The team's goal is to develop a suite of sensors that provide the same kind of capability to robotic systems.
The sensors are manufactured from an inexpensive polymer layer by photolithography. In one sample, scientists have created a 4 × 4 array (16 sensors). Each sensor measures about 200 microns and has a tiny bulge in the center. On that surface, a pressure gauge is deposited whose resistance changes as it is stretched. The pressure on the sensor is converted into digital information that is sent to a computer and is analyzed with an algorithm for signal processing. The algorithms developed by the team are advanced, allowing to quickly determine which sensors have been activated in the set, and whether the object is flat, box-shaped or X-shaped.
The scientists aim to improve efficiency in the near future by simplifying the algorithm for signal processing so that it can be executed by circuits mounted on the same substrate as the sensor.