AI increases touch screen precision

Two researchers from the Swiss Federal Institute of Technology in Zurich (ETH) have developed an artificial intelligence (AI) that enables devices to more precisely identify where a finger touches the user interface. Computer science professor Christian Holz and doctoral student Paul Streli from the Sensing, Interaction & Perception Lab are now presenting their AI solution for more precise touch screens at the specialist conference on human factors in computing systems, CHI 2021, according to a press release.

In their presentation they show that a third of the errors in current devices are due to the fact that input sensing is only low resolution. While the displays of mobile end devices have become increasingly higher-resolution, the sensors that detect input by a finger on the touch screen have barely been developed since their introduction in the mid-2000s. The resolution of the most recent iPhone is 2532 x 1170 pixels, but its touch sensor can only detect inputs with a resolution of around 32x15 pixels. This is almost 80 times lower than the screen resolution. The new solution from Holz and Streli, CapContact, eliminates these errors with a type of machine learning – deep learning.

According to the press release, the AI solution has a good chance of being used for new touch recognition in future cell phones and tablets. The system even distinguishes between touch surfaces when fingers move very closely together on a screen in order to enlarge text or images. Current devices can barely differentiate between these touches.