Artificial intelligence facilitates writing with a smartphone


Researchers at ETH Zürich University have developed a new artificial intelligence solution that allows touchscreens within smartphones to sense an accuracy eight times higher than current devices.


This could mean fewer typos on smartphones in the future, regardless of phone type and the size of the keyboard keys across the small screen.


The new artificial intelligence solution can more accurately deduce the location of the fingers on the touch screen.


The researchers in the project say: The challenge facing writing via modern smartphones is that touch sensors that detect the location of fingers on the screen have not changed much since the mid-2000s.


Although these sensors haven't changed much, the screens themselves have improved significantly with a higher resolution.


Smartphone and tablet screens now provide high visual quality, which is becoming more evident with each new generation of devices. For example, the latest generation of iPhones has a resolution of 2532 x 1170 pixels.


But the touch sensor can detect input with a resolution of only about 32 x 15 pixels, which is about 80 times less than the screen resolution.


Capacitive touch screens detect the position of the fingers by using changes in the electric field between the sensor lines to sense the proximity of the finger when touching the screen surface.


The team says: Given that the capacitive sensor detects proximity, it is unable to detect real finger contact.


The method the team found is called CapConnect, and it combines two approaches. The technology uses a touch screen as an image sensor that can see about eight millimeters and a depth camera that records an image of how close an object is.


CapConnect exploits insights to precisely discover areas of contact between fingers and surfaces, using the team's deep learning algorithm.


The team showed that the new AI solution reliably recognizes touch on surfaces even when fingers touch the screen close to each other.


The researchers believe that the AI ​​solution could pave the way for new touch sensing in future cell phones and tablets to work more reliably and accurately, but while reducing the footprint and complexity in terms of sensor manufacturing.

2 views0 comments

Recent Posts

See All

L0phtCrack is now open source

L0phtCrack is a password checking and recovery application that has been around since 1997. The program for Windows has a checkered history