bit-tech.net

Lernstift Linux-powered pen turns to Kickstarter

Lernstift Linux-powered pen turns to Kickstarter

The Lernstift pen, powered by a Linux-based computer-on-module, vibrates when you make a spelling mistake or badly-formed letter.

A German company has launched a pen with a difference on crowd-funding site Kickstarter, which promises to correct your spelling as you hand-write notes.

Lernstift - German for 'Learning Pen' - is a twist on the classic concept of a smart pen which uses some clever technology to bring the best of the digital realm to the very analogue art of handwriting. Developed, originally, as a means of improving one of the co-inventor's son's homework, the system is based around an ultra-compact computer - a variant of the Gumstix Overo computer-on-module which runs a customised Linux distribution. With integrated motion sensor, the pen is capable of recognising handwriting - and subjecting it to some very clever analysis.

Primarily, the Lernstift is designed to be used in orthography mode. In this mode, the system translates handwritten notes into text and subjects it to spelling checks. If a word has been written incorrectly, the pen vibrates - alerting the use in much the same way as the traditional red squiggly line would in a word processing application. A second mode, calligraphy mode, is designed to improve penmanship in young children by recognising when the shape of a letter may be difficult to read.

In addition to these two educationally-focused modes, the pen can also perform the same tricks as a traditional smart pen. Handwritten notes can be saved as images or turned into editable text, and images can be captured without having to scan the paper. Thanks to its reliance on a motion sensor, the pen also works on any writing surface - fountain pen and ballpoint tips are promised at launch, with a pencil tip to follow - unlike rival devices which demand expensive, custom paper with a predefined grid pattern imprint.

The team behind the project promises to release its application programming interface (API) under an open licence, to allow third parties to connect to the pen - using an embedded Wi-Fi radio - from desktops, laptops or mobile devices and create applications that can make use of its capabilities.

Thus far, the two founders have created prototypes thanks to cash from private investors who have all pledged money through the company's own website. Now, however, they are turning to Kickstarter to fund commercialisation and production. Asking for £120,000 in the next 28 days, the team hopes to have an initial production run completed by the end of October.

More information on the device is available on the team's Kickstarter page - which is, in a wonderful display of irony, littered with spelling and grammatical errors from the English-as-a-second-language pair and their eight current employees.

3 Comments

Discuss in the forums Reply
Guinevere 11th July 2013, 14:42 Quote
The trouble is accurate handwriting recognition is HARD to do.

Kids also twist and turn their pens and change their writing speed on the fly. They also go back and correct letters they've already written. A motion sensor based ben is going to have it's work cut out to translate kids writing into useful data.

Kids leaning to write and spell don't write in a way that makes it easy for handwriting recognition software to use easily ( I have 6yo twins - trust me! ). So even if you can capture the data correctly recognising it will be very difficult.

The lernstift guys are only at the stage where they're recognising 'block' characters and have managed two four character words.

They have a long way to go before this is ready for their intended uses. Looks like the hardware is okay but the software looks like it needs a LOT of work.
magnetobob 11th July 2013, 21:26 Quote
I think OCR technologies in general need to improve. However this is a nice principle but not for my handwriting as it is awful.
fluxtatic 15th July 2013, 07:06 Quote
Just like moving sidwalks, flying cars, etc - we should damn well have OCR that doesn't suck now that we're well into the 21st century.

I was just messing with it the other day, hoping to find OCR that could work with scanned documents - it's a hair better than it was when I last tried a few years ago, but it's still faster to just retype it from the hard copy >:(
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums