Library puts robots to work reading books

Citation

(2008), "Library puts robots to work reading books", Industrial Robot, Vol. 35 No. 5. https://doi.org/10.1108/ir.2008.04935eaf.001

Library puts robots to work reading books

Article Type: Mini features From: Industrial Robot: An International Journal, Volume 35, Issue 5

When the Bavarian State Library in Munich decided to digitize much of their collection of books, some more than 400 hundred years old, they “hired” two robots to start reading. The book-scanning robots developed by Treventus Mechatronics GmbH of Wein can read (digitize) at about 700 pages per hour. Markus Brantl, director of the library’s digitization center, estimates it will take until 2009 to digitize the 7.5 million pages in that part of the library’s collection planned for this project.

The goal is to digitize about 1 million of the 9 million possessions in the library holdings. The project will focus on including about 80,000 medieval manuscripts and 20,000 incunabula books (meaning printed before 1501). The materials go back to the time of Martin Luther. The goal is to make these rare books and manuscripts available to scholars on-line. The Bavarian State Library is in partnership with the internet company, Google.com, to make available digitized access to print materials that are no longer under copyright restrictions.

The ScanRobot™ SR300 (Figure 1) is the development of Treventus Mechatronics, formed by a team of robot engineers affiliated with the Vienna University of Technology. Professor Wolfgang Zagler, Department of Rehabilitation Engineering, came up with idea for the robot while on a train ride as he tried to address the needs of that visually impaired students have in trying to accessing printed materials. Development seed money from several sources help fund the project.

Figure 1 View of book reading robot with book at the ready in cradle and operator control station on the left

The robot can digitize pages up to 32 cm in height and from 12 to 30 cm in width. The book thickness can range to 12 cm maximum. The system handles page thickness from 60 to 135 g/m².

Depending on the book width and the paper quality, the robot can scan up to 25 pages/minute.

Principle of operation

Books are gently strapped to a 60° wooden slat holder frame. The straps pass over just the book covers. A robot arm with gentle vacuum action then moves down toward the spine of the book and lifts one page to read it (scan) and then turn it to the other side. The scan head includes two precision optical glass prisms that direct the image of each page to the two high-resolution image capture cameras (Figure 2). Onboard light sources evenly illuminate the pages. After the images are captured, the vacuum arm repeats its page scanning and turning assignment until all the pages have been scanned.

Figure 2 Close up of scan head gently reading a page

The 60° book cradle helps prevent damage to the book binding which is more likely to occur if the book is opened wider. The slat design of the holder allows the operator to position the hold down straps to more exactly match the dimensions of the book.

Illumination and scan technology

The pages are illuminated only as the scan passes over each particular line. The light is provide by an innovative LED system developed by Treventus, which provides only cold light and does not expose the printed materials to heat, infrared or UV energy. A very important benefit when handling priceless ancient books that might be damaged by excess exposure to light.

Computer controller details

The system is directed by a Hewlett-Packard PC-workstation running Windows XP^® operating system. A 750 GB hard drive serves as the memory storage. A 2 B RAM provides the dynamic memory. The operator display is a 53 cm wide screen color monitor. The robot reader operation is directed by the ScanGate software package.

ScanGate™ Software

SoftGate software running on the PC-platform directs the operation of the robot. Text information is converted into computer readable format using optical character recognition technology. The package includes the ability to OCR Gothic script, musical notations (optional) and up to 170 languages.

The development team

The concept was brought to “life” by the development team lead by Professor Wolfgang L. Zagler and including students Vienna University of Technology (at the time) Markus Barth, Christoph Bauer and Stephan Tratter. The three students adopted “the robot that reads” concept as the basis for their diploma theses.

Funding was partially provided by an entrepreneurship agency of the Austrian Government in combination with others. A number of automation device suppliers also contributed advice during the development.