Bot4Lib
- Call for solution type:
- Innovation Partnership
- Dates of Public Procurement:
- from 25-12-2024 to 30-01-2025
- Demo Day:
- 22-09-2025
- Status:
- In development
Identified challenge
The primary goal of the project is to automate the inventory process in the reading room of the National Library by implementing an innovative solution that integrates an autonomous robot with an intelligent document recognition system. This robot will navigate public areas to identify individual documents using computer vision. The identification is done using visual information, data stored in Data Matrix codes and / or RFID tags affixed to the documents. The robot will send the collected data to a management system that, in a final step, allows BnL staff to rectify any shelving errors.
This recognition task is complicated by the wide variety of documents, which come in different formats, can be very thin, and sometimes lack distinctive markings (such as text or images) on their spines.
Process
Launch of the call for solutions
Space Time S.A. and Data Design Engineering Sarl participated in the tender and handed in a concept note for the Bot4Lib innovation partnership. Since the number of candidates was below the maximum of five candidates that can advance to the Proof of Concept phase, there was no need for a first evaluation by the jury.
The jury was composed of the following five members :
- Carlo Blum, National Library of Luxembourg
- Michael Scharwächter, TU Dortmund University
- Dominique Riefstahl, Ministry for Digitalisation
- Rui Maciel, Ministry for Digitalisation
- Pascal Gieres, Government IT Centre
Proof of Concept phase
The Proof of Concept phase allowed the candidates to put their concept notes into practice. The companies had time to develop a prototype of the robot and test it during regular test sessions at the National Library. During this period, the BNL team started putting Data Matrix codes on the book spines of the books in the reading room.
Two test shelves with a total of 165 books were prepared for the Demo Day. The shelves were prepared in a way that each type of error could be tested. The different types of errors were:
- Books placed on the wrong shelf;
- Books present on the shelves that are not included in the theoretical model;
- Books placed on the correct board but not in the correct place;
- Books that are supposed to be in place but that are actually missing.
Demo Day
During the Demo Day, each candidate had an allocated time slot to bring their robot to the National library and perform an inventory of the books on the test shelves. They each handed in their results upon completion of the inventory process. The BNL team then proceeded to analyse the results of both candidates and identify the number of errors detected by each robot.
Both candidates were then invited to pitch their solutions in front of the jury, followed by a Q&A session and a discussion of the inventory results.
The evaluations were made according to the following criteria :
- Quality of the prototype
- IT functioning of the prototype
- Skills of the applicant/team
- Planning & budget
The Demo Day ended with the jury's deliberation and completion of the evaluation sheets.
Design phase
After the Demo Day, the winning team was announced. Data Engineering Sarl was invited to move forward to the Design phase. This phase is designed to develop the functionalities and architecture of the future solution in a series of workshops with the selected candidate and the BNL team. The objective is to establish a final offer for the implementation of the solution.