Artificial Intelligence Helps People with Assembly Tasks

In the future, a new assistance system for manual assembly stations will support workers individually thanks to artificial intelligence. Assembly faults and process times will be reduced. BIBA - Bremen Institute for Production and Logistics at the University of Bremen has presented a prototype.

The assistance system analyzes the process data and camera-based information recorded and collected at the assembly station. Image processing and machine learning processes are then implemented with regard to the ergonomic and production-related work situation. The system assesses the assembly process as well as the quality of the finished product and integrates employee-oriented assistance functions. To date, assistance systems have only focused on the products that need to be made and their quality.

“The Working Human with Individual Skills in the Foreground”

“With the assistance system, we do not want to replace human work but rather support it in the best possible way. The work is to become easier and possible for the disadvantaged on the job market,” says Christoph Petzoldt, BIBA researcher and project leader. “This assistant provides individual support depending on what is needed, it motivates people, and creates a constructive working environment. We are putting the working human with individual skills in the foreground.”

How it Works

Optical sensors (depth cameras) record the process progression at a working space from several perspectives, for example they record movements such as the removal of single components from the supply containers and the assembly task in itself. The cameras deliver said image and depth data to the system that as a first step recognizes them and analyzes them in real time with the help of an image processing procedure.

The system uses deep learning methods  - a key artificial intelligence (AI) technology - for the analysis of posture (ergonomics) and for the hand tracking with regard to the employee. The system improves with each single calculation, as it bases its analyses and prognoses on what it has already learnt. The assistant is individualized based on this analyzed information and general process data, such as process times and faults.

The information generated with the help of AI prepares the system for varied usage - initially for manual work directly at the specified assembly station. Images that accompany current work appear on the working space using a projector. If required, the workers also receive additional information and aids - on the one hand regarding the technical assembly task with options to learn a little more at the same time, and on the other hand concerning health-protecting, individual optimization of their posture whilst working.

By viewing the assembly progress, the targeted provision of information, as well as the consideration of employee needs, the system increases process efficiency as well as assembly quality and also improves the working situation by means of specific support in the form of motivation and further training strategies and techniques.

Motivation and Additional Learning through Gamification

One aim is also the better and more expedient training of staff, as well as increasing individual motivation. Gamification is the systematic motivation through incentives and refers to the process of game-based learning with techniques that have their original use in the world of computer games and have been further developed for implementation in industry. The work process is more ergonomic and stimulating for employees by means of the gamified presentation of information. Whilst carrying out their practical work they also learn in a “play-related” way. Using the data provided by the AI system, the gamified elements are controlled by the game-design concept.

Efficient, Effective, and High level of User Acceptance

With the help of novel assembly-assistance functions that stem from the combination of informational process controlling with projections, automatic supervision of assembly processes and component progression, ergonomic posture recognition, and incentive-based gamification, a clear reduction of assembly faults and process times was achieved. A great increase in efficiency was ascertained with regard to confirmation steps. Additionally, the user studies that accompanied the project showed that support measures and incentive forming lead to a high level of employee acceptance.

“Securing Participation of Small and Middle-Sizes Companies in Industry 4.0”

“With this assistant system, a solution has been created that considers social and economic aspects and meets the demands of the current situation on the job market. It makes the integration of people who still miss out during selection processes today, for example due to their age, a handicap, or lack of training, possible whilst continuing to guarantee high production quality,” states BIBA director Professor Michael Freitag. “The system can be implemented for manual assembly processes in companies of all sizes and all industries. It secures the participation of small and middle-sized companies in rapidly progressing industry 4.0 developments.”

The “AxIoM” Project

The cooperation project “Gamifiziertes KI-Assistenzsystem zur Unterstützung des manuellen Montageprozesses” (Gamified AI Assistant System for the Support of Manual Assembly Processes) (AxIoM) ran for 22 months under the direction of BIBA. Armbruster Engineering, a Bremen specialist for assembly assistance system, was involved as a development partner. The Bremer Aufbau Bank (BAB) funded the project with funds from the European Regional Development Fund (EFRE) and also accompanied it. The prototype of an AI-based assistance system for an assembly station in manual production, for example for small-scale production, is what was produced.

Sabine Nollmann


Further Information:


Prof. Dr.-Ing. Michael Freitag
Institute Director
BIBA - Bremen Institute for Production and Logistics GmbH
Phone: +49 421 218-50002


The working human with its individual skills stands in the foreground of the system developed as part of the AxIoM project.