DiProLeA

The product development process is split into three stages: concept, development and production. Companies’ product-specific information from each of these stages is still stored in various systems and databases for specific tasks. Connecting these systems and databases offers great potential to help manage more product variations and meet higher quality standards in manufacturing. By working together with process partners, a flexible and industry-wide smart learning assistance system for product development will be developed.  

Aschaffenburg UAS contributes research and concepts to designing the assembly workplace of the future. This is particularly important, as manual assembly processes carried out by people will remain indispensable in the digital factories of tomorrow due to their high flexibility and the precision they provide. It is integral to innovatively and intelligently support workers in these often repetitive and concentration-demanding tasks.

Current assistance systems for manual assembly processes are usually expensive and only optimized for specific tasks. These systems are not capable of generating dynamic assistance data from existing company data, making them inadequate for highly varied manufacturing.

Within the framework of the project a flexible, smart assistance system to support small-scale manual assembly processes is to be developed that can be used effectively in practice. The system is linked to a semantic knowledge base and a process execution system, which allows tracking and tracing of individual process steps. Data is matched with the knowledge base in real-time to ensure a seamless digital picture of the entire process.

Workers are effectively trained and supported during varied assembly tasks. The system provides guidance for correct assembly and automatically records the procedures. This results in the assistance system not only offering real-time help but also improves work processes.

During development, the main focus was on making the system economical and flexible. Instead of using expensive, manual, and fixed methods, self-learning methods through simulations were used. This allows the assembly process to be adapted to new conditions quickly and cost-effectively.

The smart assistance system consists of several cameras that cover essential areas of the workspace. The data captured by these cameras is processed by an industrial Computer and integrated into the company’s infrastructure. From this infrastructure, the data can be accessed by other devices, like augmented reality glasses, to display processes at the work station.

Various detection algorithms are used based on the needed accuracy and work step. These algorithms range from 2D detection to 3D hand recognition and 6D object pose estimation. This variety of algorithms ensures that the specific requirements of each assembly step are optimally met, providing high accuracy in detection and tracking.