ViKIMon

The increasing variety of product variants and the growing complexity of assembly processes pose significant challenges for small and medium-sized enterprises (SMEs) in quality assurance. Conventional quality inspection methods for conformity assessment of product quality are reaching their limits, as they fail to ensure sufficient inspection coverage and accuracy. The integration of artificial intelligence (AI) offers potential for improvement in this area; however, due to a lack of guidelines on the trustworthiness of AI systems, its adoption remains hesitant. In particular, uncertainties regarding performance, reliability, and transparency lead to a trust deficit among users, which hinders the implementation of AI-based quality assurance methods in assembly.

The research project ViKIMon aims to significantly enhance the digital sovereignty of small and medium-sized enterprises (SMEs) in the application of AI-based quality inspections in the field of assembly. In this context, digital sovereignty refers to a company's ability to maintain its freedom of action and decision-making within the framework of digital transformation. To achieve this, a quality management method is being developed that uses the requirements of the European Artificial Intelligence Act (AI Act) as a foundation and is further extended by specific stakeholder requirements (e.g., from companies and assembly workers).

At the beginning of the project, the requirements for trustworthy AI in quality inspection are identified. This involves systematically capturing both regulatory requirements from the Artificial Intelligence Act (AI Act) and specific demands from companies and assembly workers. Expert interviews and methodological analyses, such as the Kano model, help define practical criteria for the trustworthy design of AI-based quality inspection.

Building on this, a digital infrastructure for the collection and processing of quality data is developed. This includes the creation of a hybrid assembly system that automates the acquisition of measurement data and integrates it into an intelligent inspection process. This structure forms the foundation for a reliable and transparent AI-supported quality assessment.

The core development of the project focuses on the AI-based quality inspection itself. Algorithms are designed to ensure high precision and inspection coverage while also guaranteeing explainability and transparency. The goal is not only to achieve technical accuracy but also to strengthen users’ trust in system decisions.

Another key focus is the human-centered design of human-machine interaction. The user interface is developed to enable users to understand AI decisions and intervene if necessary. This ensures that inspection systems are intuitive to operate and seamlessly integrate into existing workflows.

Subsequently, the developed system is tested through experimental validation. In real-world scenarios, the AI-based quality inspection is evaluated to determine whether it meets the required standards of trustworthiness, transparency, and acceptance. The insights from these tests are used to iteratively optimize the methodology.

Finally, the developed concept is translated into a sector-neutral IT tool that facilitates simple and secure implementation in industrial practice. Training sessions and workshops support SMEs in independently using the technology, thereby strengthening their digital sovereignty in handling AI-based quality assurance systems.

The project ViKIMon (No. 01IF23416N) of the Gesellschaft für innovative Betriebsorganisation (GBO) e.V. - Joseph-von-Fraunhoferstraße 20, 44227 Dortmund - is funded via the DLR within the program of consortial industrial research and development (IGF) of the Federal Ministry for Economic Affairs and Climate Action (BMWK).

Portal Industrielle Gemeinschaftsforschung (IGF)

DLR Projektträger

Bundesministerium für Wirtschaft und Klimaschutz (BMWK)

Gesellschaft für innovative Betriebsorganisation e. V. (GBO)