DIEFEDA - direct error feedback during assembly

Exploring Project, Quick Check

Initial situation

Digitalization is advancing in the industrial sector, while many assembly processes are still the task of humans rather than machines due to their complexity. Manual activities are subject to human repeatability. Despite the high level of expertise of skilled personnel, the processes therefore have a low error rate, which is crucial for quality.

However, customers from both the professional and consumer sectors attach great importance to quality. As a result, these minimal potential errors must be eliminated by means of complex final inspections in order to guarantee a safe and functional product at all times. This process is financially costly compared to the low probability of occurrence. It also leads to considerable rework and places an additional burden on the final inspectors (see Quick Check "final inspection 4.0"). Modern AI solutions for assembly support are therefore a more suitable alternative.

Solution idea

Direct feedback after an error has occurred using AI feedback should enable errors to be rectified in real time at the assembly station. The final inspection can then be condensed to the relevant fault evaluations. This should reduce the workload of final inspectors and increase the safety of assembly personnel. This supportive use of AI systems can be particularly useful for complex or repetitive, monotonous assembly cycles. However, the extent to which this idea can be used in an assembly system to ensure relevant quality features must first be investigated in a preliminary analysis.

Benefit

With supportive indications from an AI system in the event of deviating behavior, the specialists on the assembly line can question the deviations and intervene to correct them at an early stage. This means that the decision-making authority remains with humans and they continue to be an important part of the solution. However, the psychological stress (e.g. due to monotony) in the work process is reduced by a greater sense of security.

Cost reduction also plays an important role. This is because there is a fixed pattern in industrial development that higher quality also means more testable quality features in the final inspection. By breaking this pattern, the final inspection can be reduced to the essentials. The skills of humans as well as the advantages of the machine are used to improve competitiveness in Germany as a high-wage location.

Implementation of the AI application

An AI system based on activity recognition was developed for the assistance system at the assembly station. The system uses image data to recognize relevant process steps in real time during execution and compares them with the target status. The comparison is based on the predefined quality characteristics of the assembly process. Deviations from the quality standard are detected in real time, enabling staff to make immediate corrections. This supports the company's zero-defect target.