Starting situation
For the Fraunhofer Institute for Production Technology and Automation IPA, evia developed an AI assistance system for evaluating components for automated assembly processes as part of the KIMont research project.
The solution analyzes CAD models already in the design phase and identifies optimization potential for automated manufacturing and assembly processes. This allows for early detection and improvement of challenges related to handling, positioning, and joining.
Challenges
Evaluate abstract component properties
Geometries and properties from CAD models had to be analyzed automatically.
Recognize potential early
Optimization opportunities should be visible even before production.
Training AI models reliably
Neural networks had to deliver valid results for different component types.
Consider complex production requirements
Handling, positioning, and availability had to be included in the assessment.
Provide scalable AI infrastructure
The training and AI services had to be operated with high performance and flexibility.
Connecting research and practice
The solution should not only work theoretically but also be directly integratable into industrial processes.
Solution
What our expert says
Steffen Tauber
Head of Research
„Designing a component and being able to assemble it automatically are two different things. In the KIMont project with the Fraunhofer IPA, we developed an AI system that closes this gap: CAD models are already checked for handling, positioning, and joinability during the design phase. This way, problems land on the designers“ desks during the design phase – and not in production, where they are ten times more expensive.“
More stories
