Design Validation Model
DFG-funded research project at Trier University
XDP-Opt
Experience-Based Design Process Optimization for Industrial Manufacturing.
XDP-Opt develops an interactive design decision support system that validates manufacturability in CAD-based product development and recommends solutions from historical design experience.
- Focus
- Design for Manufacturing
- Core methods
- Foundation Models, CBR, FL
- Funding
- DFG 543073350
The challenge
Manufacturing knowledge is hard to keep inside the design loop.
Product designers need to account for production capabilities, customer needs, prior designs, lifecycle constraints, and tacit expert knowledge. Traditional DfX software can detect rule violations, but often misses nuanced constraints and rarely proposes adapted solutions.
XDP-Opt addresses this gap with data-driven, explainable, and experience-based support directly around the CAD workflow.
Research direction
A hybrid AI core for early design decisions.
The project combines learned manufacturability assessment with reusable design cases so designers can see both the problem and a practical route toward a fix.
Design Recommendation System
Turn past cases into adapted solutions
Case-Based Reasoning retrieves similar resolved issues and proposes Engineering Change Requests that product designers can accept, adapt, or reject.Federated Learning
Collaborate without exposing raw design data
Local models can learn from company-specific knowledge while federated training explores stronger cross-company assessments for industrial manufacturing.Architecture
From design review to reusable experience.
At decision points in the development process, XDP-Opt imports the current design state, creates Problem Reports for detected issues, and proposes Engineering Change Requests linked to similar past cases.
IDDSS
- 1
Initial design
- 2
Design review
- 3
Solution finding
- 4
Solution integration
What we investigate
Research questions with industrial grounding.
The project is designed around CAD-based new product development and a prototype in the SmartFactoryKL industrial truck manufacturing test environment.
- Vision-language models for 2D technical drawings and CAD-based design artefacts
- Reusable DfM knowledge representation for problem reports and change requests
- Similarity search over complex geometry, materials, constraints, and prior outcomes
- Explainable recommendations grounded in retrieved design experience
- AI-based product design space exploration for cold-start situations
- Prototype validation in the SmartFactoryKL industrial truck test environment
Project consortium
Research by Trier University and RPTU.
XDP-Opt brings together expertise in case-based reasoning, workflow and knowledge systems, federated learning, production systems, and product development.
Trier University
Kokulan Thanabalan, Lukas Malburg, Ralph Bergmann
RPTU Kaiserslautern-Landau
Leonhard Kunz, Tatjana Legler, Simon Bergweiler, Martin Ruskowski
Connect
Follow the development of XDP-Opt.
Read the concept paper or visit the project profile for updates from the XDP-Opt team.