Thesis in Big Data & AI: Industrial Digital Twins for Physical AI (m/f/d)

Your tasks – shape the future with us:

In this master's thesis, you will develop scalable digital twins for construction and industrial applications. You will transform visual and technical information into standardized, machine-readable 3D representations that model structural elements, industrial components, and their environments with geometric, spatial, physical, and semantic consistency. This will result in physically grounded models for robust AI perception and future applications in agentic robotics and physical AI, including scene understanding, motion planning, and action execution.

The following tasks await you:

• Computer vision and 3D perception: You will further develop existing AI-supported approaches for the recognition and 3D reconstruction of selected industrial components from visual data. The information gained from this can also support downstream robotic reasoning.
• Multimodal model generation: You combine visual information with technical data, drawings and product knowledge to create and refine digital twins.
• Physical and spatial consistency: You investigate how physical, spatial, and semantic requirements can increase the consistency and reliability of the generated models.
• CAD/BIM-related standardization: You will further develop an agent-based pipeline that transforms multimodal information into machine-readable models compatible with CAD/BIM. This can support real-to-sim-to-real workflows.
• Scalability and evaluation: You evaluate the approach based on defined component classes and data conditions with regard to quality, robustness and transferability.
• Prototypical implementation: You will create a demonstrator to illustrate the results. Depending on the project progress, you can investigate how the generated representations can be used for multimodal robotics approaches, such as Vision-Language-Action (VLA) models.

Your profile – grow together:

• Master's degree in a STEM field, computer science, statistics, mathematics or comparable degree programs (physics, chemistry, economics, empirical social research, etc.).
• Basic knowledge of Microsoft Office.
• You already possess programming skills in Python, SQL and possibly another language.
• Knowledge and experience with Amazon Web Services; experience with GCP and Databricks is a plus.
• Prior knowledge of software development is also advantageous.
• You have already gained experience with common machine learning frameworks and libraries (e.g., TensorFlow, PyTorch, OpenCV, Open3D, PyMeshLab, Transformers, scikit-learn).
• Fluent English skills and basic German knowledge complete your profile.

Your benefits in an appreciative corporate culture:

• Your commitment will be rewarded. For a mandatory internship, you can expect €1,400 gross per month. If you complete a voluntary internship for four months or longer, or write your thesis with us, you will receive €2,541 gross per month.
• Stay active and healthy with our sports and relaxation courses - on-site and online - as well as the use of over 10,000 sports and wellness facilities with the employer-sponsored EGYM Wellpass.
• Our internship program “GO WÜRTH” stands for “orienting together” and offers different modules each semester. Events like “HELLO NEWCOMER” give you the opportunity to network and gain insights across departments.
• The most important word in our interactions is THANK YOU. You can expect a friendly and welcoming atmosphere where you can further develop your existing strengths and discover new ones.