Waldkirch: Thesis: AI Meets 3D - Deep Learning Methods for Real-Time Logistics Applications

Summer Semester 2026 – Limited to 6 Months

Automation in logistics increasingly relies on the analysis of 3D data. A central application is the shape and object recognition of shipped goods. However, processing large point clouds presents a major challenge, as millions of measurement points must be reduced to the most relevant information. The aim of this thesis/internship is to research and develop deep learning methods for the efficient reduction and analysis of 3D point clouds. Real-world data and a practical application scenario will be available for evaluation.

YOUR TASKS:

• Research current deep learning approaches for processing 3D point clouds and their use in shape recognition

• Familiarize yourself with existing technologies for 3D data analysis

• Develop methods to reduce point clouds to the most information-rich points

• Conduct and evaluate experiments

• Document your findings

YOUR PROFILE:

• Degree program in Computer Science, Mathematics, or a comparable field

• Programming experience in Python or C++

• Knowledge of deep learning and 3D data processing

• Good command of written and spoken English

• Independent, structured, and reliable working style

Contact: Sarah Disch