Research that Moves Into the Real World

OPTIFARM focuses on the development of AI-based image analysis and machine vision methods for agricultural monitoring, crop assessment, and automated quality inspection. The project explores robust detection, classification, and data-driven decision support to improve efficiency, consistency, and automation in real-world agri-food environments.

UV4Plants focuses on the development of UV-C illumination systems for controlled plant treatment in greenhouse environments. The project explores the effects of targeted UV-C exposure on plant growth and health, aiming to support efficient, automated, and sustainable cultivation through precise control of irradiation conditions.

Hardening increases the hardness and wear resistance of steel workpieces, but the uneven thermal and phase-transformation stresses it induces also cause shape distortion. The HARDFORM project focuses on predicting and correcting these distortions—from capturing the workpiece response during the process to modelling it with deep neural networks that predict the final shape and guide the straightening operation. The goal is more accurate, repeatable, and automated control of hardened-workpiece geometry in an industrial setting.

Translating a part’s geometry into machine-ready NC code is still a largely manual, expertise-driven task. CAD2NC investigates how geometric and process information can be encoded and passed to transformer models that generate NC code directly from CAD-derived representations. By studying how the explicitness of this representation shapes the output, the project shows that representation design drives the reproducibility of generated code far more than raw correctness—a step toward reliable, automated CAD-to-NC pipelines.