Nov 2025 – Present · 5 mos · Munich, Bavaria, Germany

• Safe Reinforcement learning for VLA models.

Oct 2025 – Present · 6 mos · Munich, Bavaria, Germany

May 2025 – Oct 2025 · 5 mos · Karlsruhe, Baden-Württemberg, Germany · On-site

• HiWi Research assistant at Intuitive Robots Lab.
• Fine-tuned and deployed VLA models (GrootN1, SmolVLA, FlowerVLA) on a Franka Emika Panda.
• Analyzed tactile embeddings from models trained with self-supervised and contrastive learning using PCA and t-SNE, and integrated them into existing policies (VLAs, BESO, Diffusion Policy).
• Implemented and tested multi-modal sensor fusion approaches, using saliency maps to verify the effective use of language, visual, and tactile modalities.