Featured Research · 2026
Academic research in 2026 across generative modeling and robotic policies. The work covers identity unlearning in diffusion models, token grounding, and expert-guided representations for vision-language-action models.
Identity-conditioned generations illustrating the PIU research direction.
01 · First author · IJCB 2026
Targeted identity unlearning for ID-conditioned diffusion models, designed to forget requested identity information while preserving generation quality.
Computer Vision Laboratory, University of Ljubljana
Language-conditioned queries expose the X-VLA image-token regions used during manipulation.
02 · Research project · 2026
Language-conditioned queries score the native X-VLA image-token grid to reveal which visual regions matter for the current manipulation step.
Visual Cognitive Systems Laboratory (ViCoS), University of Ljubljana
Expert perceptual structure recovered from frozen X-VLA representations for action guidance and alignment.
03 · Thesis work · 2026
Recover expert perceptual structure from frozen X-VLA representations, including DINOv2-like tokens and CeDiRNet-aligned maps, to support action generation and representation alignment.
Visual Cognitive Systems Laboratory (ViCoS), University of Ljubljana