Communication Dans Un Congrès Année : 2025

Bottlenecks of Vision Language Models in Image to PDDL State Extraction

Résumé

The advent of Large, Pre-trained, Vision or Language Models (LLM, VLM, etc.) has lead to their wide use in multiple robotic applications. To better understand how they can be leveraged in robotics applications in generating a precise description of an environment from an image, we evaluate the generation of Natural Language (NL) descriptions from images by comparing different prompts, image-to-PDDL data and state-of-the-art Vision-Language Models (VLMs). Results reveal specific strengths and weaknesses of individual VLMs and a strong reliance on image complexity in terms of object shape, cardinality and position, when generating NL descriptions.

Mots clés

Fichier principal
Vignette du fichier
IRCC_2025-4.pdf (127.73 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05425715 , version 1 (19-12-2025)

Licence

Identifiants

Citer

Gaëlic Bechu, Ehsan Abbasnejad, Mihai Andries, Panagiotis Papadakis. Bottlenecks of Vision Language Models in Image to PDDL State Extraction. IEEE Robotic Computing & Communication, Dec 2025, Naples, Italy. ⟨10.1109/RoboticCC68732.2025.00011⟩. ⟨hal-05425715⟩
301 Consultations
198 Téléchargements

Altmetric

Partager

  • More