Hugging Face Blog
Pollen-Vision: Unified interface for Zero-Shot vision models in robotics
Read the original on Hugging Face Blog →The Flow has not summarised this story yet — read it at Hugging Face Blog.
Related stories
arXiv Computer Vision
Oct 8
Agentic Scene Policies
arXiv:2509.19571v2 Announce Type: replace-cross Abstract: Designing or learning robot policies that generalize zero-shot across a range of language instructions and objects is a core problem in robot...
arXiv AI
Sep 10
One MLLM, One Call: Efficient Zero-Shot Vision-and-Language Navigation via Spatial-Aware Waypoints
arXiv:2609.06476v1 Announce Type: cross Abstract: Vision-and-Language Navigation in Continuous Environments (VLN-CE) requires an embodied agent to navigate unseen environments by following natural la...
arXiv Machine Learning
Sep 24
Generalizable Robotic Insertion with World Models
The paper introduces a framework that uses world models to enable robotic insertion across diverse parts. By combining proprioceptive data with wrist‑mounted camera visuals, a single model is trained on up to 90 tasks, achieving 56% zero‑shot success on unseen objects versus 7% for a model‑free baseline. The approach scales with more training objects and can be fine‑tuned for improved data efficiency and performance.
arXiv Computer Vision
Aug 25
OptiSight: Bridging Semantic Reasoning and Geometric Control for Embodied Navigation
arXiv:2608.23354v1 Announce Type: cross Abstract: Autonomous indoor navigation requires both semantic understanding and precise geometric control. We propose OptiSight, a hybrid framework that combin...
arXiv AI
Jul 21
COLIP-2: Olfaction-Vision-Language Embeddings
arXiv:2607. 17559v1 Announce Type: cross Abstract: The Contrastive Olfaction-Language-Image Pre-training 2 (COLIP-2) model is a multimodal embeddings space that places olfaction as a first-class citizen among vision and language.
Hugging Face Trending Papers
Sep 23
Generalizable Robotic Insertion with World Models
Robotic assembly in high-mixture settings requires adaptable systems that can handle diverse parts, yet current approaches typically rely on policies specialized to each insertion task. Although this...