A new survey paper systematically reviews the application of large foundation models to hand-object interaction (HOI) tasks. It categorizes existing methods into six HOI tasks and proposes a taxonomy of eight foundation-model sub-priors, including geometric, semantic, and visual priors. The paper analyzes how these priors are integrated into HOI pipelines and discusses their use in robot learning, such as skill transfer and data generation. Additionally, it provides an overview of datasets, evaluation protocols, and future research directions, supported by a continuously updated repository. AI
IMPACT This survey provides a structured overview of how large foundation models can advance hand-object interaction research, potentially accelerating progress in robotics and embodied AI.
RANK_REASON The item is a survey paper published on arXiv detailing research on foundation models for hand-object interaction. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- Geometric priors
- Hugging Face
- Large Foundation Models
- robot learning
- Semantic priors
- Visual priors
- Xu
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
