Newest papers

Learning Tactile-Aware Hand Representations from Human Hand-Object Interaction Data

Yeonguk Yu, Jaehong Kim, ByungOk Han · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.27995 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

El aprendizaje de la representación táctil de datos para la manipulación táctil de datos a menudo requiere una recopilación de datos robótica sustancial, limitando su escalabilidad a través de diferentes manos y sensores. Exploramos si los datos de la interacción de mano-objeto (HOI) a gran escala pueden proporcionar un precursor útil para el aprendizaje táctil robótico. Para este fin, proponemos HuTact, un marco que precursora las representaciones táctiles conscientes de la mano de los datos HOI humanos sin requerir señales táctiles reales durante el precursor. HuTact transfiere la estructura de interacción aprendida de los movimientos de la mano humana a las observaciones táctiles robóticas y puede ser adaptado con una pequeña cantidad de datos de interacción táctil no etiquetados. Valoramos Hu

Machine translation (Meta M2M100); the English original follows.

Abstract

Tactile representation learning for dexterous manipulation often requires substantial robot-side data collection, limiting its scalability across different hands and sensors. We explore whether large-scale human hand-object interaction (HOI) data can provide a useful prior for robotic tactile learning. To this end, we propose HuTact, a framework that pretrains tactile-aware hand representations from human HOI data without requiring real tactile signals during pretraining. HuTact transfers the interaction structure learned from human hand motions to robotic tactile observations and can be further adapted with a small amount of unlabeled robot interaction data. We evaluate HuTact on tactile perception and manipulation tasks, including grasp prediction, slip detection, object classification, and tactile-conditioned policy learning. Across these tasks, human HOI pretraining improves data efficiency and provides a strong initialization for downstream learning. Notably, with HOI pretraining followed by midtraining on unlabeled robot play data, HuTact achieves 88.26% Macro-F1 on grasp prediction using only 25\% of the labeled downstream data, comparable to training from scratch on the full labeled dataset (88.23%). These results suggest that human interaction data can serve as a scalable complementary source for robotic tactile representation learning.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: GRASP, Computer science, Artificial intelligence, Human–robot interaction, Robot, Representation (politics)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange