SignVLA: A Gloss-Free Vision-Language-Action Framework for Real-Time Sign Language-Guided Robotic Manipulation

Researchers have developed SignVLA, the first sign language-driven Vision-Language-Action framework for human-robot interaction that directly translates sign gestures into robotic commands without requiring intermediate gloss annotations. The system currently focuses on real-time alphabet-level finger-spelling for robotic control and is designed to support future expansion to word and sentence-level understanding.

• →SignVLA is the first gloss-free sign language framework that directly maps visual gestures to robotic actions without intermediate annotations.
• →The system focuses on alphabet-level finger-spelling for improved reliability and safety in robotic control applications.
• →The gloss-free approach reduces annotation costs and avoids information loss compared to conventional sign language recognition methods.
• →The framework includes geometric normalization, temporal smoothing, and lexical refinement for stable gesture-to-command translation.
• →The system is designed to integrate transformer-based models for future word-level and sentence-level sign language understanding.