MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202641025046 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on March 3, for 'a multilingual sign language interpretation system configured for real-time deployment on an iot-enabled robotic or assistive communication platform.'
Inventor(s) include Harapriya Kar; and Viswanathan P.
The application for the patent was published on March 13, under issue no. 11/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to a multilingual sign language interpretation system and method for real-time deployment on IoT-enabled robotic and assistive communication platforms are disclosed. The invention captures continuous video streams of signing gestures and extracts multimodal landmark features including hand, facial, and body pose keypoints. A transformer-based spatial encoder generates global attention representations, which are modeled using a dual- memory spatiotemporal long short-term memory (STLSTM) module to capture both short-term transitions and long-range gesture dependencies. A reinforcement adapter employing a dynamically decaying epsilon-greedy mechanism adaptively modulates attention and memory state updates to improve robustness and generalization. A multilingual embedding layer maps gestures from multiple sign languages into a shared language-invariant feature space, enabling unified recognition without separate retraining for each language. The system outputs real-time text and/or synthesized speech with low latency suitable for emergency and accessibility applications, while supporting cross-lingual transfer learning and end-to-end differentiability."
Disclaimer: Curated by HT Syndication.
