MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641095438 A) filed by Vellore Institute Of Technology on August 06, 2026, for A System And Method For Deterministic Embedded Neural Inference And Hardware-Gated Physical Actuation.
Inventors include Sivakumar Rajagopal; and Mrunmayee Mohanty.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present disclosure proposes a system (100) and method for deterministic embedded neural inference and hardware-gated physical actuation through direct-memory- access-based physiological signal acquisition, quantized neural-network execution, hardware cycle-count monitoring, and refractory-timer-based temporal validation of an actuation request. The system (100) comprises a physiological signal acquisition interface (102), an embedded controller 104 comprises at least one processor (106), a memory (108), a direct memory access controller (110), a hardware cycle counter (118), a refractory timer (122), and a physical output interface (126), a signal preprocessing module (114), an 8-bit integer quantized neural-network execution engine (116), a decision logic block (120), a hardware safety interlock (124), a telemetry interface (128), and a telemetry feedback logger (130). The system (100) records classification information and execution-duration information for verification of repeated neural-network execution cycles.
Disclaimer: Curated by HT Syndication.
