MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641075845 A) filed by Vellore Institute Of Technology on June 19, 2026, for A Physics-Guided Edge-Ai System For Landslide Pre-Failure Detection Using Multi-Modal Sensing.
Inventors include Mythili Boopathi; and Binusha S.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: A physics-guided, uncertainty-aware edge artificial intelligence system (100) for detecting pre-failure landslide conditions is disclosed. The system comprises an acoustic emission sensor (102) configured to acquire subsurface acoustic emission signals, a signal conditioning unit (104), an edge processing and sensing unit (106), and a domain-adaptive learning architecture comprising a feature extractor (302), a label predictor (304), a gradient reversal layer (306), and a domain classifier (308). Conditioned acoustic signals are transformed into spectrogram representations and analysed to generate soil-condition predictions. A physics-guided validation framework (400) comprising a physics-guided validation module (404) evaluates prediction outputs against geotechnical constraints, while an uncertainty estimation module (406) determines prediction confidence. A final decision generation module (408) consolidates prediction outputs, validation outputs, and uncertainty measures. A LoRa transceiver module (108) communicates generated decision information. The system further comprises a power management module (110) for autonomous operation on edge platforms.
Disclaimer: Curated by HT Syndication.
