MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641110989 A) filed by Pragati Engineering College on September 16, 2026, for Ai Hazardous Waste Identifier For Informal Waste Pickers.
Inventors include Dr. A. Radha Krishna; Mrs. V Anantha Lakshmi; Mr. M Ravi Kumar; and Mr. M Brahmaraju.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: AI HAZARDOUS WASTE IDENTIFIER FOR INFORMAL WASTE PICKERS The present invention discloses a portable multimodal edge-artificial-intelligence apparatus for identifying hazardous waste and providing immediate safety guidance, particularly for manual and informal waste handling environments. The apparatus comprises a sensing layer having an imaging module, a pH sensing assembly, a gas sensing array, a non-contact temperature sensor, a multispectral colour sensor and a machine-readable code scanner. An edge AI core, preferably implemented using a Raspberry Pi Zero 2W, executes a visual waste classifier, a chemical sensor-fusion classifier and a fusion arbitrator. In one embodiment, the visual classifier is a MobileNetV2 TensorFlow Lite model for fifteen waste classes, while the sensor classifier is a support-vector machine operating on pH, gas, temperature and spectral features. The fusion arbitrator applies confidence-weighted and hazard-specific precedence rules. Where sufficiently reliable physical or chemical evidence conflicts with a visual prediction, the sensor-derived hazard state may override the visual class, thereby reducing reliance on appearance alone. A low-confidence result may be designated unknown rather than being treated as safe. The final decision is communicated through a display, hazard symbol, local-language voice guidance, colour indicator and audible alarm. An optional local database supports offline barcode or QR-code hazard lookup, and a GPS and communication subsystem may log hazardous-waste events for later mapping or reporting. The apparatus is preferably powered by a rechargeable 5000 mAh battery with USB-C charging and optional solar top-up. Alternative embodiments use a higher-performance edge processor, an integrated single housing or a modular sensor board. The invention thereby provides an offline-first, field-portable and safety-oriented architecture in which direct chemical and physical sensing can correct or supersede uncertain or visually deceptive waste classification.
Disclaimer: Curated by HT Syndication.
