Title: Smart Assistive Gesture Glove and Caretaker Alert System

1. Problem statement:

Speech-impaired, partially paralyzed, or bedridden patients often struggle to communicate basic necessities (food, water, washroom, physical discomfort) or call for immediate emergency help without a caregiver continuously in the room. Standard push-button call systems lack expressive nuance, while cloud-reliant smart devices fail completely whenever local Wi-Fi or internet connectivity drops, posing a critical safety hazard during medical emergencies.

2. Solution and Technical Implementation:

​Gesture-to-Intent Recognition: The patient wears a wearable glove equipped with four flex sensors. Using trimmed-mean signal sampling and Exponential Moving Average (EMA) filtering, the system filters out muscle tremors and maps individual finger flexes to primary needs (Food, Washroom, Water, Discomfort), with a chorded fist gesture reserved for an instant Emergency SOS.

​Zero-Downtime Local Transmission: Communication between the glove and the caregiver base station uses peer-to-peer ESP-NOW radio protocol. This operates completely independent of routers or internet access, featuring an auto-channel hopping algorithm on the glove to maintain the radio link even if radio channels shift.

​Multi-Modal Feedback & Cloud Alerts: The receiver base station converts incoming packets into immediate spoken voice alerts via a DFPlayer Mini MP3 module, visually displays the request on an SSD1306 OLED, and sounds an active buzzer during SOS events.

​Non-Blocking Remote Gateway: An asynchronous FreeRTOS background task pinned to Core 0 relays every command and glove-disconnection warning directly to the caregiver’s smartphone via a Telegram Bot, ensuring family members stay informed even when away from the base station without delaying real-time local alarms.