5542

Problem The project idea comes from an accident in which my mother fell off from a ladder and was unable to move or ask for help. This hit me hard and motivated me to create a device to monitor my parents and assist when needed. This is a huge need in the market as it happens often with elder people living on their own. As the aging population grows, so do the risks associated with falls, isolation, and missed health cues, especially for seniors living alone. Existing smart devices are often intrusive, cl

My project is an advanced embedded AI system wich will use the STM32N6 development board to assist elderly individuals living alone. It will perform real-time fall detection, activity monitoring, and emotional wellness analysis, all locally on the edge. Using its AI-native architecture, the system ensures privacy, responsiveness, and low-power operation without relying on cloud infrastructure. 

The project will use AI processing and ML inference acceleration capabilities of STM32N6 to monitor daily routines, detect anomalies, and assess emotional wellbeing in real time and without transmitting sensitive data.

The device would have:

Edge AI Fall Detection: Real-time motion analysis from IMU sensors using time-series anomaly detection models optimized for the STM32N6.
Activity Recognition: Efficient classification of posture and motion states (sitting, walking, lying down) using TinyML models.
Emotional Sentiment Awareness: Periodic keyword spotting and tone analysis from ambient speech using compact MFCC-based CNNs. 
Low Power: Runs entirely on STM32N6 with built-in power management and an optional BLE module for alert transmission