Aerospace engineering

Chip

Accelerometer

Operating system

Activity recognition

Wearable computer

Engineering

Embedded system

Artificial intelligence

Gyroscope

STM32

Computer science

Real-time computing

Telecommunications

The elderly population is experiencing high growth in most countries. In Malaysia, a majority of the households are dual-earner families, leaving their elderly parents at home. Without caretakers to look after them, aging parents may be susceptible to accidents. Therefore, an elderly action recognition system which can automatically identify their actions and whereabouts at home is needed. In this paper, the development and testing of a wearable device with motion detection and indoor positioning based on Random Forest Classifier is presented. The experiment was conducted in a home environment (MMU Digital Home Lab). The action recognition system utilizes a gyroscope and an accelerometer to detect different types of motion which are walking, sitting, standing, sitting-down, standing-up, falling, and sleeping. The action recognition system can measure these motions with a 97.6% accuracy. On the other hand, the indoor positioning uses Bluetooth Low Energy beacon and scanners to pinpoint five locations which are living room, bedroom, kitchen, office and toilet with a 97.25% accuracy. The results from the measured motions and indoor locations are pushed to an online Internet of Things (IoT) platform for action recognition, data logging and analytics.

Elderly Action Recognition System with Location and Motion Data