TY - GEN
T1 - MEMS and AI for the recognition of human activities on IoT platforms
AU - Bibbo, Luigi
AU - Merenda, Massimo
AU - Carotenuto, Riccardo
AU - Romeo, Vincenzo Francesco
AU - Della Corte, Francesco
PY - 2023
Y1 - 2023
N2 - The increase in the elderly population has led to the need for new medical, social, and care services, resulting in a significant rise in health costs and the number of health workers involved. For example, IoT (Internet of Things) and wearable technologies can help contain healthcare spending and enable better living conditions of elderly. Moreover, nanotechnologies such as MEMS (micro-electromechanical system) that offer the advantage of small size, negligible need for power and motion acquisition are of considerable benefit. These technologies are able to detect and signal dangerous situations in order to ensure immediate action. In this article, we present an implementation of an IoT application on a latest-generation microcontroller. Kinematics and environmental data are transferred to a CNN (Convolutional Neural Network) to recognize the daily activities of the elderly in their homes or nursing homes. Finally, to determine the position of subjects, we associate the prototype with a positioning system on the ultrasonic platform. Finally, applying the Edge Machine Learning technique, we developed an application on the STM32L475VG microprocessor on which motion acquisition and activity recognition functions are activated.
AB - The increase in the elderly population has led to the need for new medical, social, and care services, resulting in a significant rise in health costs and the number of health workers involved. For example, IoT (Internet of Things) and wearable technologies can help contain healthcare spending and enable better living conditions of elderly. Moreover, nanotechnologies such as MEMS (micro-electromechanical system) that offer the advantage of small size, negligible need for power and motion acquisition are of considerable benefit. These technologies are able to detect and signal dangerous situations in order to ensure immediate action. In this article, we present an implementation of an IoT application on a latest-generation microcontroller. Kinematics and environmental data are transferred to a CNN (Convolutional Neural Network) to recognize the daily activities of the elderly in their homes or nursing homes. Finally, to determine the position of subjects, we associate the prototype with a positioning system on the ultrasonic platform. Finally, applying the Edge Machine Learning technique, we developed an application on the STM32L475VG microprocessor on which motion acquisition and activity recognition functions are activated.
KW - Ambient assisted living
KW - Convolutional neural network
KW - Edge machine learning
KW - Healthcare
KW - Human activity recognition
KW - Indoor positioning
KW - Internet of things
KW - Wireless sensor networks
U2 - 10.1007/978-3-031-24801-6_6
DO - 10.1007/978-3-031-24801-6_6
M3 - Conference Proceedings with Oral Presentation
VL - 1724
T3 - Communications in Computer and Information Science
SP - 73
EP - 89
BT - AII 2022: Applied Intelligence and Informatics
A2 - Mahmud, Mufti
A2 - Ieracitano, Cosimo
A2 - Kaiser, M. Shamim
A2 - Mammone, Nadia
A2 - Morabito, Francesco Carlo
PB - Springer
T2 - International Conference on Applied Intelligence and Informatics
Y2 - 1 September 2022 through 3 September 2022
ER -