Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 31, 2025

Human Activity Recognition (HAR) using wearable sensors has prompted substantial interest in recent years due to the availability and low cost of Inertial Measurement Units (IMUs). HAR IMUs can aid both ergonomic evaluation performed activities and, more recently, with development exoskeleton technologies, assist selection precisely tailored assisting strategies. However, there needs be research regarding identification diverse lifting styles, which requires appropriate datasets proper hyperparameters for employed classification algorithms. This paper offers insight into effect sensor placement, number sensors, time window, classifier complexity, IMU data types used styles. The analyzed classifiers are feedforward neural networks, 1-D convolutional recurrent standard architectures series but offer different capabilities computational complexity. is utmost importance when inference expected occur an embedded platform such as occupational exoskeleton. It shown that accurate style detection multiple sufficiently long windows, able leverage temporal nature since differences subtle from a kinematic point view significantly impact possibility injuries.

Language: Английский