Smart Worker Helmet for Human Activities Monitoring

Abstract

The diploma thesis concerns with the design and realization of a partially self-powered helmet equipped with photovoltaic elements and a battery management for harvesting purposes, including an inertial sensor for activity evaluation and a transmitter to a remote station. The experimental efficiency comparison of electronic solutions for energy harvesting allowed to choose the BQ25504 as a suitable solution. Influence of inertial sensors (MPU9250 and BNO055) on a power consumption was experimentally tested based on the measurment results the BNO055 was selected as a suitable inertial sensor for activity evaluation. A graphical user interface in the LabView program was developed for the management and monitoring of all data. The Matlab software was used for development of movement classification algorithm using the Värri method of connected windows and the proposed rule-based classifier. The accuracy of segmentation algorithm verified on 10 probands was evidently influenced by differences in the probands movements. Testing of the influence of the sensor position on the helmet proofed that the classification algorithm achieved best results for the temporal position on the helmet. The observed shortcomings affecting the accuracy of the activity classification are discussed.