An Effective Scheme of a Depth Sensor Set Up for a Real-Time Ergonomics Assessment by the Gesture Confidence Level

Abstract

Ergonomics assessment of human body movement demands a comprehensive and systematic data collection. The easy-to-use self-report (e.g. questionnaires, checklist, interview) and the observational technique (e.g. pose rating) are the commonly practiced techniques. However, these methods suffer from a high bias across different respondents and observers. Recently, the direct measurement technique by utilizing a depth sensor equipped with a modelling software is the alternative tool to facilitate a real-time digital human modelling. It gathers the 3-D human motion data with real-time ergonomics analysis and intervention features. This study aims to obtain the effective sensor setup by examining its parameters (object-to-sensor distance, horizontal field of view (FOV), and light intensity) to reach the acceptable gesture confidence level using a Kinect SDK V2.0. The standing position with a hand overhead was selected as the investigated gestures. The result showed that distance and horizontal FOV were statistically significant parameters. Thus, it proposes to place the sensor within 2 or 3 m away from the investigated object and to limit the horizontal FOV to 0 or 10°. Eventually, this proposal could be set as the reference in setting up a direct measurement studio for acquiring the human body movement data.