Kinect Validation of Ergonomics in Human Pick and Place Activities Through Lateral Automatic Posture Detection

In this paper we evaluate a system based on the Microsoft Kinect™ sensor, aimed at automatic detection of risk postures during human work activities. We first introduce pick and place task, where three different lateral standing subjects move light cardboard boxes from various levels bookcase to its top, then putting them back their original places. They repeat task over several cycles capture all natural movements in continuous way using Kinect, storing joint positions color images. Secondly, positions, our detects specific following definitions Rapid Upper Limb Assessment (RULA) method. Finally, compare posture detections by with baseline made panel five experts who used captured study find that have problems distinguish among some RULA cycle because narrow margin difficulty perceive if limb reached certain position; which is particularly true for cases wrist neck. This leads larger false positive rate lower general accuracy, detecting do not. After applying ±1° relaxation system, negligible perception, are able reach an accuracy 0.93 comparison baseline. Our results show suitability Kinect