Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
10224476 | Smart Health | 2018 | 22 Pages |
Abstract
Risk factors, such as overexertion, awkward postures, excessive repetition, and the combination of these factors are main causes of work-related musculoskeletal disorders (WMSDs). In this paper, we proposed an automatic WMSDs risk factors identification and visualization method based on Wearable and Connected Gait Analytics System (WCGAS) and Kinect skeleton models. WCGAS was capable of recording plantar pressure from which postures, force exertions, and repetitions could be recognized with algorithms such as sequential minimal optimization (SMO) algorithm and long short term memory (LSTM) network. Kinect skeleton models were used to make the WMSDs risk factors visualized. Experiments with quasi-static and sequential postures were designed to evaluate the recognition performance of work-related motion type (i.e. “lifting”, “carrying”, “bending”, “pulling”, and “pushing”). A load variable (with/without 10â¯kg load) was introduced for evaluating the performance of force exertions recognition. 5 repetitions of each motion were used for evaluating the performance of repetitions recognition. Results showed that quasi-static postures could be recognized with 100% accuracy and the accuracy for sequential motions recognition were 74%, 79%, 92%, 99% and 99% for “bending”, “carrying”, “lifting”, “pulling” and “pushing”, respectively. Force exertions were recognized with 100% accuracy. For repetitions recognition, except the accuracy in the “bending” motion was 80%, the repetitions of other motions could be recognized correctly. Kinect skeleton model showed its ability of making the WMSDs risk factors vivid which would contribute to the accuracy of WMSDs risks evaluation. These results indicated that it is possible to use WCGAS and Kinect skeleton models for WMSDs risk factors identification and visualization applications.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Networks and Communications
Authors
Diliang Chen, Yi Cai, Jiawei Cui, Jia Chen, Haotian Jiang, Ming-Chun Huang,