Automated Postural Angle Classification Using Microsoft Kinect For Ergonomics Assessment
Musculoskeletal injury is a common cause in manual material handling activities, where workers are exposed to repetitive picking and placing of materials, that therefore may lead to dangerous injuries if incorrect postures are made. It is the duty of factories to take care of the health conditions o...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://eprints.utem.edu.my/id/eprint/24681/1/Automated%20Postural%20Angle%20Classification%20Using%20Microsoft%20Kinect%20For%20Ergonomics%20Assessment.pdf http://eprints.utem.edu.my/id/eprint/24681/2/Automated%20Postural%20Angle%20Classification%20Using%20Microsoft%20Kinect%20For%20Ergonomics%20Assessment.pdf http://eprints.utem.edu.my/id/eprint/24681/ https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=117641 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Musculoskeletal injury is a common cause in manual material handling activities, where workers are exposed to repetitive picking and placing of materials, that therefore may lead to dangerous injuries if incorrect postures are made. It is the duty of factories to take care of the health conditions of their employees, and ensure the workplace is ergonomically designed. However, it is a difficult task to assess the work postures in a large number of employees all the time due to cost, lack of equipment, and lack of experience. The aim of this study is to formulate an ergonomic model to identify and classify body part motion angle ranges for upper limb postural analysis, to develop an automated real-time upper limb postural angle and classification system, and to evaluate the developed postural angle classification system using 30 participants in a lab setting and five ergonomic experts opinions. The chosen experts are individuals with experiences in ergonomics field working as academic researchers, consultancy agents, and industry management positions in Malaysia. Formulating the postural classification model applied the concepts of traffic light to categorise the work postures, where upper limb postures were classified into three classifications with mathematical models to count the number and percentage of each classification occurrence for each posture. The postural classification model was then integrated with a developed C# based software and a Microsoft Kinect sensor using heuristic approaches to do an automated real-time upper limb postural angle classification system. The developed postural classification was validated for 12 static postures, and 4 dynamic postures among 30 participants in a lab setting using Jamar goniometer (Sammons Preston Roylan, USA), a computerised protractor tool in ErgoFellow v3.0, and the statistical analysis used the root mean square error (RMSE). The evaluation was further explored by taking the ergonomic experts’ opinions through semi-structured interviews to note the needful, usefulness, applicability, effectiveness, and the details provided for the workplace. The results of validation revealed that the static postures was 7.52 RMSE, dynamic postures was 21.93 RMSE, and combined static and dynamic results was 14.48 RMSE. The study shows better mean RMSE results than Plantard et al. (2017) study by 15.6% in static phase analysis, but larger mean RMSE in dynamic analysis which might be due to the method of capturing the reference angles. The study concluded that despite the acceptable RMSE results presented by the developed system, the software architecture and detection techniques require further improvement and development for better angle measurement accuracy with added parameters for ergonomics assessment. |
|---|