Use of Kinematics to Minimize Construction Workers' Risk of Musculoskeletal Injury

Alwasel, Abdullatif . 2017
نوع عمل المنشور
مدينة النشر
واترلوو, كندا
رابط النشر على الانترنت
تاريخ المؤتمر
نوع الفرضية
جامعة واترلوو
ملخص المنشورات

Construction work requires more repetitive and highly physical effort than, for example, office work. Despite technological advancements in construction, the human factor is still an essential part of the industry. Hence, the need to maintain a healthy work environment is a shared interest between workers and industry. This thesis addresses the problem of cumulative injuries among construction workers, with emphasis on masons, and examines ways to improve safety and productivity simultaneously. Vision-based motion capture and sensor-based joint angle measurement techniques were tested against a state-of-the-art Optotrak system. Results show that the overall error in joint angle measurements was 10 deg for vision-based approaches compared to 3 deg for optical encoders. Moreover, a noninvasive fatigue detection method was developed by applying time-delay embedding and phase-space warping (PSW) techniques to a single joint angle, exerted force, and electromyography (EMG) data. Results indicate that the method can detect a slowly changing variable, fatigue in a limb, from a single kinematic signal, limb exerted force, or its EMG signals. Furthermore, twenty one masons distributed in four experience categories, ranging from novice to expert, took part in a study to evaluate safety and productivity in masonry work using inertial measurement units (IMUs). The study hypothesized that masons adopt safer and more productive work techniques with experience and that these techniques can be identified and used to train novice workers. Results indicate that journeymen appear to develop more productive and safer work techniques compared to other groups. On the other hand, the three-years experience group was found to sustain the highest joint compression forces and moments. Results also show that a clear distinction exists between expert and inexpert mason motion patterns. Support Vector Machine (SVM) classifiers were able to identify these differences with an accuracy of %92.04 in 13 seconds using a linear kernel. The thesis findings justify exploration of sensor fusion techniques to combine direct and indirect motion capture systems. The findings also suggest that PSW can be used in applications such as rehabilitation to access information about patient status hidden in the full-chain kinematics using a single kinematic signal. Finally, findings show the potential for training apprentices to excel in all three aspects: proficiency, productivity, and ergonomic safety by following the example of experts.