A Hybrid CNN-LSTM Model for Detecting Excessive Load Carrying from Workers' Body Movements

Abstract

Construction workers are often required to carry an excessive load due to schedule pressure or habituated behaviors, and frequent exposures to such excessive loads are the major source of work-related musculoskeletal disorders (WMSDs) in construction. In practice, a prevention of WMSDs is largely relying on the manual observation, which is challenging and burdensome on construction sites. In this context, this paper develops a method for detecting excessive load carrying though workers' bodily movement captured by an inertial sensor. This study proposed a hybrid convolutional neural network-long short-term memory (CNN-LSTM) to identify the excessive load carrying and carrying postures. Data collection was performed with total 14 male subjects by moving a construction material in various carrying conditions. The proposed model achieved 91% and 96% accuracies for excessive load carrying and carrying postures. The results of the study contribute on the prevention of excessive load related injuries and to promote the workers' safety and health in the workplace.