Automated recognition of construction workers’ activities for productivity measurement using wearable insole pressure system

Abstract

Continuous monitoring and automated recognition of activities performed by construction workers can help improve productivity measurements. However, manual methods are time-consuming and prone to errors; as such, they usually provide unreliable and inaccurate analyses. Therefore, an automated method can expedite the process of data collection and provide accurate analyses of activity recognition and productivity measurements. In this paper, a novel methodology is introduced to automatically recognize workers’ activities for evaluating productivity measurement based on foot plantar pressure distribution data measured by a wearable insole pressure system. Four supervised machine learning classifiers (i.e., artificial neural network (ANN), decision tree (DT), K-nearest neighbor (KNN), and support vector machine (SVM)) were used for classification performance using a 0.32s window size. Cross-validation results showed that the SVM classifier (i.e., the best classifier) obtained a classification performance with an accuracy of more than 94% and sensitivity of each category of activities was above 95% using a sliding window size of 0.32s. The findings from this preliminary study have shown great potentials to use a wearable insole pressure system to collect foot plantar pressure distribution data for automated recognition of workers’ activities and extract activity durations for evaluating productivity.