Productivity Analysis of Construction Worker Activities Using Smartphone Sensors

Low productivity has been a longstanding issue in the construction industry. A comprehensive remedy to this problem not only does require the adoption of improved construction methods and resource utilization, but also calls for a robust, effective, and standard protocol to measure productivity especially in complex and dynamic construction jobsites. Research has shown that activity-level productivity analysis can serve as one of the most reliable decision support tools in construction operations. The backbone of such analysis is collecting and mining process-level data from construction entities on a jobsite. While manual data collection methods are prone to inaccuracy and inconsistency, and are almost always time consuming, automated data collection procedures have shown a promising prospect in construction industry research. This paper discusses a novel methodology for automated activity-level productivity measurement within the construction engineering and management context. In this methodology, pervasive smartphones provide the basis for data collection from construction workers. The collected data are then used as the input of machine learning classification algorithms to detect and differentiate between several classes of human activities. While recognizing the idle/busy state of the workers provides key information required for productivity analysis, the presented research further advances this information by extracting even more particular and accurate knowledge about different activities carried out by construction workers. For validation, results of various experiments including multiple construction activities are reported in this paper.