Smart tool-based quality management system for on-site traceability

Abstract

Quality plays a significant role in the construction industry in that it not only affects consumers' satisfaction but also pertains to the lives and safety of people. To ensure the construction quality in the past decades, researchers proposed various approaches to monitor and manage the on-site activities, including wearable devices with activity recognition, surveillance cameras with computer vision, etc. However, the sensor-based methods were intrusive because the deployment added labor workloads, reduced labor productivity, and weakened the ability in handling daily construction activities; meanwhile the camera-based methods were seriously affected by the none-line-of-light effects, poor light illumination, and exposed environment. In addition, these approaches both contributed to the privacy issues at the construction sites, which resulted in a lack of independence and the depression of manpower. It was concluded that there was no efficient and effective way to record and recognize the on-site construction activities. A possible non-intrusive approach to recording the on-site events is suggested by considering the kinematic motions of construction tools used in construction tasks. As is well known, the way human beings make and use tools is perhaps what sets us apart more than anything else, employees in modern architecture, engineering and construction industry are always carrying out their behaviors with the assistance of valuable hand or power tools. This study therefore proposes a novel method that record and recognize the on-site construction processes by collecting the motion data of the equipped tools, which possibly facilitates the management of construction quality within the expectation of privacy to be legally permissible. A wireless tool tracking gadget using micro-electro-mechanical-system inertial measurement unit is presented. The prototype is capable of capturing the direct motion data, consisting of acceleration, angular velocity, and magnetic field, as well as the indirect kinematic data, including displacement, velocity, rotation angle, etc. A model for the on-site process reconstruction and activity recognition is then presented, which also enables one to obtain the construction activity indicators for the quality control and management, such as work time and orthogonality. A model for the traceability along the construction process is also presented, which is based on Bayesian Networks that has the capability of tracking forward and backward by probability delivery and belief backpropagation. These models were tested in steel rebar connection and concrete consolidation experiments, and led to satisfactory results, showing the effectiveness and efficiency of the proposed method that makes the onsite activity monitoring smarter and allows for traceability along the construction processes.