Automated Construction Worker Performance and Tool-time Measuring Model Using RGB Depth Camera and Audio Microphone Array System

Abstract

Construction productivity improvement activities in the industry has seen many developments in terms of new tools, techniques and processes been introduced through research and practices. However, on site studies and literature surveys confirm that a huge potential still exists in the industry for further improvements. The thesis discusses a novel method for developing a sustainable, reliable, integrated, automated and systematic mechanism to extract construction worker tool-time and performance information that assists project managers and planners, in developing strategies for improving labour productivity, labour allocation and developing administrative schemes related to labour performance by using audio and video surveillance techniques addressing the potential drawbacks from the manual observation on construction site. The research proposes a low cost and composite range sensing device (Microsoft Kinect) consisting of RGB camera, depth camera, and microphone array to extract multiple sensing modalities from indoor construction. A user friendly and comprehensive framework is developed to consolidate location aware information of workers, other site personnel and construction activities in order to generate productivity related data. Additionally, this provides information on worker behavioral analysis (i.e. supervisory effect on performance) and breakdown of non-tool time activities which can be used for better labor allocation strategies. The proposed set of algorithms (i.e. worker recognition, construction activity detection, direction of arrival detection and tool-time analysis) has been validated in a real work environment and experimental results witnessed over 90% precision. In brief, validated results reflect the potential for automated assessment of worker tool time in indoor construction environment and each of these implications make an important contribution to the body of knowledge in construction automation.