Xincong Yang

Dynamics of Project-Based Collaborative Networks for BIM Implementation: Analysis Based on Stochastic Actor-Oriented Models

AbstractIn the project-based construction industry, organizations are coupled with each other largely through project-specific collaborative relationships, and the industry-level networks of these ...

Location-based measurement and visualization for interdependence network on construction sites

Abstract Appropriately assigning workers to tasks is vitally important in project management. To do this, project managers need to objectively and effectively measure and visualize the spatiotemporal orders of real construction process as well as coordination structure of the workforce. However, currently there is no method/tool available to project managers to represent spatiotemporal orders of construction processes. To address this issue, this paper presents a novel approach to measuring the real spatiotemporal order of onsite tasks as well as the task interdependence by an interdependence network. This approach extracts the distance of workspace distributions as a key interdependence indicator from historical location tracks across different construction stages according to the area-restricted nature of construction activities. It then integrates generated interdependence into a network over time, to imply the cooperation patterns in stages and a task delivery across stages with a holistic view. To validate the approach, location data were collected from 31 workers working in a high-rise housing construction project for one week to construct the interdependence network of this project, which was used to quantitatively evaluate the performance of construction schedule, assignments and cooperation. Results show that the interdependence network is able to provide insightful information on how workers perform individual tasks onsite and it is also an effective tool to identify and display the interactions among site workers.

Using Switching State-Space Model to Identify Work States Based on Movement Data

A key question in the construction industry is the productivity issue. With a variety of activities on sites from pip-installation to concrete placement, how to assess the working/non-working states in order to promote time allocation and productivity next time? This paper proposes to utilize switching state-space model integrated with first-difference random walk to recognize working and non-working behavioral modes through turning angles, step length, etc., containing four main steps: eliminate errors by forward/backward-averaging filter, establish the State-Space Model of a couple states with initial prior distributions that is called Switching State-Space Model (SSSM) enabling bidirectional transitions between states, estimate posterior distributions of parameters by Bayesian inference using Gibbs sampling, and analyze the objective work rate, time allocation, and productivity. A pilot study comprised of 24 workers from a housing project is finally conducted to test the proposed model and process.

Hierarchical Bayesian Model of Worker Response to Proximity Warnings of Construction Safety Hazards: Toward Constant Review of Safety Risk Control Measures

AbstractQuickly changing and complicated workplace conditions, which are typical of construction projects, have always been contributing to the poor safety record of the construction industry. Howe...

Estimating Construction Workers' Physical Workload by Fusing Computer Vision and Smart Insole Technologies

Construction workers are commonly subjected to ergonomic risks due to awkward postures and/or excessive manual material handling. Accurate ergonomic assessment will facilitate ergonomic risk identification and the subsequent mitigation. Traditional assessment methods such as visual observation and on-body sensors rely on subjective judgement and are intrusive in nature. To cope up with the limitations of the existing technologies, a computer vision and smart insole-based joint-level ergonomic workload calculation methodology is proposed for construction workers. Accordingly, this method could provide an objective and detailed ergonomic assessment for various construction tasks. Firstly, construction workers’ skeleton data is extracted using a smartphone camera with an advanced deep learning algorithm. Secondly, smart insoles are used to quantify the plantar pressures while the worker performs a construction activity. Finally, the gathered data is fed to an inverse dynamic model in order to calculate the joint torques and workloads. The aforementioned approach was tested with experiments comprising simulations of material handling, plastering and rebar. The results reveal that the developed methodology has the potential to provide detailed and accurate ergonomic assessment. Overall, this research contributes to the knowledge of occupational safety and health in construction management by providing a novel approach to assess the risk factors of work-related musculoskeletal