Development of a system for collecting and analyzing labor environment and productivity data from construction sites using machine learning

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K04218
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 22010:Civil engineering material, execution and construction management-related
• Research Institution: Tokyo City University
• Principal Investigator: GOSO TAKASHI 東京都市大学, 建築都市デザイン学部, 准教授 (60412441)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 建設 / 労働環境 / 生産性 / 歩掛 / 動作分類 / 機械学習

Outline of Final Research Achievements

In this study, an automatic analysis system for productivity classification was constructed. Acceleration information was handled as images, and five simple actions, such as "working" and "walking", were classified first, and then classified into nine productivity categories. Although not all of the simple movements could be verified, we were able to discriminate them at a high level, but it was also found that some measures to control over-learning were required. On the other hand, the classification of the productivity motion categories was not highly accurate, because the tasks that the workers engaged in were complex and compound motions. However, the results also indicated the direction of improvement in accuracy by increasing the amount of teacher data and tuning the system.

Free Research Field

建設マネジメント，行政経営，公共調達

Academic Significance and Societal Importance of the Research Achievements

本研究で構築するシステムは，労働環境と生産性に関するデータを一定の精度で継続的に取得し，生産現場における問題点を定性的，定量的に認識し，関係者間で共有して向上策を講じるという当たり前のマネジメントを実現するための基盤である．我が国の建設産業ではこういった取り組みは全くなされておらず，建設現場の支援を得て取り組まれた本研究の成果は現場の生産性向上活動に寄与できるものである．