Real-time safety assistance to improve operators' situation awareness in crane lifting operations

Crane safety;Situation awareness;Real-time;Sensor;Point cloud;Computer vision

Fang, Yihai ; Cho, Yong Civil and Environmental Engineering Castro-Lacouture, Daniel Tien, Iris Durso, Francis Anumba, Chimay Peña-Mora, Feniosky ; Cho, Yong

University:Georgia Institute of Technology

Department:Civil and Environmental Engineering

关键词: Crane safety;    Situation awareness;    Real-time;    Sensor;    Point cloud;    Computer vision;   

Others  :  https://smartech.gatech.edu/bitstream/1853/59143/1/FANG-DISSERTATION-2016.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Operating a crane is a sophisticated job that not only requires the operators to have extensive skills and experience, but more importantly a comprehensive situation awareness (SA) of the crane and its surroundings throughout the operation. The objective of this research is to develop and test a framework for enabling real-time safety assistance for mobile crane lifting operations, and to explore a quantitative method to validate the impact of such assistance system on lift performance and the operator’s SA. Based on the framework, a practical system architecture is created featuring three major components: real-time crane motion capturing, as-is site condition modeling and updating, and hazard analysis and real-time visualization. First, crane poses are reconstructed in real-time based on the critical motions of crane parts captured by a hybrid sensor system. Second, as-is lift site conditions are modeled based on point cloud data and updated using a point cloud-vision hybrid approach. Lastly, the risk of colliding the crane parts and lifted load into nearby obstructions is pro-actively analyzed and warnings are provided to the operator through a graphical user interface. A prototype assistance system implementing the system architecture is developed and deployed on a real mobile crane. A series of field tests in realistic lift scenarios reveal that this system was able to accurately capture and visualize crane motion with minimal delay. Based on crane motion and as-is site conditions, the system was able to identify the potential collision hazards and provide timely warnings to the operator to mitigate the risk. The effectiveness of the assistance system is quantitatively validated by the improvement in lift performance and SA, where lift performance is quantified by five key performance indexes (KPIs) and SA is measured by an online query-based technique. Major contributions of this research are the creation of a technical framework and a practical system architecture toward providing effective safety assistance to crane operators during lifting operations. The findings in this research have the potential to complement existing safety measures in crane lifting practices by providing an additional layer of technology to prevent human error-related crane accidents.