Analyzing context and productivity of tunnel earthmoving processes using imaging and simulation

This study presents an integrated method of construction-process simulation and vision-based context reasoning for productivity analysis of an earthmoving process in a tunnel. Convolutional networks are used to detect construction equipment in the tunnel CCTV video and the context of the earthmoving process is inferred by the context reasoning process. The construction equipment detection model exhibited enhanced performance, with a mean average precision of 99.09%, and the error rate of the estimated context information was only 1.6% of the actual earthmoving context measured by a human. The estimated context information was used as an input for the WebCYCLONE simulation to generate a productivity and cost analysis report. Sensitivity analysis regarding construction equipment provided a new equipment allocation plan that could reduce the cost of the current earthmoving process by 12.25%.