A comprehensive methodology for vision-based progress and activity estimation of excavation processes for productivity assessment

The high level of complexity in modern construction projects causes a-priori project schedules to be highly sensitive to delays in the involved processes. At underground construction sites the earthwork processes are very vital, as most of the following tasks depend on it. This paper presents a novel method for tracking the progress of earthwork processes by combing two technologies based on computer vision: photogrammetry and video analysis. While the former is applied to determine the volume of the excavated soil in regular intervals, the latter is used to generate statistics regarding the construction activities, such as loading times and idle times. Combining these two data sources allows exact measurement of the productivity of the machinery and determining site-specific performance factors. Most importantly, reasons for low productivity - such as an insufficient number of trucks - can be identified easily. The paper presents in detail the vision-based techniques applied and the methods used for combining both data sources. The suitability of the approach has been proved by an extensive case study - a real-world excavation project in Munich, Germany.