Date of this Version
UCARE Poster session, University of Nebraska-Lincoln Research Fair, April 2016, Lincoln, NE.
The purpose of this research is to develop a smartphone based system to continually analyze construction equipment activity (e.g. a skid loader moving forward, side-ways, or raising its bucket) using a variety of different sensors and give feedback to the equipment operator or the supervisor. Such a system could detect inefficiencies in construction operations and provide valuable information to project managers.
The results have demonstrated that DTW is effective at identifying typical rotation patterns. It has been less effective for slow rotations over long duration or very fast rotations with short durations. The accuracy of DTW is improved when the data is accurately segmented. Use of standard deviation to segment the data is very promising. Current work involves determining the most effective window to calculate standard deviation on, and an appropriate threshold value to use for segmentation.