Understanding Worker Mobility within the Stay Locations using HMMs on Semantic Trajectories

Abstract

Construction is one of the most hazardous industries because it involves dynamic interactions between workers and machinery on sites. The recent technological developments in indoor positioning technologies provide a huge volume of spatio-temporal data for studying dynamic interactions of moving objects. The results from such studies can be used for enhancing safety management strategies on sites by recognizing the mobility related workers` behaviors. For understanding workers` mobility behaviors to improve site safety, a system is proposed based on semantic trajectories and the Hidden Markov Models (HMMs). Firstly, the system captures raw spatio-temporal trajectories of workers using an Indoor Positioning System (IPS) and preprocess them for determining the important stay locations where the workers are spending the majority of their time. Then, these processed trajectories are transformed into semantic trajectories to establish an understanding of the meanings behind workers` mobility behaviors in terms of the building environment. Lastly, HMMs along with the Viterbi algorithm are used for categorizing different workers` mobility behaviors within the identified stay locations. The proposed system is tested using an indoor building environment and the results show that it holds a potential to identify high-risk workers` behaviors to improve site safety.