Wearable sensor-based activity recognition for data-driven simulation of construction workers' activities

Wearable technologies are becoming the main interface between human and surrounding environment for a variety of context-aware and autonomous applications. Ubiquitous, small-size, and low-cost smartphones carried by everyone nowadays are equipped with a host of embedded sensors that provide groundbreaking opportunities to collect and use multimodal data in data-driven decision support systems. Simulation models are one of the most widely used decision support tools in project management that can highly benefit from the integration of contextual knowledge with the model design. In this paper, a discrete event simulation (DES) model of construction operations involving human activities is designed, enriched with wearable sensor data using smartphones, and validated. The model parameters are defined using 1) a data-driven activity recognition and 2) a static engineering estimation method for comparison. Results show that the output of the data-driven simulation model is in a closer agreement with the values observed in the real system.