Trajectory-based worker task productivity monitoring

Abstract

Over the past decades labour productivity in construction has been declining. The prevalent approach to estimating labour productivity is through an analysis of the trajectories of the construction entities. This analysis typically exploits four types of trajectory data: a) walking path trajectories, b) dense trajectories (posture), c) physiological rates such as heart rate (beats/minute) and respiratory rate (breaths/minute), and d) sound signals. The output of this analysis is the number of work cycles performed by construction workers. The total duration of these cycles is equal to the labour input of a task. However, all such methods do not meet the requirements for proactive monitoring of labour productivity in an accurate, non-obtrusive, time and cost efficient way for multiple workers. This paper proposes a method to address this shortcoming. It features a promising accuracy in terms of calculating the labour input.