Identification of distinct events in an assembly by automatically tracking body postures

Abstract

Identification of distinct events in an assembly process plays a vital role in assigning difficulty of assembly to specific events within the process. Literature addresses identification of events in assembly by analyzing videos, which is a manual process. We propose an approach to identify these events automatically using a method based on tracking of body postures and time. We use this method for an assembly exercise designed to introduce progressively more reach difficulty, and was performed in laboratory settings to track all limb data in time. The tracked data was used as an input to identify distinct events in an assembly. We use a technique we call slope method to smoothen the input data. Using this method helps in identifying distinct events involved in an assembly task. The result is validated by identification of events by observation of videos and correlation of these events with those identified automatically using the proposed method.