Using Marker-less Motion Capture Systems for Walk Path Analysis in Paced Assembly Flow Lines

In recent years, automotive industry is facing a turbulent environment with an increasing demand for mass-customization and shortened product life-cycles. For manual assembly, this trend has led to a rising planning complexity, since growing numbers of product variants are hitting mixed-model assembly lines. In this context, it is crucial for production planning to be aware of the actual state of an assembly line in order to identify inconsistencies between the situation in company-owned learning factories and the shop floor, especially when considering non-value-adding tasks (e.g. walk paths). However, a feedback loop for walk paths linking the assembly line with the planning department is not established in practice. Consequently, discrepancies between planned and real processes remain largely unknown since they only become apparent through production disruptions. In order to provide production planning with an objective tool for walk path assessment, this work proposes a novel tracking approach, being able to reconstruct operators' motion within an assembly line. Based on a distributed depth camera array, a scalable and marker-less tracking system is presented that can be applied in productive environments. An in-depth evaluation underlines the performance of this novel approach and assesses the overall path accuracy. Finally, the proposed system is set up in an automotive final assembly line during operation. The gathered data is investigated regarding planning inconsistencies during operation. (C) 2016 The Authors. Published by Elsevier B.V.