motiontracKIng

In recent years, high time and cost pressure in the commerce and producing industries have led to an increasing strategic importance of logistics. However, in most modern warehouse facilities, some repetitive operations are still performed manually. Because musculoskeletal disorders are disproportionately prevalent in warehouses, it is important to minimize physical stress on employees. For this purpose, movement and physical strain data can be collected by expensive motion-capturing-systems, which are usually unavailable for small and medium-sized enterprises (SMEs) due to high investment costs and great organisational effort. An alternate solution is the recording of movement data using more inexpensive and easy-to-use consumer electronics (e.g. Microsoft Kinect). The disadvantage of this optical recording is that overlaps can lead to incorrect representations of the movements and postures. The goal of the project is to develop a machine learning-based system that detects these error captures and approximates the lost motion data using process knowledge. For this purpose, data are collected not only in the laboratory, but also in cooperation with practical partners. For the first time, the project brings together insights of data collection in ergonomics with the help of consumer electronics and machine learning and thus enables SMEs to plan ergonomic workplaces in the warehouse area.