A new Kinect-enabled motion analysis approach for warehouse materials handling activities

Abstract

Manual materials handling (MMH) is an important logistics activity. It often requires human workers to repetitively stretch or bend while handling materials, exposing workers to the risk of developing musculoskeletal disorders (MSDs). A structured approach of motion data collection serves as a first essential step to evaluate physical ergonomics of workers. This paper evaluates the feasibility of Microsoft Azure Kinect for MMH motion capture in a warehouse context. We first compare Azure Kinect to Captive L7000, a professional inertial sensor-based motion capture system, and then propose two equations for the alignment of data captured by the Kinect and Captive systems. Our results show that the Captive system tends to overestimate angle sizes and that the captured values are larger than that of the Kinect system in static posture settings. For the elbow, our proposed approach is able to correct the artifacts and align the data captured by the two systems. In contrast, for the knee, our approach can correct artifacts only in case just a small share of the captured data is mismatched.