The background: Despite ongoing efforts towards automation, warehouse operations still rely heavily on manual material handling. This presents a well-known challenge: repetitive movements and unergonomic postures increase the risk of musculoskeletal disorders, which affects both employees and operational performance.

Researchers at TU Darmstadt and Saarland University demonstrate how an optical motion capture system optimised by machine learning can significantly improve the accuracy of motion tracking – even in the presence of visual obstructions that normally limit such cost-effective solutions. The Microsoft Azure Kinect sensor, originally developed for a games console, was used for this purpose.

The core idea is simple: using machine learning to correct systematic tracking errors and make cost-effective systems for ergonomic risk assessment suitable for practical use.

Our research shows that this approach:

• can significantly improve the tracking accuracy of key joint movements

• enables more reliable detection of hazardous postures

• offers a cost-effective alternative to expensive motion capture systems

This opens up new possibilities for affordable, scalable, data-driven ergonomics in warehouses and logistics operations.

The article was published in ‘Expert Systems with Applications’ and is openly accessible:https://doi.org/10.1016/j.eswa.2026.132917