A human-centric system combining smartwatch and LiDAR data to assess the risk of musculoskeletal disorders and improve ergonomics of Industry 5.0 manufacturing workers

Francesco Pistolesi,Michele Baldassini,Beatrice Lazzerini

DOI: https://doi.org/10.1016/j.compind.2023.104042

IF: 10

2024-02-01

Computers in Industry

Abstract:More than one in four workers reportedly suffer from back pain worldwide, leading to 264 million work days lost yearly. In the U.S. alone, it causes $50 billion in healthcare costs every year, up to $100 billion if including decreased productivity and lost wages. The upcoming Industry 5.0 revolution will introduce human-centric manufacturing systems where workers’ well-being comes first while safeguarding their rights: privacy, autonomy, and human dignity. This paper presents a privacy-preserving system based on artificial intelligence that tracks the posture of assembly/disassembly line workers while performing typical standardized tasks on repeat: connecting and separating parts; screwing and unscrewing using an electric screwdriver; tin soldering and unsoldering. The proposed solution assesses the upper-body (dominant arm, dominant shoulder, and trunk) and lower-body (legs) postures according to the ISO 11226 European standard, based on inertial data recorded by a smartwatch and using Laser imaging Detection and Ranging (LiDAR), respectively. These techniques preserve privacy as they collect data that cannot reveal the worker’s identity or sensitive information. Experiments showed that the system recognized a worker’s posture with a mean accuracy close to 98%. The system could help detect bad posture habits and reduce the chances of musculoskeletal disorders while preserving the workers’ privacy, in compliance with the upcoming Industry 5.0 paradigm.

computer science, interdisciplinary applications