A computer vision–based automated solution for ergonomic risk assessment using the RULA method

Work-related musculoskeletal disorders (WMSDs) remain a major occupational health concern, requiring reliable and objective postural assessment to support prevention strategies. Although observational methods such as Rapid Upper Limb Assessment (RULA) are widely used, their manual application is subjective, time-consuming, and difficult to scale. This study develops a computer vision–based automated ergonomic risk assessment solution capable of identifying and classifying upper-body postures using RULA. Guided by the Design Science Research Methodology, the artifact was implemented in Python using OpenCV and MediaPipe Pose to detect body landmarks, calculate joint angles, and automatically assign RULA scores for the upper arm, lower arm, neck, and trunk. The solution was demonstrated and evaluated in educational and office-like environments through the analysis of 3316 video frames from standing and seated tasks. Results showed that, under the evaluated conditions, the system successfully detected postural deviations and generated time-resolved RULA scores that captured variations in trunk flexion, neck inclination, and upper-limb elevation. Agreement analysis using Fleiss’ kappa demonstrated perfect concordance and predominantly moderate-to-fair agreement across demonstrations, indicating consistency with expert assessments while capturing variability in challenging postural conditions. The solution also generates spreadsheets and annotated videos, enabling the analysis of postural behaviour across tasks, reducing subjectivity, and maintaining alignment with established RULA criteria. By integrating ergonomic principles with computational automation, the proposed solution offers a replicable, low-cost, and non-intrusive approach to ergonomic risk assessment, supporting objective postural assessment and providing information that may assist ergonomic decision-making.