Study of Ergonomic Risk as a Predictor of Operational Inefficiency: Evidence from Vision-Based Monitoring in Rice Mills

Abstract

Rice milling exposes workers to repetitive lifting, bending, twisting, and upper-limb exertion, yet existing ergonomic assessment methods largely treat these postures as isolated events rather than cumulative production stresses. This study examines ergonomic risk as a predictor of operational inefficiency within vision-based monitoring environments. It first reinterprets RULA, REBA, and OWAS in computational terms. RULA is framed as a piecewise discrete additive function with high local sensitivity. REBA is presented as a coupled biomechanical matrix with stronger whole-body interaction logic. OWAS is treated as a finite-state classifier with coarse quantization and lower sensitivity. Across all three, the core limitation is the absence of temporal dynamics. They do not represent motion adaptation, exposure accumulation, or threshold-based fatigue progression. The study then proposes the Temporal Ergodic Score, TES, a time-integrated multi-model risk functional that combines frame-wise RULA, REBA, and OWAS scores and attenuates risk under active motion through angular velocity. Within this formulation, ergonomic exposure becomes a cumulative signal linked to throughput decline rather than a static compliance score. The manuscript argues that throughput remains stable until TES reaches a critical threshold, after which speed, coordination, accuracy, and output deteriorate. This reframing places ergonomics within production analysis and supports earlier intervention through continuous vision-based monitoring, targeted task rotation, and better workstation control in rice milling systems.