Ai-Powered Blazepose-Based Integrated Ergonomic and Musculoskeletal Risk Assessment Using Owas-Niosh Risk Fusion

Abstract

Manual material handling (MMH) activities, such as lifting, carrying, bending and repetitive object transfer are prevalent in industrial environments and one of the major causes of work-related MSDs. To minimize an employee's risk of injury and enhance workplace safety, it is vital to recognize unsafe postures early. This paper introduces an Artificial Intelligence based framework for ergonomic assessment with human pose estimation (HPE) using RGB video, applying the BlazePose system, combined with the Ovako Working Posture Analysis System (OWAS) and the Revised NIOSH Lifting Equation (RNIOSH). A proposed framework automatically detects the body landmarks, estimates the joint angles, classifies working postures and calculates lifting risk without the use of wearable sensors or marker-based motion capture systems. To get a complete picture, severity of posture, lifting demands and exposure to high-risk posture are combined into a single Ergonomic Risk Score (ERS). The framework was tested on four typical manual material handling operations: floor-to-waist lifting, twisted lifting, repetitive pick-and-place and load carrying. The results indicate that twisted lifting carries the highest ergonomic risk due to high levels of trunk flexion, high risk postures, and a higher lifting index, compared to load carrying. The proposed approach provides a practical, low-cost, non-invasive solution for automatic ergonomic risk assessment and can be used to continuous monitoring of the workplace, timely ergonomic interventions and enhance occupational safety in industrial environments.