Data-Driven Ergonomic Risk Assessment of Complex Hand-intensive Manufacturing Processes

Abstract

Hand-intensive manufacturing processes, such as composite layup and textile draping, require significant human dexterity to accommodate task complexity. These strenuous hand motions often lead to musculoskeletal disorders and rehabilitation surgeries. We develop a data-driven ergonomic risk assessment system with a special focus on hand and finger activity to better identify and address ergonomic issues related to hand-intensive manufacturing processes. The system comprises a multi-modal sensor testbed to collect and synchronize operator upper body pose, hand pose and applied forces; a Biometric Assessment of Complete Hand (BACH) formulation to measure high-fidelity hand and finger risks; and industry-standard risk scores associated with upper body posture, RULA, and hand activity, HAL. Our findings demonstrate that BACH captures injurious activity with a higher granularity in comparison to the existing metrics. Machine learning models are also used to automate RULA and HAL scoring, and generalize well to unseen participants. Our assessment system, therefore, provides ergonomic interpretability of the manufacturing processes studied, and could be used to mitigate risks through minor workplace optimization and posture corrections.

Figures (8)

• Figure 1: Data-driven Ergonomic Risk Assessment Pipeline. Multi-modal sensor data are used to train machine learning models and compute a biometric hand score to provide a quantitative assessment of ergonomic risks to the operators performing hand-intensive manufacturing tasks.
• Figure 2: Digital scans of the two tools used for data collection. 3D scans of the Stringer tool (left) and Convex Mold tool (right) show the difference in geometry between the many concave rows of the Stringer and the convex curvature with varying radii along the length of the Convex Mold Tool.
• Figure 3: Data Collection Testbed and Illustration of Data Synchronization. The sensor setup (left) collects data as participants perform composite hand layup, with the Leap motion sensor attached on the helmet, the goniometers attached to both the wrists, the TactileGlove worn on both the hands and two webcams (out of image) capturing stereo images. Sensors with varying frame rates are aligned and synchronized to the frame timing of the webcam (right). The smaller dots represent intrinsic sensing rates and the larger dots represent the interpolated and webcam-aligned data points.
• Figure 4: Illustration of hand poses and the relationship between wrist flexion angle $\theta$ and moment. Left: Illustration of the various hand pose and corresponding risk levels. Right: Modeled maximum isometric moments generated by wrist flexors and extensors versus wrist flexion angle from holzbaur2005model with curve fitting (scipy.optimize.curve_fit function) when applying force along the positive flexion direction. The curve can be divided into two distinct segments: the left section, representing an approximation of the linear relationship between flexion and maximum moment, and the right section, which approximates the quadratic relationship. Accordingly, a linear model and a second-degree polynomial have been separately employed for these two segments.
• Figure 5: Wrist angle and torque distributions for all the participants during hand layup tests. The wrist torque patterns emphasize right-handed dominance for the participants. The variations in the wrist angles suggest different roles for the left and right hands during hand layup.