Exploring Optimal Objective Function Weightings to Predict Lifting Postures Under Unfatigued and Fatigued States.

Objective: To explore whether the optimal objective function weightings change when using a digital human model (DHM) to predict origin and destination lifting postures under unfatigued and fatigued states.

Background: The ability to predict human postures can depend on state-based influences (e.g., fatigue). Altering objective function weightings within a predictive DHM could improve the ability to predict tasks specific lifting postures under unique fatigue states.

Method: A multi-objective optimization-based DHM was used to predict origin and destination lifting postures for ten anthropometrically scaled avatars by using different objective functions weighting combinations. Predicted and measured postures were compared to determine the root mean squared error. A response surface methodology was used to identify the optimal objective function weightings, which was found by generating the posture that minimized error between measured and predicted lifting postures. The resultant weightings were compared to determine if the optimal objective function weightings changed for different lifting postures or fatigue states.

Results: Discomfort and total joint torque weightings were affected by posture (origin/destination) and fatigue state (unfatigued/fatigued); however, post-hoc differences between fatigue states and lifting postures were not sufficiently large to be detected. Weighting the discomfort objective function alone tended to predict postures that generalized well to both postures and fatigue states.

Conclusion: Lift postures were optimal predicted using the minimization of discomfort objective function regardless of fatigue state.

Application: Weighting the discomfort objective can predict unfatigued postures, but more research is needed to understand the optimal objective function weightings to predict postures during a fatigued state.