Human lifting simulation using a multi-objective optimization approach

Abstract

This paper presents a multiobjective optimization (MOO) approach to predicting dynamic lifting for a three-dimensional, highly redundant digital human model with 55 degrees of freedom. The optimization problem is formulated to optimize two objective functions simultaneously—dynamic effort and stability—subject to basic physical and kinematical constraints. The predictive dynamics approach is used to solve for the joint angles, torque profiles, and ground reaction forces. The weighted sum approach of MOO is used to aggregate the two objective functions, and the Pareto optimal set for the problem is generated by systematically varying the weighting parameters for the objective functions. Experimental data are used to validate the final simulation. Several examples are presented to demonstrate the effect of the weighting parameters for the two objective functions on the predicted box-lifting strategies. The results show that the proposed MOO approach improves the simulation results compared to the single objective optimization formulation. Also, the formulation is less sensitive to the weighting coefficient for the stability criterion.