Experimentally Based Modeling of Whole Body Movement During Contact

In this paper, we investigated how the assumption of fixed segment lengths in two dimensional whole body dynamic models limits accuracy in reproducing experimental reaction forces and observed kinematics. A six segment whole body dynamic model of the musculoskeletal system was developed to simulate the measured forces and kinematic data during the contact phase of two somersaulting tasks performed by two Olympic level divers. Initial conditions and foot-surface model parameters were refined using optimization to ensure that change in whole body center of mass (CM) linear and angular momenta satisfied the impulse/momentum relationship for both dives and divers. Simulation results indicate that the assumption of fixed segment lengths increases error in prediction of the CM trajectory in the sagittal plane. Sensitivity analysis shows that a foot/surface model high in stiffness is more accurate in reproducing observed foot metatarsal displacement but is also more sensitive to the velocity of the metatarsals at contact than a less stiff foot-surface model. As a result, the assumption of a fixed foot segment length also affects the process of optimizing the initial conditions and foot-surface parameters. These findings suggest that a 2D representation of segment motion using fixed segment lengths is limited accuracy because the fixed length representation of segment kinematics does not reflect out of plane motion. Tracking the effect of error introduced by input kinematics on model performance is essential in the process of validating a 2D model of human movement during contact with the environment.