Validation of a three-dimensional baseball pitching model

A simulation and optimization procedure was constructed to investigate the relationships between optimal movement and muscular strength for baseball pitching. Four segments (torso, upper arms, lower arms, hands) and six torque generators (shoulders, elbows, wrists) are modeled. The torque generators have torque-angle and torque-angular velocity characteristics of Hill-type muscle function. The optimization objective function includes release velocity and negative terms penalizing joint loading and inaccuracy. The weighting coefficient for joint loads has a strong influence on the results. As this coefficient increases, the motion becomes more similar to actual measured pitches. Combining active state patterns optimized for different weighting coefficients gives larger joint loads in the simulated motion. This supports the hypothesis that well-coordinated active states are important for controlling the relationships of the different torque generators in order to create a reasonable and effective pitching motion. The model proposed here is superior to previous simulations for throwing, from the viewpoint of modeling with characteristics of Hill-type muscle function, and can be used to explore realistic baseball pitching.