Optimal throwing motion planning for a biped human mechanism

摘要: An optimal motion planning formulation of throwing for a biped human mechanism is proposed as an extension of a previous study. The unique characteristics of the throwing task—highly redundant, highly nonlinear, and highly dynamic—are addressed in this presentation within the framework of multibody dynamics and optimization. To generate physically feasible throwing motions in a fully predictive method without input reference, rigorous dynamic models are associated with the constraints. Given the target location and the object mass, the algorithm outputs the motion, required actuator torques, release parameters, balance criterion, and ground reaction forces. Overarm and sidearm throwing motions are generated as optimal solutions, which demonstrate valid kinematic and kinetic cause-effect relations.