Time-Optimal Paths for a Robotic Batting Task

摘要

This paper presents a novel method to optimize the motion of a paddle within a nonprehensile batting task. The proposed approach shows that it is possible to online predict the impact time and the configuration of the paddle, in terms of its linear velocity and orientation, to re-direct a ball towards a desired location, imposing also a desired spin during the free flight. While exploiting the hybrid dynamics of the task during the minimization process, the obtained position and orientation paths are planned by minimizing the acceleration function of the paddle in SE(3). The batting paths are then tracked by a semi-humanoid robot through a closed-loop kinematic inversion. Numerical tests axe implemented to compare different metrics to define the optimal impact time.