The purpose of this paper is to show how optimal control methods based on whole-body dynamic models of the human athlete can be very useful in generating optimal movements in sports. We present the mathematical basics of the optimal control approach as well as optimal motions for several examples in 2D and 3D : high bar gymnastics, fast track running, and platform diving with somersaults and twists. Athletes are modeled as underactuated multibody systems with torque actuators and in some cases also with spring-damper elements at the joints. In contrast to purely simulation based approaches, optimal control allows to leave all position and velocity histories as well as torque and force inputs free to be determined during optimization. In combination with biomechanical measurements, these optimization results help to understand the underlying principles of the movements and can ultimately serve to improve the individual movements and training of athletes.
© Copyright 2012 Science et motricité. EDP Sciences. Alle Rechte vorbehalten.