Comparison of feed-forward control strategies for simplified vertical hopping model with intrinsic muscle properties.

Abstract

To analyse walking, running or hopping motions, models with high degrees of freedom are usually used. However simple reductionist models are advantageous within certain limits. In a simple manner, the hopping motion is generally modelled by a spring-mass system, resulting in piecewise smooth dynamics with marginally stable periodic solutions. For a more realistic behaviour, the spring is replaced by a variety of muscle models due to which asymptotically stable periodic motions may occur. The intrinsic properties of the muscle model, i.e. preflexes, are usually taken into account in three complexities-constant, linear and Hill-type. In this paper, we propose a semi-closed form feed-forward control which represents the muscle activation and results in symmetrical hopping motion. The research question is whether hopping motions with symmetric force-time history have advantages over asymmetric ones in two aspects. The first aspect is its applicability for describing human motion. The second aspect is related to robotics where the efficiency is expressed in term of performance measures. The symmetric systems are compared with each other and with those from the literature using performance measures such as hopping height, energetic efficiency, stability of the periodic orbit, and dynamical robustness estimated by the local integrity measure (LIM). The paper also demonstrates that the DynIn MatLab Toolbox that has been developed for the estimation of the LIM of equilibrium points is applicable for periodic orbits.