Nonlinear optimal muscular control of human posture

In earlier work, an optimal control model of human posture regulation that replicated significant features of experimental observations of humans´ response to perturbations was presented. In that work the muscles were modeled as ideal torque generators and the performance measure was quartic in the states and quadratic in the controls. Here, a much more realistic model of muscles is utilized and performance criteria up to eighth order in the states are considered. The resulting closed-loop systems use less muscle force, less neural control signal, and responds faster to initial condition errors than the corresponding linear quadratic optimal control.