An optimal model predictive control model for human postural regulation

A series of convex optimal control problems is proposed as mathematical models of human postural control during quiet standing. The human body is modeled as a two-segment inverted pendulum controlled by a joint torque. Several performance criteria that are quartic in the state and quadratic in the control are utilized. The discrete-time approximation to each of these problems is a convex programming problem. These problems can be solved by the Newton-KKT method. The solutions exhibit many of the experimentally observed postural control phenomena, especially, greater sway than would occur with a linear feedback control without delay.