Application of Statistical Linearization and LQG Design to Position Control

Nonlinear Quadratic Gaussian control is applied to a position servomechanism with backlash. It is shown that as more nonlinearities are included in the system that the NQG predictions degrade. This is due to the increasingly nongaussian character of the statistics. A method for alleviating this problem is briefly mentioned. Experimental implementation of the control is shown.