Formulation of a One-Step Discrete Stochastic Controller with Application to Adaptive Control

This paper pertains to the formulation and evaluation of a discrete time control algorithm that takes into consideration current state estimates and an uncertain model with stochastic inputs. The algorithm represents a single or one-step formulation of a linear system quadratic performance index solution but introduces the flexibility of a changing weighting parameter in the performance index based on current estimates and on the control saturation limits. The algorithm represents a simple adaptive technique for real time implementation. Application of the algorithm to the inertial stabilization of a line-of-sight system with coulomb friction disturbances is presented.