Computing LQG/H/sub ∞/ bounded control with guaranteed convergence

Abstract

Fort Worth, TX 76118 Mechanical and Ampace Engineexkg University of Texas at Arlington Arlinaot~TX 76019 Moncrief-O'Donnell Chair for AutanaElectrical Engineering Department University of Texas at Arlington PkIington,TX 76019 Box 19023 Institute (ARRI) t ion and Robotics, ARRI 8 r7-794.5714Abstract A convergence theoxem is presented for an algorithm that solves an LQG conml problem which includes an pedormance bound. The mixed optimization problem setup results in a system of three Riccati equations, two of which are coupled and c8n not be solved independently. The algorithm rapidly converges to a global minimum under very general standard assumptions.