Optimal adaptive control with ΘLQG

This thesis introduces ΘLQG, the hyper extension of the Linear Quadratic Gaussian Regulator, for solving general purpose (locally optimal) nonlinear adaptive control on multiple-input multiple-output systems. This approach can be used to stabilize an under-actuated under-observed nonlinear dynamic system with noisy measurements while simultaneously identifying system parameters. It requires a parameterized dynamic model with operation about equilibrium points. Its general construction can also be used in disturbance rejection, gain scheduling and reference tracking. Specific case examples are introduced