An adaptive terminal guidance scheme based on an exponential cost criterion with application to homing missile guidance

For a linear stochastic system minimizing the expected value of an exponential function of a quadratic yields a control law for the terminal guidance problem which operates linearly on the estimated states. The control gains are explicit functions of the error variance in estimating the state. It is shown that the control gains can be calculated by combining a precalculated matrix determined by a backward integration in time with the error variance calculated forward in time. If the measurement variance is estimated in real time, then the error variance must also be calculated in real time. The control scheme will then be adaptive reflecting the estimated quality of the information. The adaptive control scheme is applied to the terminal phase of a homing missile where the measurement variance is estimated on-line.