Fusion of estimation and guidance using sequential monte carlo methods

Existing missile guidance law design methods are all based on the separation theorem, which has never been proven for realistic guidance scenarios. In such cases, only the general separation theorem may be applied, implying a separately designed estimator, but a guidance law that takes into account the conditional probability density function (PDF) resulting from this estimator. A new approach to fusion of estimation and guidance, under the guidelines of the general separation theorem, is proposed herein. Utilizing particle filtering, the entire state conditional PDF is approximated using the exact nonlinear dynamic models without constraining the analysis to the standard Gaussian noise assumptions. Under the same model assumptions (nonlinear dynamics, non-Gaussian noise), a guidance law is then derived by a geometric approach that provides the necessary conditions to guarantee a hit under ideal (noise-free) conditions, Since it is impossible to guarantee a hit under noisy measurements and uncertainties, an automatic trajectory shaping mechanism is introduced into the fused estimation/guidance algorithm. The trajectory shaping component is used to minimize the effect of the uncertainties and the noisy measurements on the overall interception performance by creating a favorable endgame geometry for the pursuer. A nonlinear, non-Gaussian numerical study is presented, which demonstrates the performance of the proposed methodology in a 3D engagement scenario with partial information