This report summarizes a multi-year investigation effort resulting in an innovative integrated estimation/guidance design algorithm for the interception of randomly maneuvering targets using noise-corrupted measurements. The algorithm is based on separating the tasks of model identification, state reconstruction and change detection, as well as assigning them to different estimators. The selection of the appropriate estimator for providing guidance information is accomplished by explicit logic-based use of the time-to-go. The homing guidance is performed by a differential game-based bounded control guidance law, modified for enhancing its efficiency in the terminal phase. The algorithm was derived using a planar linearized model, but it was implemented and validated in a generic nonlinear three-dimensional ballistic missile defense scenario. The simulation results demonstrate an exceptional homing performance improvement compared to earlier results and a potential to achieve hit-to-kill accuracy.