This paper describes a novel finite-state conditional switching structure that enables autonomous recovery for a large envelope of loss-of-control conditions. The switching control architecture mimics human pilot recovery actions through numerous discrete states that are activated individually as the recovery proceeds, based on guidelines inspired by the Federal Aviation Administration's suggestions for piloted upset recovery. The switching conditions between discrete states are optimized using interpolation algorithms that extend the recovery envelope while decreasing the average recovery time. The presented architecture is tested on a high-fidelity transport aircraft model under various initial angles of attack, sideslip and roll angles, and roll/pitch/yaw rates. Comparisons with previous methods are conducted to demonstrate that the proposed approach achieves a greater recoverability percentage and lower average recovery time. In addition, the proposed switching architecture is implemented on a high-fidelity F-16 model, and an example recovery scenario is given to demonstrate the implementation of our approach for an agile fighter aircraft model.