Comparative study of sensor combinations for autonomous landing of fixed-wing unmanned aircraft

This paper compares the suitability of onboard sensors for autonomous landing of fixed-wing aircraft in the absence of high-fidelity landing aids such as the instrument landing system. Observability analysis was used to identify the feasible sensor combinations. For each feasible combination, an extended Kalman filter-based estimator was designed and integrated with a flight control law which was, in turn, designed to execute a landing maneuver. This setup was tested through simulation to obtain a refined and arguably more realistic comparison of how well the sensor combinations enable a fully automated landing maneuver, including inner-loop flight control. The comparison took into account the differences that arise when a given bank of sensors is used with estimators deployed in a centralized versus a cascaded setting. It was seen, for instance, that GPS as well as a stereovision are adequate for facilitating landings with a centralized estimator setup. Two novel and potentially more efficient applications of the TRIAD algorithm were also demonstrated in the context of automated landing.