Onboard trajectory generation for the unpowered landing of autonomous Reusable Launch Vehicles

Abstract

Onboard trajectory generation dispenses with the pre-defined trajectories used. in Reusable Launch Vehicle (RLV) guidance since the early days of the Shuttle era. This shift, enabled by a new breed of algorithms harnessing modern computer power, offers improvements in performance, robustness, operational cost, and safety. This thesis develops a set of algorithms providing onboard trajectory generation for low lift-over-drag gliding RLVs in subsonic flight below 40,000 ft. The NASA/Orbital Sciences X-34 is used as a representative model for which feasible trajectories are designed over a range of initial conditions without human intervention. In addition to being autonomous, the guidance output of the onboard trajectory generator differs from current Shuttle-based approaches, providing a realistic "future history" in a propagated plan, rather than output commands reacting to perceived instantaneous vehicle needs. Hence, this approach serves an enabling role in a larger research effort to develop a next generation guidance system using an integrated control function. To assess feasibility, the onboard trajectory generator is benchmarked against traditional X-34 guidance for a drop test scenario. The results match in basic form, with differences showcasing the autonomous algorithms' preference for maximum robustness. The true strength of the onboard trajectory generator lies in its ability to handle off-nominal conditions. A series of test cases highlight the ability of the algorithms to effectively cope with anomalous initial drop conditions, reach the desired terminal states, and provide maximum late-trajectory robustness. Computation time is sufficiently brief to suggest a real-time application, after straightforward improvements are made.