A Fault-Tolerant Navigation Framework for Space Vehicle Coordination Landing

This paper proposes a multi-source data navigation framework designed for the formation landing of reusable launch vehicles (RLVs). The cornerstone of this framework is a centralized Dual Quaternion Unscented Kalman Filter (DQNUKF), which is specifically designed to correctly handle the Special Euclidean group, SE(3) manifold geometry of the six-degree-of-freedom (6-DoF) pose, ensuring a theoretically sound state estimation. In addition to conventional rocket sensor data, the framework incorporates relative observations between multiple rockets to enhance coordination and fault tolerance. An embedded Fault Detection and Isolation (FDI) algorithm further improves system reliability. Simulation results under various sensor failure and communication packet loss scenarios demonstrate that the proposed framework maintains high navigation accuracy and robust performance. Key advantages of this framework include high-precision state estimation on manifolds, dual robustness against both sensor and communication faults, and an innovative fault-tolerant cooperative design, making it a promising solution for RLV landing operations.