Session G2a: 40th Anniversary Track: STUDENT SESSION

Lander-Aided Differential Doppler and One-Way Ranging for Lunar Rover Navigation with a Single Relay Satellite
Kaila M. Y. Coimbra and Grace Gao, Stanford University
Location: Palm
Date/Time: Wednesday, Sep. 16, 2:58 p.m.

As lunar exploration expands, long-range lunar rovers will require positioning methods that remain available when imagery is unreliable and navigation infrastructure is sparse. This paper evaluates a minimal three-node architecture in which a rover localizes using a single lunar relay satellite, its lander, and onboard odometry. The relay is treated as a communications spacecraft rather than a navigation satellite such that the rover only measures one-way Doppler from the relay downlink. The lander provides one-way pseudorange to the rover, and we also investigate the improvement in localization performance when the lander forwards its relay Doppler observation from the satellite so that the rover can form a lander-aided differential Doppler measurement. An Extended Kalman Filter is developed for each measurement configuration, with clock states matched to the links and a Schmidt-Kalman consider-state treatment for broadcast ephemeris error when raw satellite Doppler is used. The simulation includes stochastic rover, lander, and satellite clocks; time-correlated relay ephemeris errors with periodic ODTS updates; thermal measurement noise from link budgets; and mission profiles with continuous, lost, and reacquired lander line of sight. A Fisher-information observability analysis shows that single-satellite Doppler most strongly constrains directions along the satellite sky-track sweep, while lander pseudorange supplies a stable but nearly one-dimensional surface constraint. Combining the two measurement sources removes the kilometer-scale weak horizontal direction present in the satellite-only case and improves vertical observability relative to the lander-only case. In a constant-visibility ablation, the differential satellite-and-lander configuration performs best, reaching sub-10-m three-dimensional position error after 1.8 hours from an initial position error of 100 m per-axis and maintaining that accuracy over 24 hours. In scenarios with surface-link loss or reacquisition, the estimator preserves accuracy through configuration switching. Lander loss leads to odometry-like growth until ephemeris updates restore Doppler utility, while reacquiring the lander reduces the mean error by roughly 60 m. These results show that a landed asset can strengthen lunar rover navigation without requiring a dedicated satellite navigation payload.