Sensor Fusion and PVT Techniques for Autonomous Lunar Rover Navigation

Giuseppe Tomasicchio, Luca Andolfi, Simone Giannattasio, Salvatore Cassano, Luca Ostrogovich, Alfredo Renga, Nicolò Galletta, Michèle Lavagna

Peer Reviewed

Abstract:	Lunar exploration is rapidly advancing, with initiatives like ESA’s Terrae Novae and NASA’s Artemis prioritizing technologies for sustained surface operations. Autonomous rovers are fundamental in this context, enabling scientific, technological, and resource prospecting activities on the Moon. However, current Visual-Inertial-Based Navigation (VIBN) systems suffer from drift accumulation over time, limiting long-term autonomy and necessitating frequent ground intervention. This study investigates the integration of GNSS-like lunar navigation signals with rover VIBN systems to improve rover localization and mapping accuracy. Specifically, one-way range and range-rate measurements from a simulated lunar navigation satellites constellation are fused with visual, inertial and Digital Elevation Models (DEM) data using a loosely coupled Extended Kalman Filter. Validation is conducted in a high-fidelity simulation environment based on Telespazio’s Interactive Mission Modeling Visualization & Validation (IMMV2 ) tool and its Visual Scenario Generator (VSG) module. The virtual environment reproduces the lunar South Pole with realistic lighting and terrain using Lunar Reconnaissance Orbiter (LRO) Lunar Orbiter Laser Altimeter (LOLA) DEM enhanced from 5?m/pixel to 10?cm/pixel resolution via fractal noise and texture layering. Procedural rocks and craters further increase realism. The navigation architecture includes simultaneous localization and mapping, as well as hazard detection via a YOLO-based deep neural network trained on synthetic datasets. GNSS-like observables are used when available to mitigate visual drift. Observables are generated throughout two reference trajectories. An open one, optimized with an A* pathfinding algorithm that avoids slopes exceeding 20°, and a closed one for the validation of mapping and loop closure. Results over a 500?m traverse show that multi-sensor fusion significantly enhances autonomous navigation performance, achieving a root mean square positioning error (RMSE) of 95?cm when four satellites are in view, and up to 4.6?m RMSE when limited to two satellites, thus enhancing navigation reliability in poor visibility conditions.
Published in:	Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025) September 8 - 12, 2025 Hilton Baltimore Inner Harbor Baltimore, Maryland
Pages:	2977 - 2999
Cite this article:	Tomasicchio, Giuseppe, Andolfi, Luca, Giannattasio, Simone, Cassano, Salvatore, Ostrogovich, Luca, Renga, Alfredo, Galletta, Nicolò, Lavagna, Michèle, "Sensor Fusion and PVT Techniques for Autonomous Lunar Rover Navigation," Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025), Baltimore, Maryland, September 2025, pp. 2977-2999. https://doi.org/10.33012/2025.20262
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In