Open-Source High-Fidelity Orbit Estimation for Planetary Science and Space Situational Awareness Using the Tudat Software

TL;DR

The paper presents Tudat, an open-source, modular toolkit for high-fidelity orbit propagation and state estimation that fuses radiometric and astrometric data from DSN/ESTRACK, PRIDE, MPC/NSDC, Horizons, SBDB, and SPICE. It validates real-data processing through case studies on GRAIL and MRO radiometric estimation, Eros astrometry, and Kosmos 482 reentry forecasting, achieving Doppler residuals in the millihertz range and meter- to centimeter-scale orbital accuracy. The work emphasizes open science and reproducibility, showcasing modular data ingestion, high-precision observables, and an extensible framework for planetary science and Space Situational Awareness (SSA) across diverse orbital regimes. Looking ahead, the authors plan to broaden data sources, strengthen validation, and integrate JUICE PRIDE, NSDC, and all-sky SSA pipelines, enabling broader community engagement and collaborative improvement of flight-dynamics software.

Abstract

The TU Delft Astrodynamics Toolbox (Tudat) is a free open-source software suite for research and education in astrodynamics. Initially focused on numerical simulations of orbital dynamics and state estimation, it enables combining optical and radiometric tracking data from multiple sources to estimate the dynamics and parameters of natural and artificial bodies. Recent developments have added functionality for real tracking data analysis, with applications to planetary missions and Space Situational Awareness (SSA). Tudat currently supports processing of (i) deep-space Doppler and range data from DSN and ESTRACK, (ii) Doppler and VLBI data from the PRIDE experiment, and (iii) optical astrometry from the Minor Planet Center (MPC) and Natural Satellite Data Center (NSDC). Using tracking data from the MRO and GRAIL spacecraft and astrometric data of the asteroid Eros, we present prefit residuals (from SPICE-based observables) and postfit residuals (from fitting data to the Tudat dynamical model). Postfit Doppler residuals reach 1-5 mHz for MRO and GRAIL, orbit differences are a few meters for GRAIL and about one meter for MRO, and MRO range residuals are a few meters. From eight years of Eros astrometry, we obtain an orbit difference from the JPL Horizons solution by several tens of kilometers, consistent with its 3-sigma formal error. Tudat's SSA capability is demonstrated by propagating the orbit of Kosmos 482, a Venus lander launched in 1972 that remained in Earth orbit, over 50 years, including its predicted re-entry on May 10 2025. These examples showcase Tudat's modular, flexible, high-fidelity modeling across diverse orbital regimes within a fully open-source framework. All example code is publicly available, and future Tudat analyses will be published with fully reproducible code, allowing anyone in the community to improve and expand upon our work.

Figures (12)

• Figure 1: Visual overview of the available Tudat's interfaces, downloaders, and parsers, allowing for data and metadata retrieval and processing. *Although a unified server for PRIDE data is still being established, the mission data downloader will also support downloading of the Fdets files.
• Figure 2: GRAIL (Panel a) and MRO (Panel b) averaged Doppler residuals.
• Figure 3: MRO sequential range residuals over a year. Different link end receivers have different colors.
• Figure 4: GR035 residuals for New Norcia, Urumqi, and Hart-15m stations on 28/12/2013. Left: Residuals as a function of time. Right: Residuals Distribution. The RMS is expressed in Hertz.
• Figure 5: Astrometric observations (Ra and Dec) used in the state estimation of Eros.