compute orbital positions

This study addresses the limitations of existing open-source tools for planetary position and solar/lunar event computations, which often suffer from heavy dependencies, insufficient accuracy, or lack of lightweight design. To overcome these issues, this work proposes a lightweight, pure-Python astronomical computation library with no external dependencies. Built upon analytical orbital models and coordinate transformation algorithms, the library supports geocentric and heliocentric coordinate calculations, sunrise/sunset and moonrise/moonset times, lunar phases, and conversions among common astronomical coordinate systems, with optional precession corrections for compatibility across reference frames. Validation against JPL DE440 ephemerides shows mean errors of approximately 0.44′ in planetary longitude and 0.16′ in latitude, timing errors of only a few minutes for solar and lunar events, and a lunar illumination error of about 0.2%, demonstrating a favorable balance between high precision and computational efficiency.

Modeling spacecraft–manipulator systems in non-inertial orbital reference frames remains challenging due to strong couplings among spacecraft attitude, orbital motion, and manipulator deformation dynamics. Method: This paper proposes the Lagrange–Poincaré–Kepler (LPK) framework—a novel geometric formulation that integrates Keplerian orbital dynamics and exponential joint parameterization into the Lagrange–Poincaré theory. Built upon the Lagrange–d’Alembert principle on principal bundles, it yields closed-form structural matrices explicitly incorporating orbital perturbations and external symmetry-breaking torques. Contribution/Results: The LPK framework ensures both geometric rigor and computational tractability, enabling hardware-in-the-loop simulation and autonomous control integration. Validation via a 7-DOF manipulator demonstrates significantly improved dynamical fidelity and numerical efficiency under orbital conditions compared with conventional approaches—providing a high-fidelity, computationally efficient dynamical foundation for on-orbit autonomous operations.

This study addresses the challenge of achieving both high accuracy and computational efficiency in solar radiation pressure (SRP) modeling for orbit propagation, particularly for spacecraft with complex geometries or articulated solar arrays. To overcome this, the authors propose a Vulkan-based GPU-accelerated framework that integrates a ray-tracing physical model with dynamic solar array orientation coupling, enabling, for the first time, efficient online computation of high-fidelity SRP forces. The work also systematically evaluates the applicability of precomputed interpolation strategies. Experimental results demonstrate that the proposed approach incurs less than 5×10⁻⁴ relative error compared to a reference OpenGL implementation while accelerating individual SRP computations by 9.4× and full orbit propagation by 15.2×, thereby substantially reducing long-term orbital errors.

This study addresses the challenge of accurately estimating gravitational perturbations and spacecraft states in near-binary asteroid systems using only range and Doppler measurements. To overcome this limitation, the work extends the LiAISON navigation approach—previously limited to weaker perturbation environments—to strong-gravity scenarios such as the Vesta–Ceres system. A novel joint estimation framework is proposed, integrating Multiple Model Adaptive Estimation (MMAE) with the Unscented Kalman Filter (UKF) to simultaneously identify the spacecraft’s state and the harmonic coefficients of the gravitational field. Monte Carlo simulations demonstrate that the method achieves high-precision, consistent estimates of position, velocity, and gravitational perturbations without requiring additional observations, thereby significantly enhancing deep-space autonomous navigation capabilities.

This work addresses key limitations of indirect methods in low-thrust trajectory optimization—namely, the manual derivation of transversality conditions, code reimplementation upon dynamical model changes, and the sensitivity of shooting methods to initial guesses. The authors propose an autonomous optimization agent powered by large language models that accepts natural-language mission descriptions and automatically performs symbolic derivations based on Pontryagin’s Minimum Principle (PMP), validates them via SymPy, and generates high-performance C++ solvers. Central innovations include a constraint-adaptive derivation framework that uniformly handles arbitrary terminal constraints and auto-generates smoothness conditions for free parameters, a dynamics-adaptive four-module architecture accommodating non-standard dynamics, and a comprehensive rule set covering common PMP derivation pitfalls. The approach successfully solves 11 progressively complex scenarios—including rendezvous, multi-phase hovering, gravity assists, and minimum-time solar sail transfers—with 8–48 variables, demonstrating model-agnosticism and scalability.

This study addresses the rapid growth of along-track error in low Earth orbit (LEO) satellite orbit prediction, primarily driven by atmospheric drag modeling inaccuracies, which violates the Gaussian assumption of the state covariance. To mitigate this, the authors propose a novel machine learning–based correction method that exclusively targets the dominant error dimension—the argument of latitude—without modifying the existing orbit propagator. By leveraging single-epoch vector covariance and backward-propagated errors, the approach employs a time-conditioned neural network combined with Gaussian process regression to model and correct unmodeled atmospheric drag effects. The method effectively preserves the physical propagation characteristics in all other state dimensions while significantly improving prediction accuracy, restoring the Gaussianity of the covariance, and successfully extending the validity duration of VCM ephemerides.

This study addresses the challenge of low Earth orbit (LEO) satellite anomaly detection, which is hindered by the scarcity of large-scale, high-quality labeled data. The authors propose a multi-level weakly supervised cascaded labeling framework that integrates physical orbital dynamics, an interacting multiple model–unscented Kalman filter (IMM-UKF), and Gaussian process calibration to automatically generate 8.6 million anomaly sequences from 232 million Two-Line Element (TLE) records—without requiring ground-truth labels. A two-stage Transformer model trained on this synthetic dataset incorporates time-difference features and mean orbital element engineering, substantially enhancing detection performance. Compared to a purely rule-based approach, the IMM-UKF identifies 42.6 times more anomalies. The final model achieves 55.4% maneuver recall and 62.8% decay recall on the test set, with time-difference features contributing a 107% relative improvement in decay recall.

This study addresses the challenge of three-dimensional (3D) localization for low Earth orbit (LEO) satellites by proposing a novel approach that integrates orbital dynamics modeling with interferometric 3D direction-of-arrival (3D-DOA) estimation. By leveraging the prior constraint that LEO satellite motion is confined to a two-dimensional manifold, the method effectively reduces the dimensionality of the 3D localization problem, enabling high-accuracy passive localization and tracking using only a single coplanar array of three antennas. Experimental validation on 81 Starlink satellites demonstrates a 3D-DOA estimation error below 0.7° and an orbital distance error within 5 kilometers, offering an efficient solution for applications such as spectrum management, orbit determination, and GPS-denied backup positioning.

Traditional approaches struggle to efficiently process the large-scale, high-dimensional orbital data of Saturn’s satellite system, hindering a deeper understanding of orbital stability and resonance structures. This work proposes a machine learning–based clustering framework that, for the first time, integrates advanced time-series feature extraction methods such as MiniRocket into astronomical orbital data analysis. By combining automated feature extraction with dimensionality reduction techniques (UMAP/t-SNE) and clustering algorithms (HDBSCAN/K-means), the method effectively characterizes approximately 22,300 simulated orbits. The approach successfully uncovers stable regions and resonant configurations within the system, offering an interpretable and scalable new paradigm for investigating the long-term dynamical evolution of Saturn’s satellites.