![[2602.15954] Hybrid Model Predictive Control with Physics-Informed Neural Network for Satellite Attitude Control](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.15954] Hybrid Model Predictive Control with Physics-Informed Neural Network for Satellite Attitude Control
Summary
This paper presents a hybrid model predictive control approach using physics-informed neural networks for improved satellite attitude control, demonstrating significant performance enhancements over traditional methods.
Why It Matters
The research addresses the challenges of accurate spacecraft attitude control, which is critical for mission success. By integrating physics-informed neural networks with model predictive control, the study offers a novel solution that enhances predictive reliability and robustness, which is vital in aerospace applications where precision is paramount.
Key Takeaways
- Physics-informed neural networks significantly improve predictive reliability in spacecraft attitude control.
- Hybrid control formulations combining learned dynamics with nominal models enhance system performance.
- The study achieved a 68.17% reduction in mean relative error and improved response times by up to 76.42%.
Computer Science > Robotics arXiv:2602.15954 (cs) [Submitted on 17 Feb 2026] Title:Hybrid Model Predictive Control with Physics-Informed Neural Network for Satellite Attitude Control Authors:Carlo Cena, Mauro Martini, Marcello Chiaberge View a PDF of the paper titled Hybrid Model Predictive Control with Physics-Informed Neural Network for Satellite Attitude Control, by Carlo Cena and Mauro Martini and Marcello Chiaberge View PDF HTML (experimental) Abstract:Reliable spacecraft attitude control depends on accurate prediction of attitude dynamics, particularly when model-based strategies such as Model Predictive Control (MPC) are employed, where performance is limited by the quality of the internal system model. For spacecraft with complex dynamics, obtaining accurate physics-based models can be difficult, time-consuming, or computationally heavy. Learning-based system identification presents a compelling alternative; however, models trained exclusively on data frequently exhibit fragile stability properties and limited extrapolation capability. This work explores Physics-Informed Neural Networks (PINNs) for modeling spacecraft attitude dynamics and contrasts it with a conventional data-driven approach. A comprehensive dataset is generated using high-fidelity numerical simulations, and two learning methodologies are investigated: a purely data-driven pipeline and a physics-regularized approach that incorporates prior knowledge into the optimization process. The results indic...
