Autonomous navigation method for Mars aerobraking based on fast orbit propagation via intelligent compensation

Aerobraking reduces fuel consumption and is crucial for Mars sample return. High‑precision navigation is required to ensure spacecraft safety in the complex aerobraking environment. However, the orbital altitude changes significantly during the process. For autonomous spacecraft navigation, high‑precision measurements may only be possible when the spacecraft is in the vicinity of Mars. At excessively high orbital altitudes, sensor measurement accuracy degrades, so high‑fidelity orbit propagation becomes the primary means for autonomous navigation. High-precision orbit propagation demands high-fidelity dynamics, yet onboard computational resources often limit fast high-fidelity integration, leading to poor real-time performance. To achieve high‑precision on‑board autonomous navigation with low computational load, this paper proposes an autonomous navigation method for Mars aerobraking based on fast orbit propagation via intelligent compensation. The method decomposes the high‑fidelity dynamics into a two‑body component and a nonlinear residual component, and uses a neural network to compensate for the residual between the two‑body dynamics and the high‑fidelity dynamics, thereby achieving fast and high‑fidelity orbit propagation. In the near‑Mars phase, an integrated inertial/visual navigation scheme and an angle observation model with enhanced observability are adopted, so that the impacts of nonlinear dynamics and atmospheric uncertainties are reduced and navigation accuracy is improved. Apoapsis is crucial for orbit maneuvers, but orbit propagation accuracy is insufficient. To improve apoapsis prediction, an apoapsis state prediction network is designed that uses high‑precision estimates from the end of the near‑Mars phase to predict the apoapsis state and calibrate the time. Finally, a set of numerical simulations is designed to verify the performance of the proposed method and demonstrate its feasibility for Mars aerobraking navigation.