New adaptive filtering method improves autonomous navigation for deep-space missions

Autonomous orbit determination is a critical capability for deep-space exploration, especially during close-proximity operations around small bodies such as asteroids and comets. In these environments, gravitational forces are extremely weak, non-gravitational disturbances can be significant, and communication delays limit reliance on ground-based navigation support. Traditional filtering techniques often degrade under such conditions because they rely on linearization assumptions or fixed noise models that cannot adapt to rapidly changing dynamics.

To address these challenges, the research team developed the Uncertainty-Aware Filtering framework (https://doi.org/10.1016/j.cja.2025.103971), which integrates sigma-point nonlinear state propagation with a physics-informed adaptive noise adjustment strategy. The sigma-point method avoids the linearization errors inherent in EKF, enabling more accurate state propagation in highly nonlinear regimes. At the same time, UAF continuously monitors the statistical behavior of observation residuals and dynamically adjusts noise parameters to reflect changing uncertainty levels.

Unlike conventional adaptive filters that apply uniform noise inflation, UAF distinguishes between different sources of anomalies. When residuals indicate localized and short-lived measurement errors, the algorithm selectively increases measurement noise for affected channels. When persistent, multi-channel anomalies are detected, UAF interprets them as signs of unmodeled dynamic disturbances and increases process noise accordingly. This targeted adaptation helps maintain estimator consistency and prevents unnecessary divergence.

Simulation experiments based on the Tianwen-2 mission scenario demonstrate that UAF achieves substantially improved performance. Under sustained unknown perturbations on the order of 10⁻¹¹ km/s², UAF reduces orbit prediction errors by nearly an order of magnitude compared with EKF. At the same time, it maintains more stable residual statistics than both EKF and UKF, indicating improved robustness to model uncertainty.

The algorithm also offers significant computational advantages. By eliminating the need for Jacobian matrix calculations and exhibiting faster convergence behavior, UAF reduces overall computational cost by approximately 42.5% relative to EKF. This improvement is particularly important for onboard navigation systems, where computing resources are limited and real-time performance is essential.

Robustness tests further show that UAF can reliably converge even when the initial orbit estimate contains large uncertainties. In simulations with initial position errors up to 0.7 km, UAF consistently achieves stable and accurate orbit solutions, while EKF often struggles to maintain convergence. Additional experiments confirm that UAF can effectively handle both measurement anomalies and unmodeled external forces without compromising stability.

The study also highlights the importance of orbital geometry and observability. Results indicate that at very low orbital altitudes (below about 1 km), limited observation geometry can significantly degrade estimation performance. In contrast, orbital configurations above approximately 3 km in altitude and with inclinations between 60° and 110° provide favorable observation conditions, allowing UAF to operate reliably even in the presence of modeling uncertainties.

“Our results show that uncertainty-aware adaptive filtering can substantially improve autonomous navigation in challenging deep-space environments,” said professor Jianguo Yan, lead author of the study. “The proposed method offers a practical balance between accuracy, robustness, and computational efficiency.”

The researchers believe that UAF provides a valuable navigation solution not only for Tianwen-2, but also for future small-body exploration missions, including asteroid and comet rendezvous and sampling-return missions.

Original Source

YANG Wanling, YAN Jianguo*, SUN Shangbiao, et al. Robust autonomous orbit determination with dynamical uncertainty awareness: Application to Tianwen-2 at 311P[J]. Chinese Journal of Aeronautics, 2026, 39(6): 103971. https://doi.org/10.1016/j.cja.2025.103971

About Chinese Journal of Aeronautics 

Chinese Journal of Aeronautics (CJA) is an open access, peer-reviewed international journal covering all aspects of aerospace engineering, monthly published by Elsevier. The Journal reports the scientific and technological achievements and frontiers in aeronautic engineering and astronautic engineering, in both theory and practice. CJA is indexed in SCI (IF = 6.9, Q1), EI, IAA, AJ, CSA, Scopus.