Adaptive Estimation Achieves Precise Spacecraft Attitude and 3D Star-Tracker Alignment

Accurate attitude determination is crucial for deep-space missions where GPS signals are unavailable, and even minor errors can accumulate rapidly. Ridma Ganganath, Simone Servadio, and David Daeyoung Lee, from Iowa State University and California State University, Long Beach, address this challenge with a novel framework for simultaneously estimating spacecraft attitude and compensating for misalignments in star-trackers. Their research introduces an adaptive multi-model estimation technique combining a Multiplicative Extended Kalman Filter with a Bayesian Multiple-Model Adaptive Estimation layer, allowing for robust performance even with imperfect sensor data. This approach significantly improves the accuracy of in-flight calibration for spacecraft with limited computational resources, and importantly, demonstrates the feasibility of estimating dual star-tracker misalignments , a capability that enhances reliability for critical deep-space operations. Through Monte Carlo simulations, the team demonstrates arcsecond-level misalignment estimation and sub-degree attitude errors, proving the consistency and robustness of their system.

This process constructs a six-dimensional grid over the two misalignment quaternions, crucially without augmenting the continuous-state dimension. A novel diversity metric, denoted as Ψ, is introduced to instigate adaptive refinement of the misalignment.

Spacecraft Attitude and Star Tracker Alignment

Scientists have developed an adaptive multi-model framework for accurately determining spacecraft attitude and estimating star-tracker misalignments during deep-space missions where GPS signals are unavailable. This metric ensures the filter maintains a diverse set of hypotheses, enhancing robustness and consistency. The core of the system relies on the MEKF processing gyroscope measurements and TRIAD-based attitude observations, while the MMAE layer operates on a discrete grid representing potential misalignment vectors.

In configurations involving dual-misalignments from two star trackers, the MMAE bank is driven by stacked line-of-sight measurements, forming a six-dimensional grid over the misalignment quaternions without increasing the continuous-state dimension. Tests confirm that estimation errors remain well-bounded throughout the simulations, validating the filter’s stability and reliability under challenging conditions. This breakthrough delivers a practical solution for accurate, autonomous, and computationally efficient in-flight calibration, particularly beneficial for deep-space CubeSat missions.

Spacecraft Attitude and Misalignment Estimation Achieved

This research details a novel adaptive multi-model framework designed for the joint estimation of spacecraft attitude and star-tracker misalignments, specifically for deep-space missions where GPS signals are unavailable. A key innovation lies in a hypothesis-diversity metric, coupled with weighted-mean grid centering, which facilitates adaptive refinement of misalignment estimates without premature convergence and maintains computational tractability for resource-constrained spacecraft.

The authors acknowledge limitations relating to the current model’s inability to account for slowly time-varying misalignment drifts, star-identification errors, or hardware-related issues. Future work intends to address these points and validate the framework through hardware-in-the-loop demonstrations, paving the way for operational deep-space attitude determination systems.