Computational Methods to Improve Satellite Attitude Determination and Control with a Focus on Autonomy, Generalizability, and Underactuation
Author(s)
McKeen, Patrick
DownloadThesis PDF (12.41Mb)
Advisor
Cahoy, Kerri
Terms of use
Metadata
Show full item recordAbstract
The attitude determination and control system (ADCS) onboard a satellite uses sensors to measure orientation and angular velocity, enabling the satellite to manage angular momentum, counteract disturbances, and point in the desired directions. Many historical ADCS approaches are designed for constant pointing goals, high accuracy sensors, powerful actuators, or larger, high-inertia satellites. Many modern satellites are small satellites (tens of kilograms or less), with lower-cost actuators and sensors, and may have more complicated attitude goals. This dissertation presents a variety of computational approaches to improve ADCS performance by leveraging detailed satellite dynamics modeling and estimation, disturbance inclusion, and trajectory planning–all optimized for efficient onboard computation suitable for small satellites. The proposed framework generalizes ADCS operations, allowing it to adapt automatically to different satellite types, mission requirements, and operational goals, reducing reliance on predefined ground-based commands. This framework can be used in place of standard control laws to make ADCS more autonomous and “hands-off,” calculating its own slews and desaturation while meeting pointing goals, even in cases of underactuation or large disturbances. This generalized and autonomous framework is a contribution of this work, alongside each of its components, which can be individually used in their own right. One key component of this work is a generalized state estimator that integrates a dynamic model of the spacecraft. This estimator demonstrates high accuracy across various satellite configurations, achieving angular error as low as 0.01◦ in low Earth orbit (LEO) with highquality sensors (but no star trackers), compared to the typical 1◦ error of conventional methods. The estimator can account for biases, sensor errors, and external disturbances, ensuring robust performance (e.g., 0.1◦ error in LEO) even with lower-quality sensors (MEMS gyroscopes, plus magnetometers and sun sensors). This adaptability highlights the increased autonomy of the system, as it requires minimal human intervention to maintain high accuracy across diverse mission scenarios. Another major contribution is the integration of disturbance modeling into control laws. By accounting for disturbances directly (either individually or as an all-in-one value tracked by the estimator), rather than through reactive measures like integral control, the proposed methods improve stability and performance, particularly for underactuated systems–improving pointing accuracy by up to 20 degrees. The developed control laws are adaptable to various actuator configurations, disturbance environments, and pointing objectives. This flexibility extends to modifying pointing goals, such as aligning specific vectors rather than requiring a fully specified orientation, enhancing mission adaptability. This work also implements a novel trajectory planning method that generates efficient pointing trajectories for both constant and time-varying goals. The method, based on the Augmented Lagrangian iterated-LQR (ALTRO) approach, creates sequential mission trajectories that optimize performance even under underactuation or disturbance conditions. The planned trajectories are followed by two types of robust closed-loop controllers, applicable across satellite architectures ranging from large weather satellites to 3U CubeSats. By enabling onboard trajectory planning and adaptive control adjustments, this method significantly reduces the need for ground-based planning and interventions, further advancing autonomous operation. The combined framework of estimation, disturbance-aware control, and trajectory planning achieves significantly higher accuracy than traditional ADCS approaches. This enables the use of commercial off-the-shelf components in high-performance missions, overcoming the limitations of low-cost sensors and actuators. The proposed methods allow satellites to operate with weaker or fewer actuators, such as magnetic-only control, while still achieving precise pointing, thereby expanding the feasibility of more autonomous, robust, and cost-effective satellite operations.
Date issued
2025-02Department
Massachusetts Institute of Technology. Department of Aeronautics and AstronauticsPublisher
Massachusetts Institute of Technology