Time-Varying Model Predictive Attitude Control for Magnetically Actuated Dual-Spin Satellites

Robert D. Halverson and Ryan J. Caverly · 2025

Read Time-Varying Model Predictive Attitude Control for Magnetically Actuated Dual-Spin Satellites by Robert D. Halverson and Ryan J. Caverly (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-07858.txt.

Introduction

Robert Halverson and Ryan Caverly tackle steering for a small satellite that spins to stay stable, in a 2025 paper. The satellite uses magnetic rods that push against Earth's field, saving the power and space that wheels or thrusters would take. The catch is that magnetic steering is weak and changes as the satellite orbits, so the controller must plan ahead carefully.

The team built three planning controllers that predict the satellite's motion with a time-varying model. One accounts for the satellite's position along its orbit, and two refine their guesses step by step, one with straight-line math and one with the full curved math. In tests aimed at holding a steady pointing direction, the full-math version used the least effort, broke no limits, and stayed cheap to run.

Catalog record

CREATORRobert D. Halverson and Ryan J. Caverly
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-07858

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Time-Varying Model Predictive Attitude Control for Magnetically Actuated Dual-Spin Satellites Robert D. Halverson and Ryan J. Caverly Abstract Attitude control hardware for small satellites is often limited in power and space availability given the importance of the science instruments they exist to transport. To mitigate this, a dual-spin stabilized satellite actuated via magnetic torque rods reduces the space and power required of the attitude control system, but may require advanced control policies. This paper explores the attitude control of a magnetically actuated dual-spin stabilized CubeSat with model predictive control using time-varying prediction dynamics. An inertial pointing objective is used as a representative mission, where the satellite is able to deviate from its nominal orientation within some allowable amount. Three time-varying MPC policies are developed and compared to ensure the system does not violate constraints while minimizing control effort. These policies include a prediction model that accounts for the orbital position of the satellite and two iterative approaches that incorporate control inputs through either propagation of the linear dynamics, or nonlinear propagation with successive linearization. Results demonstrate that the nonlinear prediction policy outperforms other prediction methods with regards to not only minimal actuation, but to constraint satisfaction as well while imparting minimal computational burden. 1 Introduction T he small satellite revolution is bringing forth a new age of space exploration technologies, causing spacecraft to become increasingly small while continuing to deliver important science payloads and instruments to low-Earth orbit (LEO) and beyond [ 1 ] . Cube Satellites (CubeSats) are standardized and cost-effective platforms that are accessible to not only the spacecraft manufacturing industry, but many universities and research institutions alike [ 2 , 3 ] . Science instruments often use a great amount of power and space within CubeSats, which limits the power available for use by other important components including attitude determination and control system (ADCS) hardware. The goal of the ADCS is to determine the satellite’s orientation with respect to some inertial reference frame in real time while simultaneously controlling the satellite via available actuators. CubeSats are often equipped with three reaction wheels that are nominally non-spinning and exchange momentum with the satellite bus to produce a torque in a desired direction. Magnetic torque rods are usually then included to interact with the Earth’s magnetic field when actuated to produce a small torque for the purpose of momentum management. This paper explores the problem where a CubeSat does not have enough space or power available for three reaction wheels. Rather, the spacecraft is equipped with one nominally spinning wheel (i.e., a ‘momentum wheel’) and three magnetic torque rods mounted orthogonally within the spacecraft. The spacecraft is spun around the same axis of the wheel’s momentum vector, which defines the dual-spin stabilized configuration [ 4 , 5 ] . This configuration provides passive stability properties such that a gyroscopic stiffness is provided to the spin axis. This work is inspired by the EXACT (Experiments in X-ray Characterization and Timing) satellite being designed and built at the University of Minnesota [ 6 ] . This mission employs a 3U (30cm x 10cm x 10cm) CubeSat that gathers data from the HAFX (Hard and Fast X-Ray Spectrometer) science instrument. The primary objective of this mission requires the instrument boresight attached to the CubeSat to point at the Crab Pulsar within a specified half-cone angle. Figure 1 illustrates the different pointing cone configurations for this representative mission. Importantly, the science instrument allows the CubeSat to spin about its boresight axis, where a minimum spin rate is required based on the novel attitude determination system that was developed for the satellite [ 7 ] . Figure 1 : An illustration of three configurations for a 3U CubeSat pointing at the distant Crab Nebula. There is a considerable amount of literature regarding the use of solely magnetic actuation beginning with some of the first publications investigating magnetic actuation for spinning satellites in the 1960’s [ 8 ] . A useful survey on the use of various magnetic control approaches for spacecraft including linear and nonlinear control theory, as well as predictive control is found in Ref. [ 9 ] , while a deep-dive into magnetic attitude control for small satellites is presented in Ref. [ 10 ] . Particularly relevant literature in this area include an investigation into periodic linear quadratic control for nadir-pointing spacecraft in a bias momentum configuration using magnetic actuation [ 11 ] , sliding-mode control for satellites in the presence of disturbance torques [ 12 ] , and an investigation into controllability and stability via linear quadratic regulation (LQR) with magnetic actuation [ 13 ] . Other work includes the application of periodic LQR with proven steady-state errors of less than one degree in a nadir-pointing mission and robustness to model uncertainties and unmodeled dynamics [ 14 ] , as well as the application of full-state feedback via LQR with an observer-controller design for magnetic attitude control [ 15 ] . Much of existing literature uses some form of LQR control for magnetic actuation, however this often requires extensive tuning and risks an aggressive and sometimes over-actuating control law. Further, the use of a static or pre-scheduled control gain introduces the risk of limited robustness to uncertainty in system and environment dynamics. To mitigate this, a predictive control policy can be utilized. Model predictive control (MPC) has been used for a myriad of aerospace control applications including autonomous station keeping of satellites [ 16 , 17 ] , spacecraft attitude control and relative motion control [ 18 , 19 ] , and many others as outlined in the survey in Ref. [ 20 ] . The benefit of MPC comes from the fact that it selects the optimal control inputs via a user-defined optimization cost function with the capability of incorporating state and input control constraints while optimizing over a receding-time-horizon, subject to expected disturbances and state response. The application of MPC to the LEO magnetic actuation control problem has been explored using only magnetic torque rods [ 21 , 22 , 23 ] , as well as under-actuated systems with two reaction wheels as opposed to three or four [ 24 ] . Nonlinear control of dual-spin and bias-momentum spacecraft via magnetic actuation has also been investigated [ 25 , 26 ] . Ref. [ 27 ] presented an LTV (linear time-varying) MPC policy for the stabilization of a magnetically actuated dual-spin spacecraft, and applied their policy to an LTV simulation that includes a gravity gradient torque. This paper expands on the literature by introducing complex constraints to the dual-spin stabilized system, including minimum and maximum roll rate limitations, second-order cone constraints on the spacecraft’s pointing vector, and the inclusion of soft constraints that allow for continued feasibility in the MPC problem. Crucially, this expansion enables implementation of dual-spin satellites beyond a regulation problem to allow for state deviations that are acceptable within the pointing constraints of the EXACT mission. Further, the predictive control policies are tested in a nonlinear simulation that includes three important disturbance torques: gravity gradient, aerodynamic, and residual magnetic dipole torques. The iterative approach to the LTV predictions used in the MPC policies in this paper provides improved prediction accuracy that likewise improves performance, while

Read the complete text