- The paper demonstrates that integrating star tracker operational constraints within a nonlinear MPC framework prevents sensor dropouts during aggressive CubeSat maneuvers.
- The approach uses a quaternion-based SO(3) attitude model with horizon truncation and state linearization to meet real-time performance and actuator limits.
- Simulation studies confirmed that the proposed controller achieves mission-required tracking accuracy while maintaining robust STR availability under high angular rates.
Star-Tracker-Constrained Attitude MPC for CubeSats
The paper presents a Model Predictive Control (MPC) framework for CubeSat attitude control that explicitly constrains star tracker (STR) operation, enabling reliable attitude maneuvers while maintaining continuous STR functionality. The integration of star trackers as primary attitude sensors has become prevalent in precision CubeSat missions, but their operation is degraded or interrupted under high angular rates or adverse illumination, motivating a need for constraint-aware guidance and control paradigms. Legacy linear or geometric control approaches typically neglect STR operational boundaries, potentially leading to attitude estimation dropouts and loss of closed-loop control authority.
Methodology
A tailored nonlinear MPC scheme is formulated, incorporating explicit state and input constraints reflecting STR rate tolerances, field-of-view limits, and actuator bounds. The spacecraft attitude dynamics are modeled on SO(3) using quaternion representations, capturing full nonlinear kinematics and control torques. The MPC objective penalizes state deviations and control effort, while nonlinear constraints are imposed to ensure the attitude trajectory always keeps the STR within its operational envelope.
To tackle computational complexity and real-time feasibility, the proposed approach employs horizon truncation, state linearization for initial guess computation, and warm starting based on receding horizon strategies. The authors leverage the open-source “42” NASA attitude simulator for validation, ensuring credible dynamic propagation and sensor/actuator models. The utility of the approach is highlighted in scenarios where STR outage or blinding by bright celestial objects is an operational risk.
Simulation Studies and Results
Extensive nonlinear simulation studies are performed under realistic CubeSat parameters. Comparative analysis with unconstrained MPC and reference governor-based schemes demonstrates the efficacy of the STR-constrained MPC in avoiding estimation dropouts—guaranteeing continuous availability of attitude measurements without sacrificing maneuver agility or steady-state accuracy. The controller maintains fast tracking performance while respecting slew rate and pointing constraints, accommodating actuator saturation and enforcing strict STR operational boundaries. The robust constraint satisfaction is observed across a diverse set of reference trajectories, including aggressive slews and large-angle maneuvers.
Quantitatively, the proposed scheme achieves reference tracking errors within mission requirements throughout all evaluated scenarios, with no loss of attitude estimation despite conditions that precipitate dropouts in non-STR-aware controllers. Control effort remains within practical RW (reaction wheel) limits, and constraint satisfaction is shown to be robust to initial condition uncertainty and model disturbances.
Theoretical and Practical Implications
The incorporation of sensor operational constraints directly into the predictive controller's optimization problem establishes a reliable foundation for safe, aggressive, and autonomous CubeSat operations. This addresses a critical gap in current attitude control architectures by bridging guidance, estimation, and control in a unified constrained optimization framework.
The method is immediately applicable to missions requiring STR-based navigation during agile maneuvers, including Earth observation, laser communications, and constellation phasing. By formalizing sensor constraints, the architecture diminishes the need for ad hoc safety margins or estimation mode switching, thus enabling higher agility and mission utility under hardware-imposed limits. The generality of the approach allows extension to other sensor constraints beyond STRs, and integration with autonomous mission planning and fault recovery.
Future Directions
Potential advancements include autonomous horizon adaptation to further optimize performance-resource trade-offs, integration with real-time onboard optimization hardware, and extension to cooperative multi-agent CubeSat missions. The explicit handling of sensor constraints can inform broader applications in autonomous aerospace and robotic systems where estimation-control coupling is critical. Directions for future research may involve real-space experimental validation, stochastic constraint formulation for robustness to unmodeled disturbances, and hybridization with learning-based or adaptive guidance modules.
Conclusion
The research delivers a significant methodological advancement in constrained attitude control by integrating star tracker limitations within an MPC framework tailored for CubeSat platforms. The approach enables aggressive yet robust maneuvers with continuous high-accuracy attitude estimation—a capability essential for modern autonomous space missions. This framework establishes new standards for safety and performance in constrained spacecraft attitude control, and offers a foundation for further innovation in autonomous, sensor-aware aerospace control systems.