Instability Dependent on Absolute Orientation in Basilisk ADCS Simulation

[BennerThanYou96]

Hello all,

I am a master’s student at Portland State University working with the Portland State Aerospace Society (PSAS) on the OreSat1.0 CubeSat mission (2U, scheduled for launch in early 2026). My thesis work focuses on the design, implementation, and validation of OreSat’s ADCS using Basilisk. We are using a pyramid configuration of reaction wheels for attitude control.

Over the past several weeks, we have been investigating a puzzling instability in our simulations. Initially, we thought the issue was tied to the size of the commanded rotation (e.g., stable up to ~89°, but unstable at 90° or larger). However, after further testing we now believe the problem is independent of rotation angle and instead depends on the spacecraft’s absolute orientation in Basilisk’s coordinate system.

Specifically:

• If the spacecraft begins in what we might call the “stable hemisphere” and is commanded to hold its attitude, it remains stable indefinitely.
• If the spacecraft begins in the “unstable hemisphere” and is simply told to hold its attitude (no large slews), the system becomes unstable even without motion.
• Even more telling: if the spacecraft is already unstable and is then commanded to slew into the stable hemisphere, it performs the maneuver perfectly and remains stable.
• This behavior is consistent across different control modes, including both our LQR state–space controller (based on Markley & Crassidis but implemented in state–space form rather than as a 3-axis PID) and a bang–bang pre-slew controller.
• The controller operates only on the relative error quaternion and body rates; it has no knowledge of absolute orientation. Thus, the controller itself should behave uniformly regardless of absolute attitude.

This strongly suggests the instability is related to how Basilisk handles absolute orientation in the background, rather than to the control law itself.

The OreSat1.0 flight software and Basilisk setup are fully open source. If anyone would like to review the minimal reproducible case, the code can be found here: OreSat Simulation Folder

We would greatly appreciate any insight into whether this is:

• a known artifact of Basilisk’s attitude representation
• an issue with our usage of the simulation framework
• or something deeper that may need attention in Basilisk itself

If anyone needs more information I would of course be happy to answer any questions.

Thank you very much for your time and for maintaining this excellent tool.

Best regards,
Sebastian Benner