⚡ electrical engineering

Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(3)

This paper demonstrates that the error dynamics of thrusting spacecraft are nearly group affine on the $SE_2(3)$ Lie group and can be effectively linearized via a dynamic inversion control law, a finding validated by numerical simulations showing agreement between predicted log-dynamics and classical Newtonian trajectories.

Original authors: Micah K. Condie, Abigaile E. Woodbury, Li-Yu Lin, Kartik A. Pant, Michael Walker, James Goppert

Published 2026-02-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Micah K. Condie, Abigaile E. Woodbury, Li-Yu Lin, Kartik A. Pant, Michael Walker, James Goppert

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: The "GPS" Problem for Spacecraft

Imagine you are trying to guide a fleet of thousands of new satellites (like the Starlink constellation) to move around Earth. They need to perform delicate dance moves: changing orbits, avoiding collisions, and docking with other ships.

To do this, engineers usually use simplified maps. Think of these maps like a flat drawing of a globe. They work great if you are walking a few steps in a park, but if you try to use that flat map to navigate a flight around the entire Earth, you get lost because the map doesn't account for the curve of the planet or the changing pull of gravity.

In space, these "flat maps" are called linear models (like the HCW equations mentioned in the paper). They assume that if you push a satellite a little bit, it moves in a straight line. But in reality, space is curved, gravity changes depending on how high you are, and the satellite is constantly spinning. When you try to use the "flat map" for complex maneuvers, the errors pile up, and the satellite might miss its target or crash.

The Solution: A "Shape-Shifting" Coordinate System

The authors of this paper propose a new way to look at the problem. Instead of using a flat map, they use a 3D, shape-shifting coordinate system based on a mathematical concept called a Lie Group (specifically, something called $SE_2(3)$ ).

The Analogy: The Roller Coaster vs. The Train

Old Way (Linear Models): Imagine trying to describe a roller coaster ride by drawing a straight line on a piece of paper. You can describe the start and the end, but you can't capture the loops, the drops, or the G-forces. If you try to steer a train based on that straight line, it will fly off the tracks.
New Way (The Paper's Method): Imagine the roller coaster track itself is your coordinate system. You don't fight the curves; you ride along them. The math in this paper creates a system where the "track" (the math) naturally bends and twists just like the spacecraft's path does.

The Three Key Discoveries

The paper makes three main points, which we can break down like this:

1. The "Gravity Glitch"

The authors discovered that if you use this new 3D coordinate system, the spacecraft's movement is almost perfectly predictable, except for one thing: Gravity.

The Metaphor: Imagine you are driving a car on a perfectly smooth, self-adjusting road that knows exactly how to turn. The only thing messing up your perfect drive is a sudden, unpredictable gust of wind (Gravity) that pushes you slightly off course.
The Finding: In this new math system, the "wind" (gravity) is the only thing that makes the equations messy. Everything else (how the ship spins, how it accelerates) fits perfectly into the system.

2. The "Safety Net" (Bounding the Error)

The paper proves that even though gravity is a "glitch," it's a predictable glitch.

The Metaphor: Think of gravity as a rubber band. If you stretch the rubber band (move the spacecraft far away from its target), the pull gets stronger, but it never snaps. The authors calculated exactly how strong that pull can get.
The Result: They created a "safety net" (a mathematical bound). They showed that as long as the spacecraft stays within a certain distance, the gravity error won't get big enough to break the system. This means engineers can use standard, simple control tools without needing to calculate complex gravity corrections for every single second.

3. The "Magic Eraser" (Dynamic Inversion)

If you need the system to be perfectly linear (with zero errors), the authors offer a "Magic Eraser."

The Metaphor: Imagine you are driving on that self-adjusting road, and the wind (gravity) keeps pushing you. The "Magic Eraser" is a control system that instantly calculates exactly how hard the wind is pushing and applies an equal force in the opposite direction.
The Result: By using this specific control law, the gravity "glitch" is completely canceled out. The spacecraft's error dynamics become perfectly linear. It's as if the spacecraft is floating in deep space with no gravity at all, making it incredibly easy to steer.

The Proof: The "Molniya" Test

To prove this works, the authors ran a computer simulation.

The Scenario: They simulated two spacecraft in a Molniya orbit. This is a weird, highly oval-shaped orbit where the spacecraft zooms very close to Earth (where gravity is strong) and then flies very far away (where gravity is weak). It's the "hardest test case" imaginable.
The Result: They compared their new "shape-shifting" math against the old, heavy-duty physics calculations. The results matched almost perfectly (within a tiny fraction of a percent). Even when the spacecraft was moving at vastly different speeds and distances, the new math held up.

Why Does This Matter?

As we launch thousands of new satellites, we can't rely on old, simplified maps. We need a way to control these ships that is:

Accurate: It works even when gravity changes wildly.
Simple: It doesn't require supercomputers to solve complex equations every second.
Safe: It guarantees the ships won't drift apart or crash.

This paper provides the mathematical "GPS" that allows us to navigate the complex, curved reality of space using simple, reliable rules. It turns a chaotic, non-linear problem into a neat, linear one that engineers can easily solve.

1. Problem Statement

The rapid expansion of satellite constellations (e.g., Starlink, OneWeb) has increased the frequency of orbital maneuvers, necessitating precise, continuous-thrust guidance and control. Classical relative-motion models, such as the Hill–Clohessy–Wiltshire (HCW) and Tschauner–Hempel/Yamanaka–Ankersen (TH/YA) equations, rely on local linearization around a reference orbit. These models suffer from two primary limitations in modern thrusting scenarios:

Limited Validity: They are only accurate within a small neighborhood of the reference trajectory.
Nonlinear Coupling: They fail to explicitly capture the nonlinear coupling between translational acceleration (commanded in the body frame) and rotational attitude, which becomes significant during long-duration or attitude-coupled thrusting maneuvers.

The authors aim to develop a geometric framework that describes the dynamics of a thrusting spacecraft globally, avoiding the limitations of local linearization while maintaining a structure amenable to linear control design.

2. Methodology

The paper utilizes Lie Group theory, specifically the group $SE_2(3)$ (the group of double direct spatial isometries), to model the spacecraft state.

State Representation: The spacecraft state $X \in SE_2(3)$ is defined as a matrix containing the attitude $R \in SO(3)$ , inertial velocity $v$ , and inertial position $p$ .
Mixed-Invariant Dynamics: The authors formulate the equations of motion as a "mixed-invariant" system on $SE_2(3)$ . In this framework, the system dynamics are shown to be group-affine (a property where error dynamics evolve independently of the trajectory) except for one specific term: gravity.
Error Definition: A left-invariant error $\eta = X^{-1}\bar{X}$ is defined, where $\bar{X}$ is the reference trajectory. The tracking error is mapped to the Lie algebra $\mathfrak{se}_2(3)$ via the logarithmic map, resulting in a 9-dimensional vector $\xi = \text{Log}(\eta)^\vee$ representing position, velocity, and attitude errors.
Gravity as a Perturbation: The analysis identifies that the position-dependent gravity term $g(p)$ is the sole obstruction to the system being perfectly group-affine. The difference in gravity between the actual and reference positions ( $g(p) - g(\bar{p})$ ) acts as a nonlinearity.

3. Key Contributions

The paper makes three primary theoretical and practical contributions:

Geometric Embedding: It demonstrates that the translational and rotational dynamics of a thrusting spacecraft can be embedded in $SE_2(3)$ as a mixed-invariant system. This provides a structurally consistent representation of tracking errors that does not rely on local linearization.
Bounded Nonlinearity: The authors derive a rigorous bound on the gravity-induced mismatch. They prove that for bounded tracking errors and orbital radii above a minimum altitude, the gravity mismatch acts as a linear perturbation proportional to the position error. This allows the use of standard linear stability and reachability tools without needing to compensate for gravity explicitly.
Exact Log-Linearization via Control: A dynamic inversion control law is proposed that completely eliminates the gravity mismatch term. When applied, the closed-loop error dynamics become exactly log-linear (i.e., $\dot{\xi} = A(t)\xi$ ), independent of the reference trajectory's specific path. This yields a linear time-varying (or time-invariant if reference velocity is constant) system.

4. Results

The theoretical findings were validated through numerical simulations using a highly elliptical Molniya-like orbit (eccentricity $e=0.74$ ), which presents a severe test case due to large variations in orbital radius and gravity gradients.

Validation of Log-Dynamics: The error dynamics predicted by the derived log-linear equation (Eq. 18) were compared against classical Newtonian integration. Over two full orbits, with position errors growing from 219 m to 2,407 km and velocity errors reaching 1.97 km/s, the two methods agreed to within a relative error of $1.6 \times 10^{-8}\%$ . The residuals were attributed solely to numerical integration tolerances.
Gravity Mismatch Bound: The derived bound on the gravity mismatch (Proposition 1) was verified. The actual mismatch remained well below the theoretical pointwise bound throughout the simulation. A global constant bound was calculated (11.6 m/s²), which is sufficient for mission planning without requiring real-time knowledge of both trajectories.
Control Law Effectiveness: The dynamic inversion control law successfully canceled the gravity term, confirming that the system can be rendered exactly linear in the log-error coordinates.

5. Significance

This work bridges the gap between complex nonlinear spacecraft dynamics and linear control theory:

Unified Framework: It places thrusting spacecraft dynamics within the invariant systems framework, offering a geometrically consistent alternative to HCW/TH/YA models that degrades over large distances or long durations.
Simplified Control Design: By showing that gravity acts as a bounded perturbation (or can be canceled), the paper enables the direct application of Linear Matrix Inequality (LMI) based tools for safety analysis, reachable set computation, and controller synthesis.
Scalability: The approach is particularly relevant for mega-constellations where frequent, precise maneuvers are required. It allows for the design of robust controllers that remain valid even when the spacecraft deviates significantly from the reference trajectory, a scenario where classical linearized models often fail.
Future Applications: The authors note that this framework can be extended to recover classical dynamics through specific frame changes and to analyze impulsive $\Delta v$ planning within a unified geometric context.

In summary, the paper provides a mathematically rigorous method to treat thrusting spacecraft dynamics as nearly linear systems on a Lie group, significantly simplifying the analysis and control of complex orbital maneuvers.