Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(3)
summary
The gist
The error dynamics of a thrusting spacecraft are nearly group affine on the SE2(3) Lie group, and the nonlinearity can be bounded, or removed with the application of a dynamic inversion control law.
In short
The paper shows that a thrusting spacecraft's motion can be modeled using SE2(3), a Lie group structure, which is nearly group-affine. The main challenge is gravity, which introduces nonlinearity. The authors derive an error dynamics model and propose a dynamic inversion control law to eliminate this gravity mismatch entirely, resulting in exactly log-linear closed-loop error dynamics for precise tracking.
Key concepts
- SE2(3) Lie Group
- This is a mathematical structure used to represent the spacecraft's motion. It describes the spatial isometries of 3D space, providing a consistent framework for analyzing how the spacecraft moves relative to its reference path. This representation helps simplify complex dynamics into a structured form.
- Group-Affine Dynamics
- The paper suggests that thrusting spacecraft dynamics are nearly group-affine when embedded in SE2(3). Group-affinity means the system's evolution can be described linearly within this mathematical framework, which is crucial because it allows tracking errors to evolve independently of the spacecraft's absolute position.
- Dynamic Inversion Control Law
- This is a specific control strategy used to counteract the gravity-induced nonlinearity. By applying this law, the system actively compensates for deviations from the ideal group-affine dynamics, ensuring that the resulting error dynamics are perfectly log-linear and predictable.
Terminology used across episodes
This episode discusses
The paper
Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(3) · Read on arXiv
Purdue University · Cromulence
Body-fixed spacecraft thrust couples attitude tracking error to translational motion because the applied inertial acceleration depends on spacecraft attitude. In this work, we formulate the dynamics of a continuously thrusting spacecraft on SE 2(3) to make this coupling explicit while retaining a structured form suitable for analysis. We represent attitude, velocity, and position within a single Lie-group state and derive the logarithmic dynamics of a left-invariant tracking error with position-dependent gravity. The resulting dynamics show explicitly how attitude error affects translational velocity through the applied acceleration and separate the structured linear dynamics from nonlinear contributions due to body-frame input mismatch and position-dependent gravity difference. We specialize the gravity contribution to the two-body model and derive an analytical bound in terms of the reference orbital radius, logarithmic position error, and attitude error. Numerical simulations of a continuous-thrust maneuver in a highly elliptical Molniya-like orbit show agreement between the derived logarithmic error dynamics and errors reconstructed from independent nonlinear state propagation.
Transcript
Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.
Rosa: We're looking at the paper: "Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(3)".
Dev: Here's the gist of the paper. The error dynamics of a thrusting spacecraft are nearly group affine on the SE2(3) Lie group, and the nonlinearity can be bounded, or removed with the application of a dynamic inversion control law.
Rosa: Foundational Framework This technical note shows that the disturbance-free dynamics of a thrusting spacecraft can be embedded in the Lie group SE2(3), also known as the group of double direct spatial isometries in a nearly mixed-invariant form that is approximately group-affine.
Dev: Under this representation, the left-invariant configuration-tracking error evolves independently of the system trajectory and admits a linear representation in the associated Lie algebra, rather than arising from a local linearization as in HCW or TH/YA.
Rosa: This property follows from results in invariant systems theory and provides a structurally consistent way to represent tracking error for thrusting spacecraft.
Title and authors: Rosa: So let’s talk about who wrote this. We have Micah K Condie, Abigaile E Woodbury, Li-Yu Lin, Kartik A Pant, and James Goppert from Purdue University. They are the folks doing the heavy lifting here.
Dev: Yeah, they’re a solid team with different backgrounds probably covering both the geometric group theory side and the actual control engineering aspects of this problem. It's good to see that mix when you tackle something this deep into Lie groups.
Taro: I always look for papers where people are trying to bridge theoretical geometry with practical orbital mechanics, because that’s usually where the hardest problems live in terms of real-world failure modes.
Rosa: Yeah, and this paper seems to be doing just that by taking something abstract like SE2(three) and applying it directly to a concrete problem: thrusting spacecraft motion <ref:2512.05888#pg1>. It shows how these dynamics fit into an invariant systems framework.
Dev: The title itself points us toward the solution they propose, Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(three) <ref:2512.05888#pg1,for Thrusting Spacecraft on SE2(3>. That inversion part is what’s really interesting to me—how you can use control laws to remove that nonlinearity.
Taro: I wonder how robust this log-linear representation is when you push it outside the perfect conditions they might be testing in the lab. Does it hold up when the world misbehaves?
Rosa: That’s a big question for me. They are showing that even with bounded tracking errors and orbital radii above a minimum altitude, gravity mismatch acts as a perturbation that can be bounded on page two of THIS PAPER <ref:2512.05888#pg1>. It suggests we get some stability guarantees even if the conditions aren't perfect yet.
Dev: That perturbation idea is key because it moves us away from needing a perfect model for every tiny movement. It means we can treat the gravity issue as something we can manage, not something that breaks the whole structure of our control design.
The paper's summary: Rosa: Okay, so to summarize what’s going on in this paper about Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(three) <ref:2512.05888#pg1,for Thrusting Spacecraft on SE2(3>. They start by showing the dynamics are embedded in a nearly mixed-invariant form on SE2(three), which is approximately group affine <ref:2512.05888#pg1>.
Dev: And they establish that this representation means the left-invariant configuration tracking error evolves independently of the actual system trajectory, and it has a linear representation in the associated Lie algebra, not some local linearization.
Taro: That independence from the trajectory sounds fantastic for planning because it decouples some of those coupling effects we usually have to worry about when things move around.
Rosa: Right. They then show how they derive the log-linear spacecraft error dynamics on page three of THIS PAPER, which is xi evolving according to terms involving the adjoint operator and mismatch terms like ntilde and mtilde <ref:2512.05888#pg1>.
Dev: Those mismatch terms are defined by how the reference path doesn't match the actual system parameters, specifically gravity mismatches. They define ntilde as something related to angular errors and mtilde as something related to position errors from gravity differences on page three of THIS PAPER <ref:2512.05888#pg1>.
Taro: So they’ve quantified exactly what’s causing the deviation from perfect log-linearity—it's just those specific mismatch terms, not some unknown high-order nonlinear term.
Rosa: Precisely. And the big result is that they can eliminate this gravity mismatch entirely using a dynamic inversion control law, which is what they call u one on page three of THIS PAPER <ref:2512.05888#pg1,a dynamic inversion control law>.
Dev: That control law effectively sets the gravity mismatch to zero, meaning the error dynamics become exactly log-linear after you apply that specific feedback.
The paper's improvements: Rosa: Now for the improvements they propose, which is where this paper gets really useful for us in practice. They show two main things that make it better than just having a standard log-linear model.
Dev: First, they show that under a specific feedback control law, u one which is defined as negative adjoint map of gravity minus the reference gravity on page three of THIS PAPER, the gravity mismatch is eliminated completely <ref:2512.05888#pg1>.
Taro: Eliminating it entirely sounds like a huge win for stability. It means the error dynamics are exactly log-linear without any residual gravitational noise interfering with the control loop.
Rosa: And second, they combine that with another stabilizing feedback term, u two which is defined using some gain K on page three of THIS PAPER, and you get this closed-loop error dynamics that is exactly linear in time, like A t plus B K xi <ref:2512.05888#pg1>.
Dev: That’s the payoff. You go from a nearly group affine system with a bounded nonlinearity to an exactly linear closed-loop error dynamics, which is much easier to analyze for stability and reachability.
Taro: So it gives us a clean mathematical object—the linear error dynamics—that we can use standard tools on, like LQR or Kalman filtering, knowing the underlying structure is sound because of the SE2(three) embedding <ref:2512.05888#pg1>.
Rosa: And they validate this stuff numerically in a highly elliptical orbit, showing that their log-error dynamics exactly reproduce the true relative motion to within numerical integration tolerances.
Conclusion: Dev: So wrapping up on Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(three) <ref:2512.05888#pg1,for Thrusting Spacecraft on SE2(3>. The paper establishes that the dynamics can be embedded in a nearly group affine form, with gravity as the main issue.
Rosa: And the key result is that by using a dynamic inversion control law, you eliminate this gravity mismatch entirely, leading to exactly log-linear closed-loop error dynamics. It’s a powerful way to handle tracking errors without needing complex local linearizations.
Taro: From my testing side, it’s good to see that the actual gravity mismatch stays within the derived bound throughout a full orbital period, which confirms their analysis holds up in real motion.
Dev: And if we use standard tools like LQR with that resulting linear error dynamics, we can get a very stable and predictable system where the error falls with the square of the sampling interval.
Rosa: It’s a solid foundation for building reliable control systems on these spacecraft, especially when you are dealing with those frequent maneuvers needed for mega-constellations.
Taro: I think what this means is we get better guarantees on how well our autonomy performs when we push the system into those complex orbital regimes where gravity effects are significant.
Dev: Exactly. We move from guessing about nonlinearity to having a mathematical structure that we can actively control and bound precisely.
Rosa: And that’s what this paper, Log-linear Dynamic Inversion for Thrusting Spacecraft on SE2(three), gives us <ref:2512.05888#pg1,for Thrusting Spacecraft on SE2(3>. We’ll be looking at how these structural properties help us design the next generation of autonomous space systems next time.
More episodes
- 2608.13438-ContactGuard: Pre-Contact Execution Monitoring with Action-Conditioned Latent World Models
- 2610.34593-Future Prediction as Training-Only Video Supervision for Pretrained VLA Policies
- 2610.35385-EmbodiedLab: Beyond Manipulation, Benchmarking Embodied Agents That Act to Know
- 2610.36378-Data-Driven Hamiltonian for Neural Optimal Control
- 2610.37275-A Heuristic for Scheduling Autonomous Vehicles at Intersections
- 2610.37288-Dynamic Coordination of Autonomous Vehicles at Intersections
- 2610.37632-Real-Time Coordination of Autonomous Vehicles at Intersections
- 2611.13779-Softness is something you also need! Why compliance helps control: information metrics of a passive walker with compliant ankle
- 2611.13411-Residual-Corrected Physics-Informed Neural Network KKL Observers with Input-to-State Stability Guarantees
- 2611.14488-Is Your Robot's Memory Just a Clock? Temporal Shortcuts in Memory-Augmented VLA Policies