Pith. sign in

REVIEW 4 major objections 4 minor 49 references

Learning-Based Stable Optimal Guidance for Spacecraft Close-Proximity Operations

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single neural network can be trained to certify stability and generate near-optimal guidance for spacecraft rendezvous.

desk verdict Promising supervised-CLF guidance pipeline whose 'certified stability' claim outruns the sampled loss evidence; needs verification or rewording. read the letter →

arxiv 2501.01096 v1 pith:K6CKVD4S submitted 2025-01-02 eess.SY cs.SYmath.OC

Kun Wang , Roberto Armellin , Adam Evans , Harry Holt , Zheng Chen This is my paper · ORCID
classification eess.SYcs.SYmath.OC
keywords spacecraftrendezvouscontrolLyapunovfunctionneuralnetworkguidancenear-optimalbang-bangfuel-optimalClohessy-Wiltshireequationscertifiedstabilitypolynomialmaps
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a single neural network can simultaneously act as a control Lyapunov function—a certificate that the system is stable—and generate near-optimal guidance for spacecraft rendezvous. It constructs the Lyapunov function as \(V_\$\theta$(\mathbf{x}) = [\phi_\$\theta$(\mathbf{x}) - \phi_\$\theta$(\mathbf{x}_e)]^2\), which is positive definite by construction, and derives the thrust direction as the greedy minimizer of \(\dot{V}\) and the throttle as the minimal value needed to enforce decay. A single loss function supervises both the certificate and the policy using optimal state-control pairs obtained offline. If correct, this gives onboard spacecraft a stability guarantee and near-optimal performance for time- and fuel-optimal bang-bang problems in about a millisecond of computation.

What carries the argument

The central object is the squared-difference neural Lyapunov function \(V_\$\theta$(\mathbf{x}) = [\phi_\$\theta$(\mathbf{x}) - \phi_\$\theta$(\mathbf{x}_e)]^2\), which is positive definite by construction. It is coupled to a derived controller: the greedy thrust direction \(\boldsymbol{\$\alpha$} = -(\partial V/\partial \mathbf{x}\,\mathbf{B})^T / \lVert \partial V/\partial \mathbf{x}\,\mathbf{B}\rVert\) minimizes \(\dot{V}\), and the minimal required throttle \(u = (\partial V/\partial \mathbf{x}\,\mathbf{A}\mathbf{x} + \gamma V - \partial V/\partial t_g)/\lVert \partial V/\partial \mathbf{x}\,\mathbf{B}\rVert\) turns the decay condition into a scalar inequality. The decay rate \(\gamma\) is a state-dependent output of the same network. This mechanism carries the argument by replacing all separate CLF loss terms with a single supervised loss on optimal state-control pairs.

What would settle it

Compute the minimal required throttle \(u\) from Eq. (36) on a dense grid of states within the reachable set (especially near the target) using the trained network. If any grid point has \(u > 1\), the decay inequality \(\dot{V} \le -\gamma V\) is violated for the admissible control set \(u \in [0,1]\), disproving the certificate claim at that state. The paper's own Fig. 19 shows such points, so this calculation would settle whether the certificate holds beyond the training samples.

Watch

Extended reading notes

Core claim

The central discovery is that the constraints of a control Lyapunov function can be folded into the control-policy loss rather than treated as separate terms. The squared-difference Lyapunov function \(V_\$\theta$(\mathbf{x}) = [\phi_\$\theta$(\mathbf{x}) - \phi_\$\theta$(\mathbf{x}_e)]^2\) automatically satisfies positive definiteness and vanishes at the equilibrium; the greedy direction \(\boldsymbol{\$\alpha$} = -(\partial V/\partial \mathbf{x}\,\mathbf{B})^T / \lVert \partial V/\partial \mathbf{x}\,\mathbf{B}\rVert\) minimizes the Lyapunov derivative; and the minimal required throttle formula \(u = (\partial V/\partial \mathbf{x}\,\mathbf{A}\mathbf{x} + \gamma V - \partial V/\partial t_g)/\lVert \partial V/\partial \mathbf{x}\,\mathbf{B}\rVert\) replaces the standard decay-inequality penalty. The paper shows that this construction lets one network produce both the certificate and a near-optimal policy, and it reports time-optimal final errors around 6 m and 0.002 m/s, and a 0.38% fuel penalty for the nominal fuel-optimal case.

Load-bearing premise

The load-bearing assumption is that the neural Lyapunov function, trained on finitely many samples, satisfies the decay inequality on the full continuous reachable set; the paper only checks the inequality at sampled states and in simulation, not as a mathematical guarantee.

Editorial extensions

If this is right

  • If the framework works as claimed, spacecraft rendezvous guidance can run in about 0.4 ms per update on a laptop and 6–13 ms on a 200 MHz flight processor, making on-board certified-stable optimal guidance feasible.
  • The same certificate-plus-policy construction should transfer to any control-affine system (the paper says it is easily extensible), potentially covering powered-descent landing, orbital transfer, and attitude control.
  • The state-dependent decay rate appears necessary for the fuel-optimal problem: constant decay rates either converge slowly or fail to converge, while the state-dependent \(\gamma\) enables the 0.38% fuel penalty result.
  • The method's ability to replicate bang-bang throttle policies suggests that supervised learning can handle discontinuous optimal controls when the policy structure is informed by the Lyapunov function, countering previous difficulties noted in the literature.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The 'certified stability' claim is weaker than a formal certificate: the loss only enforces the decay inequality at sampled states, so the paper's guarantee is empirical. A true certificate would require verifying inequality (28) over the whole reachable set.
  • The learned state-dependent decay rate may act as a slack variable that grows near the target to keep the loss small even where the greedy direction cannot satisfy the decay condition—consistent with the observed \(u > 1\) in Fig. 19.
  • A practical onboard implementation would likely combine this learned policy with a two-stage handoff to a local controller near the target, since the paper notes the learned policy chatters due to a small divisor in Eq. (35).
  • One testable extension is to apply the same squared-difference Lyapunov structure to a nonlinear control-affine benchmark (e.g., powered-descent landing) and check whether the minimal-required-throttle condition continues to hold on a dense grid; if not, the framework's stability guarantee would need additional verification.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes a supervised-learning framework for spacecraft close-proximity rendezvous under Clohessy-Wiltshire dynamics, combining a neural control Lyapunov function (CLF) with a greedy thrust-direction policy and a minimal-required-throttle condition. Optimal state-control pairs for training are generated offline via polynomial-map based solution of two-point boundary-value problems. The paper claims that the trained neural network simultaneously certifies a CLF and provides near-optimal time- and fuel-optimal guidance, with stability guaranteed by the CLF decay inequality. Numerical simulations for both problems, including robustness tests and computational-cost estimates, are presented as validation.

Significance. If the stability-certificate claim were rigorously supported, the framework would be a valuable contribution to onboard spacecraft guidance: it unifies CLF certification and supervised policy learning in a single network, avoids separate certificate/policy optimization, and reports millisecond-level inference times. The paper also deserves credit for its efficient polynomial-map dataset generation, the analytically correct derivation of the greedy thrust direction (Eq. 27) and minimal required throttle (Eq. 29), and the extensive numerical experiments including robustness studies and fuel-consumption penalty histograms. The central weakness is that the 'certified' stability claim is not actually certified: the CLF conditions are only enforced at sampled training points and are explicitly violated near the target in the fuel-optimal case. Thus the significance is that of a promising empirical framework, not that of a formally verified guidance scheme.

major comments (4)
  1. [IV.B.1, Eqs. (28), (31)] The CLF decay inequality V_dot <= -gamma V is not verified on the continuous state space; the training loss only penalizes violations at finitely many sampled states via lambda1 max(0, u_i - 1). The reported maximum minimal required throttle of 0.9242 on one nominal trajectory (Fig. 10) is not a certificate over the whole domain. To support the word 'certified,' the authors must either provide a formal verification over the reachable set (e.g., interval analysis, sum-of-squares, or SMT-based checking) or explicitly weaken the claim to empirical stability.
  2. [IV.B.2, Fig. 19 and surrounding text] For the fuel-optimal problem, the minimal required throttle exceeds 1 near the target, as the paper itself states. Since the admissible throttle is bounded by u <= 1, condition (28) cannot hold at those states even with u = 1. The observation that the Lyapunov function decreases along the simulated trajectory (Fig. 17b) is a trajectory check, not a Lyapunov certificate. This directly invalidates the abstract's claim of 'certifiably stable ... fuel-optimal guidance' and the claim that all CLF loss terms are satisfied or replaced by the derived policy.
  3. [IV.B.1, Eq. (25)] The constructed candidate V(x) = (phi(x) - phi(x_e))^2 is positive semidefinite, not automatically positive definite. Positive definiteness on X \ {x_e} requires phi(x) != phi(x_e) for all x != x_e, which is not established anywhere in the paper. Without this condition, even a verified decay inequality would not guarantee asymptotic stability to the target, because V could vanish at other states.
  4. [III.C and IV.B.2, Eq. (33)] For the fuel-optimal problem the Lyapunov candidate depends on the time-to-go t_g, so Definition 2, which applies to the autonomous system in Eq. (1), is not directly applicable. The paper should either state the standard time-varying Lyapunov conditions for the non-autonomous system in (t_g, x), or prove a uniform decay bound for V(t_g, x); otherwise the stability argument for the fuel-optimal case is not formally grounded.
minor comments (4)
  1. [IV.B.1, text after Eq. (31)] The sentence 'The first term encourages the minimal required throttle in Eq. (36) to comply with the admissible control set' should refer to Eq. (29), since Eq. (36) is the minimal required throttle for the fuel-optimal problem.
  2. [IV.C, text near Eq. (30)] The expression 'u = sgn(u)+1/2' is ill-posed because u appears on both sides. It should presumably read u = (sgn(\bar{u}) + 1)/2, where \bar{u} is the minimal required throttle from the preceding derivation.
  3. [V.B.2, text after histogram description] The sentence 'we apply the indirect shooting method to solve the shooting function in Eq. (21)' refers to the smoothing function, not the shooting function; the shooting function for the fuel-optimal problem is Eq. (23).
  4. [V.A.3 and V.B.2] For the time-optimal robustness study, only the success count is reported; reporting a distribution of final-state errors or optimality penalties, analogous to the fuel-optimal histogram in Fig. 25, would strengthen the near-optimality claim across perturbed initial conditions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the supervised losses transparently use optimal labels; the unverified stability certificate is a verification gap, not a circularity.

full rationale

The claimed derivation chain is not circular. The neural Lyapunov candidate V=(phi-phi_e)^2 (Eq. 25) is an explicit construction; the greedy direction (Eq. 27) and minimal required throttle (Eq. 29) are derived from V and the CW dynamics in the paper, with [33] used only as a pointer. The time-optimal loss (Eq. 31) explicitly penalizes sampled violations of the CLF decay inequality through max(0, ubar-1) and clones the optimal direction; the fuel-optimal loss (Eq. 37) explicitly fits ubar to the open-loop optimal throttle u*, so near-matching u* is a supervised-learning outcome, not a hidden prediction. Validation uses held-out initial conditions and compares against an independent indirect-shooting benchmark, so the near-optimality numbers are not forced by construction. The paper's substantive weakness is that 'certified stability' is not actually certified: losses are enforced only at finitely many samples, Eq. (25) enforces only positive semidefiniteness (the text first says 'positive semi-definite' and later claims positive definiteness), and Section V.B.1 admits the admissible throttle constraint is 'occasionally exceeded' near the target in the fuel-optimal case (Fig. 19). These are correctness/verification gaps, not circular reductions, and the self-citations ([33], [43], [44]) do not carry a uniqueness or ansatz argument that makes the conclusion equal to its inputs. Hence no circular step is identified.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim rests on a learned certificate. The neural network parameters and loss weights are fitted to the data, and the validity of the Lyapunov function over the domain is assumed, not proven. The only standard background is Pontryagin's principle and the CLF theorem.

free parameters (7)
  • Neural network parameters theta (phi and gamma) = Not provided (trained weights)
    Trained via Adam on loss (31)/(37); the final Lyapunov function and policy depend entirely on these fitted weights.
  • Loss weights lambda1, lambda2, lambda3 = Time-optimal: (1,1,0.1); fuel-optimal: (1.5,1,0.1)
    Chosen after 'a number of preliminary trials' (Section IV.C); they control the trade-off between stability feasibility and optimality imitation.
  • Bang-bang smoothing constant rho = 600
    Used in Eq. (21) to approximate the discontinuous optimal throttle; chosen to make the dataset smooth while approximating bang-bang.
  • Normalization nominal state = x_nom=[500,-500,1,-1] (time); (t_g,nom,x_nom)=[14400,500,-500,1,-1] (fuel)
    The loss term lambda3 (V(x_nom)-1)^2 sets the scale of the Lyapunov function; the paper claims scaling does not introduce conservativeness.
  • Decay rate gamma = state-dependent network output (learned); constant values 0, 0.001, 0.01, 0.1 in ablation
    For the time-optimal problem, gamma is a learned function; the guidance policy direction is independent of gamma, but the feasibility loss depends on it. In the fuel-optimal case, gamma is critical for training convergence.
  • Policy update period = 3.6 s
    Chosen for closed-loop simulation; affects discretization and final accuracy.
  • Polynomial-map initial domain = center [500,-500,1,-1]; position uncertainties +-75 m (x) and +-150 m (y); velocity +-0.05 m/s
    Defines the distribution of training trajectories; all robustness tests use perturbations inside this domain.
assumptions (6)
  • domain assumption The relative motion is accurately described by the linearized Clohessy-Wiltshire equations with in-plane motion only.
    Section III.A; assumes relative distance much smaller than orbital radius and a circular target orbit.
  • domain assumption The mass variation is negligible during training and for the time-optimal problem.
    Remark 2; the paper includes mass dynamics in dataset generation but ignores it in the learned policy.
  • domain assumption The optimal control solutions generated by the polynomial maps method are accurate and representative of the true optimal policy over the domain.
    Section IV.A; the method is cited to Ref. [44] and is used to generate all labels; no validation of the map accuracy is given in this paper.
  • ad hoc to paper A neural network of 3 to 4 hidden layers can represent a valid Lyapunov function satisfying the CLF conditions on the continuous domain after training on finite samples.
    Section IV.B; the certificate claim rests on this unproven representational and generalization assumption.
  • domain assumption The optimal throttle for the fuel-optimal problem is bang-bang and the smoothing approximation with rho=600 is adequate.
    Section IV.A.2, Eqs. (19)-(21).
  • standard math Pontryagin's Minimum Principle and the standard CLF theorems apply.
    Used to derive the shooting functions and the CLF decay condition.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning-Based Stable Optimal Guidance for Spacecraft Close-Proximity Operations." pith.science (2026). https://pith.science/paper/K6CKVD4S

@misc{pith2026250101096,
  author       = {Pith},
  title        = {Pith review of: Learning-Based Stable Optimal Guidance for Spacecraft Close-Proximity Operations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K6CKVD4S}},
  note         = {Machine review of arXiv:2501.01096}
}
read the original abstract

Machine learning techniques have demonstrated their effectiveness in achieving autonomy and optimality for nonlinear and high-dimensional dynamical systems. However, traditional black-box machine learning methods often lack formal stability guarantees, which are critical for safety-sensitive aerospace applications. This paper proposes a comprehensive framework that combines control Lyapunov functions with supervised learning to provide certifiably stable, time- and fuel-optimal guidance for rendezvous maneuvers governed by Clohessy-Wiltshire dynamics. The framework is easily extensible to nonlinear control-affine systems. A novel neural candidate Lyapunov function is developed to ensure positive definiteness. Subsequently, a control policy is defined, in which the thrust direction vector minimizes the Lyapunov function's time derivative, and the thrust throttle is determined using minimal required throttle. This approach ensures that all loss terms related to the control Lyapunov function are either naturally satisfied or replaced by the derived control policy. To jointly supervise the Lyapunov function and the control policy, a simple loss function is introduced, leveraging optimal state-control pairs obtained by a polynomial maps based method. Consequently, the trained neural network not only certifies the Lyapunov function but also generates a near-optimal guidance policy, even for the bang-bang fuel-optimal problem. Extensive numerical simulations are presented to validate the proposed method.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 27 canonical work pages

  1. [1]

    A review of space robotics technologies for on-orbit servicing,

    Flores-Abad, A., Ma, O., Pham, K., and Ulrich, S., “A review of space robotics technologies for on-orbit servicing,”Progressin AerospaceSciences, Vol. 68, 2014, pp. 1–26. https://doi.org/10.1016/j.paerosci.2014.03.002

    doi:10.1016/j.paerosci.2014.03.002 2014

  2. [2]

    Technology of automated rendezvous and capture in space,

    Polites, M. E., “Technology of automated rendezvous and capture in space,”Journal of Spacecraft and Rockets, Vol. 36, No. 2, 30 1999, pp. 280–291. https://doi.org/10.2514/2.3443

    doi:10.2514/2.3443 1999

  3. [3]

    Commercial avenues for space utilization,

    Dittmar, M., “Commercial avenues for space utilization,”AIAA Space2003 Conference& Exposition, 2003, p. 6234

    work page 2003

  4. [4]

    Survey of numerical methods for trajectory optimization,

    Betts, J. T., “Survey of numerical methods for trajectory optimization,”Journal ofGuidance,Control,andDynamics, Vol. 21, No. 2, 1998, pp. 193–207. https://doi.org/10.2514/2.4231

    doi:10.2514/2.4231 1998

  5. [5]

    Introducing computational guidance and control,

    Lu, P., “Introducing computational guidance and control,”Journal ofGuidance,Control,andDynamics, Vol. 40, No. 2, 2017, pp. 193–193. https://doi.org/10.2514/1.G002745

    doi:10.2514/1.g002745 2017

  6. [6]

    System test results from the GNC experiments on the PRISMA in-orbit test bed,

    Bodin, P., Noteborn, R., Larsson, R., and Chasset, C., “System test results from the GNC experiments on the PRISMA in-orbit test bed,”ActaAstronautica, Vol. 68, No. 7-8, 2011, pp. 862–872. https://doi.org/10.1016/j.actaastro.2010.08.021

    work page doi:10.1016/j.actaastro.2010.08.021 2011

  7. [7]

    Model predictive control approach for guidance of spacecraft rendezvous and proximity maneuvering,

    Di Cairano, S., Park, H., and Kolmanovsky, I., “Model predictive control approach for guidance of spacecraft rendezvous and proximity maneuvering,”International Journal of Robustand NonlinearControl, Vol. 22, No. 12, 2012, pp. 1398–1427. https://doi.org/10.1002/rnc.2827

    work page doi:10.1002/rnc.2827 2012

  8. [8]

    Model predictive control for spacecraft rendezvous and docking: Strategies for handling constraints and case studies,

    Weiss, A., Baldwin, M., Erwin, R. S., and Kolmanovsky, I., “Model predictive control for spacecraft rendezvous and docking: Strategies for handling constraints and case studies,”IEEE Transactionson Control SystemsTechnology, Vol. 23, No. 4, 2015, pp. 1638–1647. https://doi.org/10.1109/TCST.2014.2379639

    work page arXiv 2015

Show all 49 references
  1. [9]

    A tutorial on model predictive control for spacecraft rendezvous,

    Hartley, E. N., “A tutorial on model predictive control for spacecraft rendezvous,”2015 European Control Conference(ECC), IEEE, 2015, pp. 1355–1361. https://doi.org/10.1109/ECC.2015.7330727

    2015

  2. [10]

    A Pontryagin-based NMPC approach for autonomous rendezvous proximity operations,

    Pagone, M., Boggio, M., Novara, C., and Vidano, S., “A Pontryagin-based NMPC approach for autonomous rendezvous proximity operations,”2021 IEEE AerospaceConference(50100), IEEE, 2021, pp. 1–9. https://doi.org/10.1109/AERO50100. 2021.9438226

    2021

  3. [11]

    Fast model predictive control for spacecraft rendezvous and docking with obstacle avoidance,

    Bashnick, C., and Ulrich, S., “Fast model predictive control for spacecraft rendezvous and docking with obstacle avoidance,” Journal of Guidance,Control, andDynamics, Vol. 46, No. 5, 2023, pp. 998–1007. https://doi.org/10.2514/1.G007314

    2023 doi

  4. [12]

    A tube-based approach to nonlinear explicit MPC,

    Bayer, F. A., Brunner, F. D., Lazar, M., Wijnand, M., and Allgöwer, F., “A tube-based approach to nonlinear explicit MPC,” 2016 IEEE 55th Conferenceon Decision and Control (CDC), IEEE, 2016, pp. 4059–4064. https://doi.org/10.1109/CDC.2016. 7798884

    2016 doi

  5. [13]

    Autonomous trajectory planning for rendezvous and proximity operations by conic optimization,

    Lu, P., and Liu, X., “Autonomous trajectory planning for rendezvous and proximity operations by conic optimization,”Journal of Guidance,Control, andDynamics, Vol. 36, No. 2, 2013, pp. 375–389. https://doi.org/10.2514/1.58436

    2013 doi

  6. [14]

    Autonomous optimal trajectory planning for orbital rendezvous, satellite inspection, andfinalapproachbasedonconvexoptimization,

    Ortolano, N., Geller, D. K., and Avery, A., “Autonomous optimal trajectory planning for orbital rendezvous, satellite inspection, andfinalapproachbasedonconvexoptimization,” TheJournaloftheAstronauticalSciences,Vol.68,No.2,2021,pp.444–479. https://doi.org/10.1007/s40295-021-00...

    2021 doi

  7. [15]

    Fast homotopy for spacecraft rendezvous trajectory optimization with discrete logic,

    Malyuta, D., and Acikmese, B., “Fast homotopy for spacecraft rendezvous trajectory optimization with discrete logic,”Journal of Guidance,Control, andDynamics, Vol. 46, No. 7, 2023, pp. 1262–1279. https://doi.org/10.2514/1.G006295

    2023 doi

  8. [16]

    On the stability analysis of deep neural network representations of an optimal state feedback,

    Izzo, D., Tailor, D., and Vasileiou, T., “On the stability analysis of deep neural network representations of an optimal state feedback,” IEEE Transactions on Aerospace and Electronic Systems, Vol. 57, No. 1, 2020, pp. 145–154. https: //doi.org/10.1109/TAES.2020.3010670

    2020

  9. [17]

    Fuel-optimal powered descent guidance for lunar pinpoint landing using neural networks,

    Wang, K., Chen, Z., and Li, J., “Fuel-optimal powered descent guidance for lunar pinpoint landing using neural networks,” Advancesin SpaceResearch, Vol. 74, No. 10, 2024, pp. 5006–5022. https://doi.org/10.1016/j.asr.2024.07.019

    2024 doi

  10. [18]

    Stable optimal feedback control for landers based on machine learning,

    Mulekar, O. S., Cho, H., and Bevilacqua, R., “Stable optimal feedback control for landers based on machine learning,”AIAA Journal, Vol. 62, No. 5, 2024, pp. 1932–1945. https://doi.org/10.2514/1.J063682

    2024 doi

  11. [19]

    Real-time guidance for low-thrust transfers using deep neural networks,

    Izzo, D., and Öztürk, E., “Real-time guidance for low-thrust transfers using deep neural networks,”Journal of Guidance, Control, andDynamics, Vol. 44, No. 2, 2021, pp. 315–327. https://doi.org/10.2514/1.G005254

    2021 doi

  12. [20]

    Real-time optimal control for attitude-constrained solar sailcrafts via neural networks,

    Wang, K., Lu, F., Chen, Z., and Li, J., “Real-time optimal control for attitude-constrained solar sailcrafts via neural networks,” ActaAstronautica, Vol. 216, 2024, pp. 446–458. https://doi.org/10.1016/j.actaastro.2024.01.026

    2024 doi

  13. [21]

    Deep learning techniques for autonomous spacecraft guidance during proximity operations,

    Federici, L., Benedikter, B., and Zavoli, A., “Deep learning techniques for autonomous spacecraft guidance during proximity operations,”Journal of Spacecraftand Rockets, Vol. 58, No. 6, 2021, pp. 1774–1785. https://doi.org/10.2514/1.A35076

    2021 doi

  14. [22]

    Attitude takeover control for noncooperative space targets based on Gaussian processes with online model learning,

    Liu, Y., Wang, P., Lee, C.-H., and Tóth, R., “Attitude takeover control for noncooperative space targets based on Gaussian processes with online model learning,”IEEE Transactionson Aerospace and Electronic Systems, Vol. 60, No. 3, 2024, pp. 3050–3066. https://doi.org/10.1109/T...

    2024

  15. [23]

    Spacecraft rendezvous guidance in cluttered environments via reinforcement learning,

    Broida, J., and Linares, R., “Spacecraft rendezvous guidance in cluttered environments via reinforcement learning,”29th AAS/AIAASpace FlightMechanicsMeeting, American Astronautical Society, 2019, pp. 1–15

    2019

  16. [24]

    Spacecraft proximity maneuvering and rendezvous with collision avoidance based on reinforcement learning,

    Qu, Q., Liu, K., Wang, W., and Lü, J., “Spacecraft proximity maneuvering and rendezvous with collision avoidance based on reinforcement learning,”IEEE Transactionson Aerospaceand Electronic Systems, Vol. 58, No. 6, 2022, pp. 5823–5834. https://doi.org/10.1109/TAES.2022.3180271

    2022

  17. [25]

    Meta-reinforcement learning for adaptive spacecraft guidance during finite-thrust rendezvous missions,

    Federici, L., Scorsoglio, A., Zavoli, A., and Furfaro, R., “Meta-reinforcement learning for adaptive spacecraft guidance during finite-thrust rendezvous missions,”ActaAstronautica, Vol. 201, 2022, pp. 129–141. https://doi.org/10.1016/j.actaastro.2022.08. 047

    2022 doi

  18. [26]

    Reinforcementlearninginspacecraftcontrolapplications: Advances,prospects, and challenges,

    Tipaldi,M.,Iervolino,R.,andMassenio,P.R.,“Reinforcementlearninginspacecraftcontrolapplications: Advances,prospects, and challenges,”AnnualReviewsinControl, Vol. 54, 2022, pp. 1–23. https://doi.org/10.1016/j.arcontrol.2022.07.004. 32

    2022 doi

  19. [27]

    Meta-reinforcementlearningforspacecraftproximityoperationsguidanceandcontrol in cislunar space,

    Fereoli, G., Schaub, H., andDiLizia, P., “Meta-reinforcementlearningforspacecraftproximityoperationsguidanceandcontrol in cislunar space,”Journal of SpacecraftandRockets, 2024, pp. 1–13. https://doi.org/10.2514/1.A36100

    2024 doi

  20. [28]

    Global stabilization of linearized spacecraft rendezvous system by saturated linear feedback,

    Zhou, B., and Lam, J., “Global stabilization of linearized spacecraft rendezvous system by saturated linear feedback,”IEEE Transactions on Control Systems Technology, Vol. 25, No. 6, 2017, pp. 2185–2193. https://doi.org/10.1109/TCST.2016. 2632529

    2017 doi

  21. [29]

    Survey of machine learning techniques in spacecraft control design,

    Shirobokov, M., Trofimov, S., and Ovchinnikov, M., “Survey of machine learning techniques in spacecraft control design,”Acta Astronautica, Vol. 186, 2021, pp. 87–97. https://doi.org/10.1016/j.actaastro.2021.05.018

    2021 doi

  22. [30]

    Neural network optimal feedback control with guaranteed local stability,

    Nakamura-Zimmerer, T., Gong, Q., and Kang, W., “Neural network optimal feedback control with guaranteed local stability,” IEEE Open Journal ofControl Systems, Vol. 1, 2022, pp. 210–222. https://doi.org/10.1109/OJCSYS.2022.3205863

    2022

  23. [31]

    Optimalstrategyforlow-thrustspiraltrajectoriesusingLyapunov-basedguidance,

    Dalin,Y.,Bo,X.,andYoutao,G.,“Optimalstrategyforlow-thrustspiraltrajectoriesusingLyapunov-basedguidance,” Advances in SpaceResearch, Vol. 56, No. 5, 2015, pp. 865–878. https://doi.org/10.1016/j.asr.2015.05.030

    2015 doi

  24. [32]

    Optimal Q-laws via reinforcement learning with guaranteed stability,

    Holt, H., Armellin, R., Baresi, N., Hashida, Y., Turconi, A., Scorsoglio, A., and Furfaro, R., “Optimal Q-laws via reinforcement learning with guaranteed stability,”ActaAstronautica, Vol. 187, 2021, pp. 511–528. https://doi.org/10.1016/j.actaastro.2021. 07.010

    2021 doi

  25. [33]

    Reinforcement learning enhanced LQR and control Lyapunov functions for proximity operations,

    Holt, H., and Armellin, R., “Reinforcement learning enhanced LQR and control Lyapunov functions for proximity operations,” International SymposiumonSpace FlightDynamics(29th: 2024: Darmstadt, Germany), 2024, pp. 1–12

    2024

  26. [34]

    Safe control with learned certificates: A survey of neural Lyapunov, barrier, and contraction methods for robotics and control,

    Dawson, C., Gao, S., and Fan, C., “Safe control with learned certificates: A survey of neural Lyapunov, barrier, and contraction methods for robotics and control,”IEEE Transactions on Robotics, Vol. 39, No. 3, 2023, pp. 1749–1767. https://doi.org/10.1109/TRO.2022.3232542

    2023

  27. [35]

    The Lyapunov neural network: Adaptive stability certification for safe learning of dynamical systems,

    Richards, S. M., Berkenkamp, F., and Krause, A., “The Lyapunov neural network: Adaptive stability certification for safe learning of dynamical systems,”ConferenceonRobotLearning, PMLR, 2018, pp. 466–476

    2018

  28. [36]

    Formal synthesis of Lyapunov neural networks,

    Abate, A., Ahmed, D., Giacobbe, M., and Peruffo, A., “Formal synthesis of Lyapunov neural networks,”IEEEControlSystems Letters, Vol. 5, No. 3, 2020, pp. 773–778. https://doi.org/10.1109/LCSYS.2020.3005328

    2020

  29. [37]

    Stability analysis using quadratic constraints for systems with neural network controllers,

    Yin, H., Seiler, P., and Arcak, M., “Stability analysis using quadratic constraints for systems with neural network controllers,” IEEE TransactionsonAutomaticControl, Vol. 67, No. 4, 2021, pp. 1980–1987. https://doi.org/10.1109/TAC.2021.3069388

    2021

  30. [38]

    10, Springer Science & Business Media, 2013

    Sastry, S.,Nonlinearsystems: analysis,stability,andcontrol, Vol. 10, Springer Science & Business Media, 2013

    2013

  31. [39]

    NeuralLyapunovcontrol,

    Chang,Y.-C.,Roohi,N.,andGao,S.,“NeuralLyapunovcontrol,” AdvancesinNeuralInformationProcessingSystems,Vol.32, 2019. 33

    2019

  32. [40]

    Terminal guidance system for satellite rendezvous,

    Clohessy, W., and Wiltshire, R., “Terminal guidance system for satellite rendezvous,”Journal of the Aerospace Sciences, Vol. 27, No. 9, 1960, pp. 653–658. https://doi.org/10.2514/8.8704

    1960 doi

  33. [41]

    Nonlinear optimal guidance for intercepting stationary targets with impact-time constraints,

    Wang, K., Chen, Z., Wang, H., Li, J., and Shao, X., “Nonlinear optimal guidance for intercepting stationary targets with impact-time constraints,”Journal of Guidance, Control, and Dynamics, Vol. 45, No. 9, 2022, pp. 1614–1626. https://doi.org/10.2514/1.G006666

    2022 doi

  34. [42]

    S.,Mathematicaltheory of optimal processes, Routledge, 2018

    Pontryagin, L. S.,Mathematicaltheory of optimal processes, Routledge, 2018. https://doi.org/10.1201/9780203749319

    2018 doi

  35. [43]

    High-orderguidancefortime-optimallow-thrusttrajectorieswithaccuracy control,

    Evans,A.,Armellin,R.,Pirovano,L.,andBaresi,N.,“High-orderguidancefortime-optimallow-thrusttrajectorieswithaccuracy control,”Journal of Guidance,Control, andDynamics, Vol. 47, No. 2, 2024, pp. 279–290. https://doi.org/10.2514/1.G007540

    2024 doi

  36. [44]

    Fuel-optimal guidance using costate supervised learning with local refinement,

    Evans, A., Armellin, R., Holt, H., and Pirovano, L., “Fuel-optimal guidance using costate supervised learning with local refinement,”ActaAstronautica, Vol. 228, 2025, pp. 17–29. https://doi.org/10.1016/j.actaastro.2024.11.031

    2025 doi

  37. [45]

    High order optimal feedback control of space trajectories with bounded control,

    Di Lizia, P., Armellin, R., Morselli, A., and Bernelli-Zazzera, F., “High order optimal feedback control of space trajectories with bounded control,”ActaAstronautica, Vol. 94, No. 1, 2014, pp. 383–394. https://doi.org/10.1016/j.actaastro.2013.02.011

    2014 doi

  38. [46]

    Control design along trajectories with sums of squares programming,

    Majumdar, A., Ahmadi, A. A., and Tedrake, R., “Control design along trajectories with sums of squares programming,”2013 IEEEInternational ConferenceonRoboticsandAutomation, IEEE, 2013, pp. 4054–4061. https://doi.org/10.1109/ICRA.2013. 6631149

    2013 doi

  39. [47]

    Spacecraft rendezvous using constant-magnitude low thrust,

    Gurfil, P., “Spacecraft rendezvous using constant-magnitude low thrust,”Journal ofGuidance,Control,andDynamics, Vol. 46, No. 11, 2023, pp. 2183–2191. https://doi.org/10.2514/1.G007472

    2023 doi

  40. [48]

    Guidanceandcontrolnetworksfortime-optimalquadcopter flight,

    Origer,S.,DeWagter,C.,Ferede,R.,deCroon,G.C.,andIzzo,D.,“Guidanceandcontrolnetworksfortime-optimalquadcopter flight,”arXiv, 2023. https://doi.org/10.48550/arXiv.2305.02705

    2023 doi arXiv

  41. [49]

    Customized real-time interior-point methods for onboard powered-descent guidance,

    Dueri, D., Acikmese, B., Scharf, D. P., and Harris, M. W., “Customized real-time interior-point methods for onboard powered-descent guidance,” Journal of Guidance, Control, and Dynamics, Vol. 40, No. 2, 2017, pp. 197–212. https: //doi.org/10.2514/1.G001480. 34

    2017 doi

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.