Pith. sign in

REVIEW 3 major objections 5 minor 52 references

Design And Optimization Of Multi-rendezvous Manoeuvres Based On Reinforcement Learning And Convex Optimization

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A three-stage optimizer pairs learned routing with convex refinement to plan multi-rendezvous missions near-optimally.

desk verdict Solid engineering framework, but the near-optimality claim rests on an unvalidated tour-level cost model. read the letter →

arxiv 2411.11778 v1 pith:QADIKLP4 submitted 2024-11-18 eess.SY cs.SY

Antonio López Rivera , Lucrezia Marcovaldi , Jesús Ramírez , Alex Cuenca , David Bermejo This is my paper · ORCID
classification eess.SYcs.SY
keywords multi-rendezvoustrajectoryoptimizationspacetravelingsalesmanproblemreinforcementlearningattentionmodelsequentialconvexprogrammingon-orbitservicingJ2perturbationmissiondesign
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 tries to establish that a modular, three-stage pipeline can solve the space traveling salesman problem—choosing both the order of targets and the transfer trajectories between them—for a realistic on-orbit servicing and deployment mission. The key demonstrated result is that an attention-based routing policy trained with reinforcement learning, used to seed population-based heuristic search, finds tours within a mean 3.02% of the heuristic benchmark's fuel cost when combined with beam search, beating the handcrafted heuristics considered. Sequential convex programming then re-optimizes each transfer arc under thruster constraints, producing feasible trajectories, and one non-coplanar case reduced propellant by about 8.5% relative to the analytical estimate. If correct, this gives mission designers a practical way to explore many mission scenarios quickly while retaining a path to feasible, constraint-satisfying trajectories.

What carries the argument

The load-bearing machinery is a two-level cost pipeline. At the combinatorial level, tours are scored by an analytical estimator that sequences Multiple Hohmann Transfers and Nodal Inclination Changes, choosing the node order to perform the inclination change at the highest semi-major axis, and propagates the J2 secular drift of RAAN and argument of perigee. This estimator produces the fuel, delta-V, and time-of-flight values that the heuristic search and the reinforcement learning policy optimize. At the trajectory level, a sequential convex programming solver takes each combinatorial arc as a warm start, normalizes state and control variables, and enforces thrust-magnitude and duty-cycle constraints while minimizing terminal state error. The third element is the RL attention policy itself: a graph-attention encoder and pointer decoder trained with policy-gradient methods, whose samples and beam-search trajectories seed the combinatorial search.

What would settle it

Generate a random sample of tours, compute each tour's cost with the analytical estimator and then with the high-fidelity SCP or functional simulator, and compare the rankings: if the analytical order disagrees with the refined order on a meaningful fraction of tour pairs, the RL training signal and heuristic search inherit a biased cost model and the reported gaps do not reflect true mission performance.

Watch

Extended reading notes

Core claim

The central claim is that reinforcement learning for combinatorial optimization transfers successfully to spacecraft routing, and that a decomposition-based framework can exploit it. The framework combines heuristic combinatorial optimization over target sequences, a fast analytical transfer-cost estimator based on sequential Multiple Hohmann Transfer and Nodal Inclination Change maneuvers under J2 secular drift, and a sequential convex programming layer that converts each analytical transfer into a feasible, thrust-constrained trajectory in modified equinoctial elements. On the OSSIE mission case study, the paper reports that the learned policy with beam search reaches a mean 3.02% optimality gap relative to near-optimal heuristic solutions, outperforms the four handcrafted heuristics, and that the convex refinement layer meets orbit injection accuracy requirements while keeping delta-V close to the combinatorial estimate. The paper also reports a Monte Carlo campaign of 5000 mission scenarios showing the framework can map the mission design envelope and identify the main cost drivers.

Load-bearing premise

The load-bearing premise is that the analytical transfer-cost model, which uses J2-only secular propagation and idealized MHT-NIC maneuvers, ranks candidate tours in the same order as the true high-fidelity dynamics would, and this ranking is never validated tour-by-tour.

Editorial extensions

If this is right

  • Multi-rendezvous missions with a dozen or more targets can be optimized in a modular pipeline, so new target sets or propulsion limits require swapping modules rather than reformulating the whole problem.
  • An RL-trained routing policy can serve as a learned heuristic that accelerates and improves population-based search on this space traveling salesman problem variant, with beam search giving the best final solutions.
  • The same framework can be run in a Monte Carlo loop to map the feasible mission-design envelope, identifying which mission parameters drive cost—here, number of deployment bundles and inclination spread.
  • SCP re-optimization can turn idealized impulsive transfer sequences into feasible trajectories under actuator on/off constraints, and high-fidelity simulation confirms that the optimized trajectory is followable in at least one test case.

Reading between the lines

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

  • Beyond the paper: because the RL policy is trained on the analytical cost model, replacing that model with a higher-fidelity propagator would likely change tour rankings, so measuring the policy's gap against high-fidelity-corrected tours is a natural next test.
  • Beyond the paper: the same decomposition—search over sequences scored by a cheap estimator followed by convex refinement of arcs—should transfer to other drifting-target problems such as active debris removal or constellation deployment, where the mass-deployment sequence also affects cost.
  • Beyond the paper: the SCP layer's warm-start assumption, in which combinatorial timing fixes the actuator windows, may be the main source of suboptimality when several arcs are optimized together, and relaxing it could recover additional propellant savings.
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

3 major / 5 minor

Summary. This manuscript proposes a three-stage framework for the multi-target rendezvous trajectory optimization problem, demonstrated on the UARX Space OSSIE mission. Stage one performs population-based heuristic combinatorial optimization over visit sequences, with optional seeding from an attention-based routing policy trained by REINFORCE, A2C, or PPO; transfer costs are estimated by an analytical sequential MHT-NIC model with secular J2 drift. Stage two re-optimizes individual transfer arcs with a Sequential Convex Programming solver tailored to OSSIE thrust constraints. Stage three verifies selected trajectories in the SENER Functional Engineering Simulator. The authors report a 3.02% mean gap between their best RL-policy solutions and heuristic solutions, SCP injection errors within mission requirements, and a successful preliminary FES verification of one coplanar transfer arc.

Significance. The modular architecture with standardized interfaces is a genuine strength, and the idea of using an RL-trained attention policy to seed heuristic population-based search is a sensible way to combine neural combinatorial optimization with an existing optimization pipeline. The SCP formulation is standard, and the four test cases provide an initial demonstration of actuator-constrained re-optimization; the FES verification, although preliminary, grounds the work in an industry mission context. If the near-optimality and mission-feasibility claims were fully supported, the framework would be a useful practical design tool for multi-rendezvous missions. At present, however, the tour-level evidence is insufficient: the combinatorial and RL layers rely entirely on an analytical cost model that is not validated at tour level, and the reported optimality gaps are measured against heuristic solutions produced with that same model.

major comments (3)
  1. [§4, §5.2.2, §6.1, §7.6] The MHT-NIC/J2 cost model (Eqs. 9-15 and Algorithm 1) is the only tour-ranking oracle for the heuristic optimizer, the RL training signal, and the 5000-scenario Monte Carlo analysis, but the paper provides no tour-level validation of this estimator against high-fidelity dynamics; the only FES verification (Section 7.6) covers a single coplanar transfer arc. Since SCP re-optimizes individual arcs from warm starts and does not reorder tours, any ranking bias in the analytical model is inherited by the final sequence. Table 5 itself shows that the decomposition is approximate, with an 8.53% ΔV reduction on one non-coplanar arc, so the missing tour-level validation is load-bearing. Please add a tour-level comparison of estimator costs against FES or an equivalent high-fidelity propagator for a representative set of tours, and quantify rank correlation or maximum cost inversion.
  2. [§7.3, Table 3] The quantity labeled 'optimality gap' is not an optimality gap in the usual sense: it is the relative difference between the RL policy's tours and heuristic solutions computed with the same MHT-NIC cost model. A 3.02% gap therefore shows that the learned policy is competitive with the considered heuristics in-domain, but it does not establish near-optimality of the tours for the actual mission. Please rename this metric (e.g., 'gap to heuristic baseline') and, if near-optimality is claimed, compare against an exact method, a lower bound, or at least a substantially different independent solver.
  3. [§7.4-§7.6] The statement that OSSIE is 'on average capable of fulfilling its mission and decommissioning afterwards' (Section 7.4) is based on 5000 scenarios solved with the analytical estimator only. SCP re-optimization is demonstrated for four transfer arcs, and FES verification for one arc, so the full-dynamics feasibility of the mission-level Monte Carlo envelope is not established. Please either temper the mission-feasibility claim to 'feasible under the analytical model' or extend SCP/FES verification to a representative sample of tours covering the scenario distribution.
minor comments (5)
  1. [§4.1.1, Eq. (9)] In Eqs. (9b) and (9c), the symbol 'n' inside the square roots appears to be a typo for 'ξ'; as printed, the expressions are dimensionally inconsistent with the standard Hohmann transfer formulas.
  2. [Table 5] The header of Table 5 lists 'Δe_target[deg]' and 'Δi_target[-]', but the units appear to be swapped: eccentricity error is dimensionless and inclination error is in degrees.
  3. [§7.3] The reported training batch size of 5096 is suspicious; please confirm whether this is 512, 5096, or another value, and report the other RL hyperparameters and the heuristic-optimizer settings used to generate the benchmark solutions.
  4. [§5.1, Eq. (19)] The phrase 'strict feasibility guarantees' overstates what Eq. (19) provides, because the final-state constraint is relaxed into a weighted penalty; the later acknowledgment that attitude constraints are omitted should also be reflected in this phrasing.
  5. [§7.6] The text states that the FES test 'is compliant with client requirements' but does not give the quantitative requirement or the achieved errors; please report the actual numbers so the verification can be assessed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the optimization loop is internal to a stated cost model, with independent arc-level SCP and FES verification; the only self-citation ([23]) is for solver implementation details and is not load-bearing.

full rationale

The derivation chain is self-contained rather than circular. The combinatorial objective (Eq. 16) is defined by the MHT-NIC/J2 cost model, and both the heuristic optimizer and the RL policy minimize exactly that objective; the reported 3.02% "optimality gap" in Table 3 is a gap to heuristic solutions computed with the same model, so it is an in-domain algorithmic benchmark, not a fitted quantity disguised as a prediction. No parameter is fit to a subset and then used to predict a quantity it was fit to: the RL policy is trained by REINFORCE/A2C/PPO on the cost signal, and the heuristic benchmark is a separate solver on the same cost. The SCP block (Sec. 5.4) re-optimizes arcs under nonlinear MEE dynamics and thrust constraints from warm starts; this is an independent refinement, and one coplanar arc is checked in the high-fidelity FES (Sec. 7.6), which includes drag, higher-order zonals, third-body and SRP. The only self-citation, [23], supplies SCP solver details and is not used to forbid alternatives or to import a uniqueness theorem; the algorithm is described as classical SCP with trust regions. The absence of tour-level validation of the MHT-NIC estimator against full-dynamics propagation is a genuine external-validity/correctness gap, but it is not circularity: the estimator's ranking bias would not be hidden by the in-domain gap, yet the paper's claims are explicitly relative to its model and arc-level verification. Hence no step reduces by construction to its inputs.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The sequence-level claims rest on the analytical MHT-NIC cost model and the J2-only perturbation model; the RL policy is trained on the same cost model used for evaluation, so its 'optimality gap' is an in-domain measure. The SCP layer and the high-fidelity simulator provide external grounding only for individual transfer arcs, not for the tour ordering. The four fitted/hand-chosen parameter groups above are what the central results depend on beyond standard astrodynamics.

free parameters (4)
  • SCP weight matrices P and R (Equation 19)
    Hand-tuned to prioritize final-state accuracy over control cost; directly influences all SCP re-optimization results (Table 5) and the reported Delta V changes.
  • RL policy network weights (attention model) = Trained on 100,000 scenarios
    The central learned component; its quality dictates the claimed 3.02% optimality gap, and the evaluation uses the same cost model that generated the training signal.
  • Mission scenario distribution parameters (Table 2) = SMA spread 50 km; SSO inclination range; exponential mass with floor; uniform bundles between 2 and 13
    Hand-chosen statistical model for OSSIE clients; defines the RL training/test distribution and the Monte Carlo mission feasibility analysis, so all results are conditional on it.
  • RL training hyperparameters = batch size 5096; 50 epochs; 128-dim embedding; search strategies greedy/stochastic/beam
    Chosen by hand without ablations; affects the reported policy performance and training times (Table 3).
assumptions (5)
  • domain assumption J2-only secular perturbation model (Equation 8) is sufficient to propagate target states and the OTV state over mission durations.
    Invoked in the combinatorial cost estimator and RL environment (Sections 4.4, 5.2.2, 6.1); ignores drag, higher-order harmonics, and third-body effects for the sequence-level search.
  • domain assumption The sequential MHT-NIC decomposition (Algorithm 1), performing the inclination change at the highest semi-major axis, is a near-optimal guidance policy for combined altitude and inclination changes under OSSIE's thrust constraints.
    Underpins the analytical transfer cost in Section 4.3; the decomposition is intuitive (lower velocity reduces NIC cost) but not proven optimal for multi-burn constrained vehicles.
  • domain assumption The SCP solver from SOTB (based on trust-region sequential convex programming and interior-point methods) converges to a feasible, locally optimal trajectory for the discretized OCP.
    The paper relies on the cited solver behavior [23] without providing convergence guarantees; the results in Section 7.5 are empirical.
  • domain assumption The OSSIE mission scenario model (Table 2) is representative of real client payload requests.
    The RL policy and the 5000-scenario Monte Carlo study (Section 7.4) are built on these distributions; deviations in real operations would change the reported feasibility rates and policy performance.
  • standard math Standard two-body dynamics with J2 perturbation in Modified Equinoctial Elements (Equations 1-7) are sufficient for SCP trajectory re-optimization.
    Standard astrodynamics formulation; not re-derived. Used in the SCP dynamics model and the high-fidelity check adds perturbations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Design And Optimization Of Multi-rendezvous Manoeuvres Based On Reinforcement Learning And Convex Optimization." pith.science (2026). https://pith.science/paper/QADIKLP4

@misc{pith2026241111778,
  author       = {Pith},
  title        = {Pith review of: Design And Optimization Of Multi-rendezvous Manoeuvres Based On Reinforcement Learning And Convex Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QADIKLP4}},
  note         = {Machine review of arXiv:2411.11778}
}
read the original abstract

Optimizing space vehicle routing is crucial for critical applications such as on-orbit servicing, constellation deployment, and space debris de-orbiting. Multi-target Rendezvous presents a significant challenge in this domain. This problem involves determining the optimal sequence in which to visit a set of targets, and the corresponding optimal trajectories: this results in a demanding NP-hard problem. We introduce a framework for the design and refinement of multi-rendezvous trajectories based on heuristic combinatorial optimization and Sequential Convex Programming. Our framework is both highly modular and capable of leveraging candidate solutions obtained with advanced approaches and handcrafted heuristics. We demonstrate this flexibility by integrating an Attention-based routing policy trained with Reinforcement Learning to improve the performance of the combinatorial optimization process. We show that Reinforcement Learning approaches for combinatorial optimization can be effectively applied to spacecraft routing problems. We apply the proposed framework to the UARX Space OSSIE mission: we are able to thoroughly explore the mission design space, finding optimal tours and trajectories for a wide variety of mission scenarios.

Figures

Figures reproduced from arXiv: 2411.11778 by the authors.

Figure 1
Figure 1. UARX Space OSSIE OTV. Credit: UARX Space [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. ECI and LVLH reference frames. 4 Guidance Policies and Transfer Cost Estimation As discussed in Section 2, OSSIE is constrained to multi-revolution impulsive manoeuvres. For the pay￾load deployment mission scenario under consideration, OSSIE must achieve high accuracy SMA, inclination and phase convergence: RAAN targeting is not of interest to the current mission, as RAAN drift is very large for LEO orbits. This sec… view at source ↗
Figure 4
Figure 4. STSP solver architecture. In black: complex components (integrated using standardized interfaces) [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: On-time constraints scheme and warm￾starting procedure Note that this decision enforces the actuation windows based on the combinatorial solution and Hohmann transfers. For multiple arcs optimization, this might create a source of suboptimality. Nonetheless, for the si…
Figure 6
Figure 6. Figure 6: Best fuel mass achieved as a function of gen [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 8
Figure 8. Figure 8: Training validation reward curve (expressed [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Cost of 5000 optimized mission scenarios. Fuel mass, [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 11
Figure 11. Figure 11: SCP-tailored noncoplanar maneuver (cor￾responding to the Test Case 2 in [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 10
Figure 10. Figure 10: Fuel consumption as a function of number [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 12
Figure 12. Figure 12: Preliminary verification results of case 1. [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 36 canonical work pages

  1. [1]

    Evolving Solutions to TSP Variants for Ac- tive Space Debris Removal: Genetic and Evo- lutionary Computation Conference (GECCO),

    D. Izzo, I. Getzner, D. Hennes, and L. Simões, “Evolving Solutions to TSP Variants for Ac- tive Space Debris Removal: Genetic and Evo- lutionary Computation Conference (GECCO),” Proceedings of the 17th annual conference on Genetic and evolutionary computation (GECCO 2015), S. Silva, Ed., pp. 1207–1214, 2015, Pub- lisher: ACM Press, issn: 9781450334723. do...

    arXiv 2015

  2. [3]

    Evo- lutionary Optimization of Multirendezvous Im- pulsive Trajectories,

    L. Federici, A. Zavoli, and G. Colasurdo, “Evo- lutionary Optimization of Multirendezvous Im- pulsive Trajectories,” International Journal of Aerospace Engineering, vol. 2021, pp. 1–19, May

    work page 2021

  3. [4]

    Trajectory optimization for multi-target Active Debris Removal missions,

    L. Medioni et al., “Trajectory optimization for multi-target Active Debris Removal missions,” Advances in Space Research, Space Environment Management and Space Sustainability, vol. 72, no. 7, pp. 2801–2823, Oct. 2023,issn: 0273-1177. doi: 10.1016/j.asr.2022.12.013 . [Online]. Available: https://www.sciencedirect.com/ science / article / pii / S0273117722...

    work page doi:10.1016/j.asr.2022.12.013 2023

  4. [5]

    Large- scale object selection and trajectory planning for multi-target space debris removal missions,

    A. Barea, H. Urrutxua, and L. Cadarso, “Large- scale object selection and trajectory planning for multi-target space debris removal missions,” Acta Astronautica, vol. 170, pp. 289–301, May 2020, issn: 0094-5765. doi: 10 . 1016 / j . actaastro.2020.01.032 . [Online]. Available: https://www.sciencedirect.com/science/ article/pii/S0094576520300436 (visited o...

    work page 2020

  5. [6]

    Low-thrust rendezvous trajectory generation for multi-target active space debris removal using the RQ-Law,

    S. Narayanaswamy, B. Wu, P. Ludivig, F. Soboczenski, K. Venkataramani, and C. J. Damaren, “Low-thrust rendezvous trajectory generation for multi-target active space debris removal using the RQ-Law,”Advances in Space Research, vol. 71, no. 10, pp. 4276–4287, May 2023, issn: 0273-1177. doi: 10 . 1016 / j . asr . 2022.12.049.[Online].Available: https://www. ...

    work page 2023

  6. [7]

    Review of Ac- tive Space Debris Removal Methods,

    C. P. Mark and S. Kamath, “Review of Ac- tive Space Debris Removal Methods,” Space Policy, vol. 47, pp. 194–206, Feb. 2019,issn: 0265-9646. doi: 10.1016/j.spacepol.2018. 12 . 005. [Online]. Available: https : / / www . sciencedirect.com/science/article/pii/ S0265964618300110 (visited on 09/18/2024)

    work page doi:10.1016/j.spacepol.2018 2019

  7. [8]

    Active debris removal: Recent progress and cur- rent trends,

    C. Bonnal, J.-M. Ruault, and M.-C. Desjean, “Active debris removal: Recent progress and cur- rent trends,” Acta Astronautica, vol. 85, pp. 51– 60, Apr. 2013,issn: 0094-5765. doi: 10.1016/j. actaastro.2012.11.009 . [Online]. Available: https://www.sciencedirect.com/science/ article/pii/S0094576512004602 (visited on 09/18/2024)

    doi:10.1016/j 2013

  8. [9]

    On-Orbit Servicing Missions: Challenges and Solutions for Space- craft Operations,

    F. Sellmaier, T. Boge, J. Spurmann, S. Gully, T. Rupp, and F. Huber, “On-Orbit Servicing Missions: Challenges and Solutions for Space- craft Operations,” en, inSpaceOps 2010 Con- ference, Huntsville, Alabama: American Insti- tute of Aeronautics and Astronautics, Apr. 2010, p. 11, isbn: 978-1-62410-164-9. doi: 10.2514/6. 2010-2159. [Online]. Available:http...

    doi:10.2514/6 2010

Show all 52 references
  1. [11]

    Multi- orbit routing and scheduling of refuellable on- orbit servicing space robots,

    S. E. Sorenson and S. G. N. Pinkley, “Multi- orbit routing and scheduling of refuellable on- orbit servicing space robots,”Computers & In- dustrial Engineering, vol. 176, p. 108852, Feb. 2023, issn: 0360-8352. doi: 10 . 1016 / j . cie . 2022.108852.[Online].Available: https://...

    2023

  2. [12]

    A Survey on Low-Thrust Trajectory Optimization Approaches,

    D. Morante, M. Sanjurjo Rivo, and M. Soler, “A Survey on Low-Thrust Trajectory Optimization Approaches,” Aerospace, vol. 8, no. 3, 2021,issn: 2226-4310. doi: 10.3390/aerospace8030088 . [Online]. Available: https://www.mdpi.com/ 2226-4310/8/3/88

    2021 doi

  3. [13]

    Petropouloset al., GTOC9: Methods and Re- sults from the Jet Propulsion Laboratory Team , May 2017

    A. Petropouloset al., GTOC9: Methods and Re- sults from the Jet Propulsion Laboratory Team , May 2017

    2017

  4. [14]

    GTOC 11: Results found at Beijing Institute of Technology and China Academy of Space Technology,

    S. Luet al., “GTOC 11: Results found at Beijing Institute of Technology and China Academy of Space Technology,”Acta Astronautica, vol. 202, pp. 876–888, Jan. 2023,issn: 0094-5765. doi: 10.1016/j.actaastro.2022.07.039. [Online]. Available: https://www.sciencedirect.com/ science...

    2023 doi

  5. [15]

    Deep networks as approximators of optimal low-thrust and multi-impulse cost in multitarget missions,

    H. Li, S. Chen, D. Izzo, and H. Baoyin, “Deep networks as approximators of optimal low-thrust and multi-impulse cost in multitarget missions,” Acta Astronautica, vol. 166, pp. 469–481, Jan. 2020, ADS Bibcode: 2020AcAau.166..469L,issn: 0094-5765. doi: 10.1016/j.actaastro.2019. ...

    2020 doi

  6. [16]

    François, Q

    A. François, Q. Cappart, and L.-M. Rousseau, How to Evaluate Machine Learning Approaches for Combinatorial Optimization: Application to the Travelling Salesman Problem,Sep.2019.[On- line]. Available: https : / / arxiv . org / abs / 1909.13121

    2019 arXiv

  7. [17]

    GTOC 9, Multiple Space Debris Rendezvous Trajectory Design in the J2 environment,

    M. Hallmann et al., “GTOC 9, Multiple Space Debris Rendezvous Trajectory Design in the J2 environment,” inActa Futura 11, Jun. 2017, p. 9

    2017

  8. [18]

    A parallel global multi- objective framework for optimization: Pagmo,

    F. Biscani and D. Izzo, “A parallel global multi- objective framework for optimization: Pagmo,” en, Journal of Open Source Software , vol. 5, no. 53, p. 2338, Sep. 2020, issn: 2475-9066. doi: 10 . 21105 / joss . 02338. [Online]. Avail- able: https://joss.theoj.org/papers/10. 2...

    2020

  9. [19]

    Pymoo: Multi-Objective Optimization in Python,

    J. Blank and K. Deb, “Pymoo: Multi-Objective Optimization in Python,”IEEE Access, vol. 8, pp. 89497–89509, 2020, Conference Name: IEEE Access, issn: 2169-3536. doi: 10.1109/ ACCESS . 2020 . 2990567 . [Online]. Available: https : / / ieeexplore . ieee . org / document / 9078759...

    2020

  10. [20]

    Solving non-convex optimal con- trol problems by convex optimization,

    P. L. X. Liu, “Solving non-convex optimal con- trol problems by convex optimization,”Journal of Guidance, Control, and Dynamics , vol. 37, no. 3, pp. 750–765, 2014

    2014

  11. [21]

    Ecos: An socp solver for embedded systems,

    S. B. A. Domahidi E. Chu, “Ecos: An socp solver for embedded systems,” European Control Con- ference (ECC), 2013

    2013

  12. [22]

    Optimal guid- ance and control with nonlinear dynamics us- ing sequential convex programming,

    F. Y. H. R. Foust S. Chung, “Optimal guid- ance and control with nonlinear dynamics us- ing sequential convex programming,”Journal of Guidance Control & Dynamics , vol. 43, no. 4, pp. 633–644, 2020

    2020

  13. [23]

    Sequential convex program- ming for optimal line of sight steering in agile missions,

    L. H. J. Ramirez, “Sequential convex program- ming for optimal line of sight steering in agile missions,” 9th European Conference for Aero- nautics and Aerospace Sciences (EUCASS) , 2022

    2022

  14. [24]

    Rapid low-thrust trajectory optimization in deep space based on convex programming,

    F. T. C. Hofmann, “Rapid low-thrust trajectory optimization in deep space based on convex programming,” Journal of Guidance, Control, and Dynamics, 2021

    2021

  15. [25]

    S. J. W. J. Nocedal,Numerical Optimization. New York, NY, USA: Springer, 2nd edition, 2006

    2006

  16. [26]

    ESA, ESA Space Debris Mitigation Require- ments, English, Oct. 2023. [Online]. Available: https : / / technology . esa . int / upload / media / DGHKMZ _ 6542582e18e33 . pdf(visited on 09/23/2024)

    2023

  17. [27]

    Survey of Orbit Element Sets,

    G. Hintz, “Survey of Orbit Element Sets,”Jour- nal of Guidance Control and Dynamics - J GUID CONTROL DYNAM , vol. 31, pp. 785– 790, May 2008.doi: 10.2514/1.32237

    2008 doi

  18. [28]

    Chapter 2 - Fundamen- tal Astrodynamics,

    K. T. Alfriend, S. R. Vadali, P. Gurfil, J. P. How, and L. S. Breger, “Chapter 2 - Fundamen- tal Astrodynamics,” in Spacecraft Formation Flying, K. T. Alfriend, S. R. Vadali, P. Gur- fil, J. P. How, and L. S. Breger, Eds., Oxford: Butterworth-Heinemann, Jan. 2010, pp. 13–38, i...

    2010 doi

  19. [29]

    2015,isbn: 978-94-6186-419-

    K.Wakker, Fundamentals of Astrodynamics.TU Delft Library, Jan. 2015,isbn: 978-94-6186-419-

    2015

  20. [30]

    Sampling Permutations for Shapley Value Esti- mation,

    R. Mitchell, J. Cooper, E. Frank, and G. Holmes, “Sampling Permutations for Shapley Value Esti- mation,” Journal of Machine Learning Research, vol. 23, no. 43, pp. 1–46, 2022,issn: 1533-7928. 16 RL and SCP for MRTO Presented at IAC 2024 [Online]. Available:http://jmlr.org/pape...

    2022

  21. [31]

    Fisher-Yates shuffle,

    M. Eberl, “Fisher-Yates shuffle,” Arch. Formal Proofs , 2016. [Online]. Available: https : / / www . semanticscholar . org / paper / Fisher - Yates - shuffle - Eberl / 5e24ffebdc35e8e11af823505cbd5c6d5407f23e (visited on 09/19/2024)

    2016

  22. [32]

    Available: https : / / repository

    [Online]. Available: https : / / repository . tudelft . nl / islandora / object / uuid % 5C % 3A3fc91471-8e47-4215-af43-718740e6694e

  23. [33]

    Non-Null Ranking Models. I,

    C. L. Mallows, “Non-Null Ranking Models. I,” Biometrika, vol. 44, no. 1/2, pp. 114–130, 1957, Publisher: [Oxford University Press, Biometrika Trust], issn: 0006-3444. doi: 10.2307/2333244. [Online]. Available:https://www.jstor.org/ stable/2333244 (visited on 09/19/2024)

    1957

  24. [34]

    Sampling and learning distance- based probability models for permutation spaces,

    E. Irurozki, “Sampling and learning distance- based probability models for permutation spaces,” es, Ph.D. dissertation, Universidad del País Vasco - Euskal Herriko Unibertsitatea,

  25. [35]

    Group Representations in Proba- bility and Statistics,

    P. Diaconis, “Group Representations in Proba- bility and Statistics,”Lecture Notes-Monograph Series, vol. 11, pp. i–192, 1988, Publisher: Insti- tute of Mathematical Statistics,issn: 0749-2170. [Online]. Available:https://www.jstor.org/ stable/4355560 (visited on 09/19/2024)

    1988

  26. [36]

    Novel Random Key Encoding Schemes for the Dif- ferential Evolution of Permutation Problems,

    P. Krömer, V. Uher, and V. Snášel, “Novel Random Key Encoding Schemes for the Dif- ferential Evolution of Permutation Problems,” IEEE Transactions on Evolutionary Computa- tion, vol. 26, no. 1, pp. 43–57, Feb. 2022, Con- ference Name: IEEE Transactions on Evolu- tionary Comput...

    2022

  27. [37]

    Biased random-key genetic algorithms: A review,

    M. A. Londe, L. S. Pessoa, C. E. Andrade, and M. G. C. Resende, “Biased random-key genetic algorithms: A review,”European Journal of Op- erational Research, Mar. 2024,issn: 0377-2217. doi: 10.1016/j.ejor.2024.03.030 . [Online]. Available: https://www.sciencedirect.com/ science...

    2024 doi

  28. [38]

    The open-source astrodynamics Tudatpy software - overview for planetary mis- sion design and science analysis,

    D. Dirkxet al., “The open-source astrodynamics Tudatpy software - overview for planetary mis- sion design and science analysis,” en,European Planetary Science Congress , EPSC2022–253, Sep. 2022. doi: 10.5194/epsc2022-253 . [On- line]. Available:https://ui.adsabs.harvard. edu /...

    2022 doi

  29. [39]

    Genetic Algorithms and Random Keys for Sequencing and Optimization,

    J. C. Bean, “Genetic Algorithms and Random Keys for Sequencing and Optimization,”ORSA Journal on Computing, vol. 6, no. 2, pp. 154–160, May 1994, Publisher: ORSA,issn: 0899-1499. doi: 10.1287/ijoc.6.2.154 . [Online]. Avail- able: https : / / pubsonline . informs . org / doi / ...

    1994 doi

  30. [40]

    Inter- planetary transfers via deep representations of the optimal policy and/or of the value func- tion,

    D. Izzo, E. Öztürk, and M. Märtens, “Inter- planetary transfers via deep representations of the optimal policy and/or of the value func- tion,” in Proceedings of the Genetic and Evo- lutionary Computation Conference Companion , ser. GECCO ’19, event-place: Prague, Czech Republ...

    2019

  31. [41]

    Bertoet al., RL4CO: An Extensive Reinforce- ment Learning for Combinatorial Optimization Benchmark, arXiv:2306.17100 [cs], Jun

    F. Bertoet al., RL4CO: An Extensive Reinforce- ment Learning for Combinatorial Optimization Benchmark, arXiv:2306.17100 [cs], Jun. 2024. doi: 10.48550/arXiv.2306.17100 . [Online]. Available: http : / / arxiv . org / abs / 2306 . 17100 (visited on 09/18/2024)

    2024 doi arXiv

  32. [42]

    Atten- tion, Learn to Solve Routing Problems!

    W. Kool, H. v. Hoof, and M. Welling, “Atten- tion, Learn to Solve Routing Problems!” en, in International Conference on Learning Rep- resentations 2019, 2019, p. 25. [Online]. Avail- able: https : / / openreview . net / forum ? id = ByxBFsRqYm (visited on 07/24/2024)

    2019

  33. [43]

    A survey on artificial intelligence trends in spacecraft guid- ance dynamics and control,

    D. Izzo, M. Märtens, and B. Pan, “A survey on artificial intelligence trends in spacecraft guid- ance dynamics and control,” en,Astrodynamics, vol. 3, no. 4, pp. 287–299, Dec. 2019,issn: 2522-

    2019

  34. [44]

    Actor-Critic Algorithms,

    V. Konda and J. Tsitsiklis, “Actor-Critic Algorithms,” in NeurIPS Proceedings, vol. 12, MIT Press, 1999, p. 7. [Online]. Avail- able: https : / / proceedings . neurips . cc / paper _ files / paper / 1999 / hash / 6449f44a102fde848669bdd9eb6b76fa - Abstract.html (visited on 09/20/2024)

    1999

  35. [45]

    Mnih et al., Asynchronous Methods for Deep Reinforcement Learning, arXiv:1602.01783 [cs], Jun

    V. Mnih et al., Asynchronous Methods for Deep Reinforcement Learning, arXiv:1602.01783 [cs], Jun. 2016. doi: 10.48550/arXiv.1602.01783. [Online]. Available: http://arxiv.org/abs/ 1602.01783 (visited on 09/20/2024)

    2016 doi arXiv

  36. [46]

    Schulman, F

    J. Schulman, F. Wolski, P. Dhariwal, A. Rad- ford, and O. Klimov, Proximal Policy Opti- mization Algorithms, arXiv:1707.06347 [cs], Aug

    arXiv

  37. [47]

    T. M. Inc., Simulation and Model-Based De- sign, Natick, Massachusetts, United States, 2023. [Online]. Available:https://mathworks.com/ products/simulink.html

    2023

  38. [48]

    Simple statistical gradient- following algorithms for connectionist reinforce- ment learning,

    R. J. Williams, “Simple statistical gradient- following algorithms for connectionist reinforce- ment learning,” en, Machine Learning, vol. 8, no. 3, pp. 229–256, May 1992,issn: 1573-0565. doi: 10.1007/BF00992696. [Online]. Available: https://doi.org/10.1007/BF00992696 (vis- it...

    1992 doi

  39. [54]

    T. M. Inc.,MATLAB version: 23.2.0 (R2023b) , Natick, Massachusetts, United States, 2023. [On- line]. Available:https://www.mathworks.com. 18

    2023

  40. [98]

    [On- line]

    doi: 10.1007/s42064-018-0053-6. [On- line]. Available: https://doi.org/10.1007/ s42064-018-0053-6 (visited on 09/18/2024)

    2024 doi

  41. [2014]

    Available: https : / / dialnet

    [Online]. Available: https : / / dialnet . unirioja . es / servlet / tesis ? codigo = 213087 (visited on 09/19/2024)

    2024

  42. [2017]

    [On- line]

    doi: 10.48550/arXiv.1707.06347. [On- line]. Available:http://arxiv.org/abs/1707. 06347 (visited on 09/20/2024). 17 RL and SCP for MRTO Presented at IAC 2024

    2024 doi arXiv

  43. [2021]

    doi: 10.1155/2021/9921555

    2021 doi

  44. [4650]

    [Online]

    doi: 10.2514/1.A35094. [Online]. Avail- able: https://doi.org/10.2514/1.A35094 (visited on 09/18/2024)

    2024 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.