pith. sign in

arxiv: 2606.06011 · v1 · pith:3NW764V4new · submitted 2026-06-04 · 💻 cs.RO · cs.LG· cs.MA

Merging model-based control with multi-agent reinforcement learning for multi-agent cooperative teaming strategies

Christian Llanes , Spencer W. Jensen , Samuel Coogan This is my paper

Pith reviewed 2026-06-28 01:45 UTC · model grok-4.3

classification 💻 cs.RO cs.LGcs.MA
keywords multi-agent reinforcement learningmodel predictive controlcooperative multi-agent systemshardware experimentspursuit-evasionlanding task
0
0 comments X

The pith

Merging multi-agent reinforcement learning with model-predictive control produces safe cooperative actions that reach 100 percent hardware success where pure neural policies reach only 60 percent.

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

The paper establishes that extending actor-critic model-predictive control to the multi-agent setting yields actions that remain both cooperative over long horizons and dynamically feasible over short replanning cycles. It demonstrates the result in a pursuit-evasion task against an advanced adversarial pursuer and in a heterogeneous drone-rover landing task. The approach matters because learned multi-agent policies alone often produce unsafe or infeasible commands, while pure model-based methods struggle with long-horizon coordination from discrete rewards. A reader would therefore see a practical route to deploying learned team strategies on physical robots.

Core claim

The multi-agent actor-critic model predictive control (MA-AC-MPC) algorithm couples a learned multi-agent policy with short-horizon model-predictive control to generate safe, dynamically feasible actions; in hardware it achieves a 100 percent success rate on a cooperative drone-rover landing task versus 60 percent for the multi-layer perceptron baseline, and it maintains performance against augmented proportional navigation in pursuit-evasion scenarios.

What carries the argument

The MA-AC-MPC algorithm, which places the learned multi-agent policy inside a receding-horizon model-predictive controller that replans feasible trajectories at each step.

If this is right

  • Actions generated by MA-AC-MPC remain both cooperative across long horizons and dynamically feasible within short replanning windows.
  • The same controller succeeds against established adversarial laws such as augmented proportional navigation.
  • Heterogeneous teams achieve repeatable, successful hardware landings without further policy retraining.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The separation of long-horizon cooperation learning from short-horizon safety enforcement may reduce the reward-shaping burden typical in pure MARL.
  • The same coupling structure could be tested on additional cooperative tasks such as formation flight or joint manipulation where dynamic constraints are tight.

Load-bearing premise

A short-horizon model-predictive controller can be stably coupled to the learned multi-agent policy without introducing instability or requiring extensive additional tuning.

What would settle it

A hardware trial in which the MA-AC-MPC controller produces control inputs that violate vehicle dynamics or cause collisions during the landing maneuver.

Figures

Figures reproduced from arXiv: 2606.06011 by Christian Llanes, Samuel Coogan, Spencer W. Jensen.

Figure 1
Figure 1. Figure 1: Screenshots of both a homogeneous and heterogeneous multi-agent environment where multi-agent actor-critic model predictive control (MA-AC-MPC) is implemented. The top image is a screenshot at the final step of a two pursuer versus two evader environment trajectory before the pursuers collide into each other. The bottom screenshot is the final step of a drone landing on a rover environment where both agent… view at source ↗
Figure 2
Figure 2. Figure 2: Architecture diagram for multi-agent actor-critic model predictive control (MA-AC-MPC). externally. However, this approach scales poorly with the number of agents as the number of decision variables is increased and is not effective for autonomous environments where a centralized computer is required. Additionally, this approach requires reliable communications as control actions are published through the … view at source ↗
Figure 4
Figure 4. Figure 4: Evader win rate for independent evader mass variations for both MA-AC-MPC [256×2 / 256×2] and MA-AC-MLP [512×2 / 512×2]. Both models are trained with a nominal mass of 40.6 g. The mass is only changed within the environment and not the MPC dynamics which uses the nominal mass. The third plot is the difference in win rate between both methods where darker red implies MA-AC-MPC win rate is much higher than M… view at source ↗
Figure 3
Figure 3. Figure 3: Training curves for mean total reward, evader win rate, and curriculum level for various MA-AC-MLP actor and critic neural network sizes and the proposed MA-AC-MPC. Curriculum level advancement occurs when the evader win rate reaches a threshold of 70%. AMD Ryzen 9 5950x desktop processor and tabulate the training time and inference time in [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 5
Figure 5. Figure 5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Landing rate for MA-AC-MPC and MA-AC-MLP during training. Solid curves show the smoothed landing rate, faint curves show the raw logged values, and dashed vertical lines indicate curriculum level advancements. commands rather than direct wheel-speed commands. The model is discretized with RK4 inside the MPC. The wheel limits are enforced as linear inequality constraints on the commanded body velocity: −ωma… view at source ↗
Figure 8
Figure 8. Figure 8: Three-dimensional landing trajectories for the two policies across five hardware trials, with MA-AC-MPC shown on top and MA-AC-MLP on the bottom. In each plot, the drone trajectory is drawn with solid lines and the rover trajectory with dashed lines, while endpoint markers indicate landing outcome. MA-AC-MPC exhibits consistently tight convergence to the rover landing pad across all trials, whereas MA-AC-M… view at source ↗
read the original abstract

In this work, we propose a framework that combines multi-agent reinforcement learning (MARL) with model-based control to achieve safe, dynamically feasible actions in cooperative multi-agent tasks. Multi-agent reinforcement learning provides the advantage of learning cooperative policies for multi-agent teams from discrete non-differentiable rewards in a long planning horizon. Model-predictive control is robust and offers safe, dynamically feasible actions in a fast replanning framework for short horizons. We propose an algorithm that extends actor-critic model predictive control for MARL which we refer to as multi-agent actor-critic model predictive control (MA-AC-MPC). We demonstrate the capabilities of this algorithm by applying it to a multi-agent pursuit-evasion scenario. Specifically, we compare the evader team's strategy using the MA-AC-MPC model and a multi-layer perceptron model (MA-AC-MLP). The pursuer team uses augmented proportional navigation as it is accepted as an advanced adversarial control law. We also provide an example with a heterogeneous environment where a drone and omni-wheeled rover cooperate to achieve repeatable and successful landing with 100% success rate in hardware for MA-AC-MPC compared to 60% for MA-AC-MLP. We demonstrate the robustness of the proposed MA-AC-MPC algorithm in hardware for both environments.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The paper proposes MA-AC-MPC, an extension of actor-critic model-predictive control to multi-agent reinforcement learning settings. It combines MARL for learning cooperative long-horizon policies from non-differentiable rewards with short-horizon MPC for generating safe, dynamically feasible actions. The approach is demonstrated in a multi-agent pursuit-evasion scenario (evader team vs. augmented proportional navigation pursuers) and a heterogeneous drone-rover landing task, where MA-AC-MPC achieves 100% hardware success compared to 60% for MA-AC-MLP.

Significance. If the coupling between the learned multi-agent policy and the MPC layer proves stable under model mismatch and varying replanning rates, the framework could offer a practical route to safe cooperative behaviors in robotics without requiring fully differentiable rewards or long-horizon optimization. The hardware demonstration in a heterogeneous team is a positive step, but the absence of supporting analysis limits assessment of generality.

major comments (2)
  1. [Abstract] Abstract: The central claim that MA-AC-MPC produces safe, dynamically feasible cooperative actions rests on the 100% vs. 60% hardware landing success comparison, yet the text supplies no number of trials, variance, or statistical test; without these, the performance difference cannot be distinguished from tuning or sampling effects.
  2. [Abstract] Abstract and hardware results: No analysis is provided on the stability of feeding the actor-critic output into the short-horizon MPC (e.g., sensitivity to model mismatch, replanning frequency, or closed-loop eigenvalues), which directly governs whether the reported success generalizes beyond the specific tested conditions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and indicate the planned revisions.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that MA-AC-MPC produces safe, dynamically feasible cooperative actions rests on the 100% vs. 60% hardware landing success comparison, yet the text supplies no number of trials, variance, or statistical test; without these, the performance difference cannot be distinguished from tuning or sampling effects.

    Authors: We agree that the hardware results require additional statistical detail to support the reported success rates. In the revised manuscript we will specify the number of trials conducted for each method, report observed variance or standard deviation, and include a statistical test (e.g., two-proportion z-test) to evaluate whether the 100% versus 60% difference is significant. These details will be added to the abstract and the hardware results section. revision: yes

  2. Referee: [Abstract] Abstract and hardware results: No analysis is provided on the stability of feeding the actor-critic output into the short-horizon MPC (e.g., sensitivity to model mismatch, replanning frequency, or closed-loop eigenvalues), which directly governs whether the reported success generalizes beyond the specific tested conditions.

    Authors: We acknowledge that the manuscript lacks a dedicated analysis of the learned policy to MPC interface. While the hardware demonstrations show reliable performance under the tested conditions, we agree this restricts assessment of broader applicability. In revision we will add simulation-based sensitivity studies on model mismatch and replanning frequency, together with a qualitative discussion of observed stability margins. A full closed-loop eigenvalue analysis lies outside the present scope and will not be included. revision: partial

Circularity Check

0 steps flagged

No circularity; empirical validation of MA-AC-MPC extension stands independent of inputs.

full rationale

The paper proposes MA-AC-MPC as an extension of actor-critic MPC to the multi-agent setting and validates it via hardware experiments (100% landing success for MA-AC-MPC vs 60% for MA-AC-MLP). No equations, fitted parameters, or self-citations are presented that reduce the performance claims to quantities defined by the authors' own inputs or prior results. The derivation consists of an algorithmic combination followed by direct empirical comparison; no self-definitional, fitted-input-called-prediction, or uniqueness-imported patterns appear. The central claims rest on observable hardware outcomes rather than any closed mathematical loop.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no explicit free parameters, axioms, or invented entities can be extracted.

pith-pipeline@v0.9.1-grok · 5770 in / 996 out tokens · 33450 ms · 2026-06-28T01:45:34.386121+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

55 extracted references · 8 canonical work pages

  1. [1]

    A Comprehensive Survey of Multiagent Reinforcement Learning , year=

    Busoniu, Lucian and Babuska, Robert and De Schutter, Bart , journal=. A Comprehensive Survey of Multiagent Reinforcement Learning , year=

  2. [2]

    Development of an industrial Internet of Things (IIoT) based smart robotic warehouse management system , author=

  3. [3]

    and Katz, Benjamin and Di Carlo, Jared and Wensing, Patrick M

    Bledt, Gerardo and Powell, Matthew J. and Katz, Benjamin and Di Carlo, Jared and Wensing, Patrick M. and Kim, Sangbae , booktitle=. MIT Cheetah 3: Design and Control of a Robust, Dynamic Quadruped Robot , year=

  4. [4]

    Champion-level drone racing using deep reinforcement learning , journal=

    Kaufmann, Elia and Bauersfeld, Leonard and Loquercio, Antonio and M. Champion-level drone racing using deep reinforcement learning , journal=. 2023 , month=. doi:10.1038/s41586-023-06419-4 , url=

    work page doi:10.1038/s41586-023-06419-4 2023

  5. [5]

    Toward a Fully Autonomous UAV: Research Platform for Indoor and Outdoor Urban Search and Rescue , year=

    Tomic, Teodor and Schmid, Korbinian and Lutz, Philipp and Domel, Andreas and Kassecker, Michael and Mair, Elmar and Grixa, Iris Lynne and Ruess, Felix and Suppa, Michael and Burschka, Darius , journal=. Toward a Fully Autonomous UAV: Research Platform for Indoor and Outdoor Urban Search and Rescue , year=

  6. [6]

    Samvelyan, Mikayel and Rashid, Tabish and Schroeder de Witt, Christian and Farquhar, Gregory and Nardelli, Nantas and Rudner, Tim G. J. and Hung, Chia-Man and Torr, Philip H. S. and Foerster, Jakob and Whiteson, Shimon , title =. Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems , pages =. 2019 , isbn =

    2019

  7. [7]

    Stanley: The Robot That Won the DARPA Grand Challenge

    Thrun, Sebastian and Montemerlo, Mike and Dahlkamp, Hendrik and Stavens, David and Aron, Andrei and Diebel, James and Fong, Philip and Gale, John and Halpenny, Morgan and Hoffmann, Gabriel and Lau, Kenny and Oakley, Celia and Palatucci, Mark and Pratt, Vaughan and Stang, Pascal and Strohband, Sven and Dupont, Cedric and Jendrossek, Lars-Erik and Koelen, C...

    2005

  8. [8]

    and Blackmore, Lars , journal=

    Açıkmeşe, Behçet and Carson, John M. and Blackmore, Lars , journal=. Lossless Convexification of Nonconvex Control Bound and Pointing Constraints of the Soft Landing Optimal Control Problem , year=

  9. [9]

    IEEE Transactions on Robotics , year=

    Actor-Critic Model Predictive Control: Differentiable Optimization meets Reinforcement Learning for Agile Flight , author=. IEEE Transactions on Robotics , year=

  10. [10]

    and Wu, Xinzhou , booktitle=

    Lubars, Joseph and Gupta, Harsh and Chinchali, Sandeep and Li, Liyun and Raja, Adnan and Srikant, R. and Wu, Xinzhou , booktitle=. Combining Reinforcement Learning with Model Predictive Control for On-Ramp Merging , year=

  11. [11]

    2025 , eprint=

    Synthesis of Model Predictive Control and Reinforcement Learning: Survey and Classification , author=. 2025 , eprint=

    2025

  12. [12]

    Safe Reinforcement Learning Using Robust MPC , year=

    Zanon, Mario and Gros, Sebastien , journal=. Safe Reinforcement Learning Using Robust MPC , year=

  13. [13]

    2024 , eprint=

    DeepSafeMPC: Deep Learning-Based Model Predictive Control for Safe Multi-Agent Reinforcement Learning , author=. 2024 , eprint=

    2024

  14. [14]

    Zico , title =

    Amos, Brandon and Rodriguez, Ivan Dario Jimenez and Sacks, Jacob and Boots, Byron and Kolter, J. Zico , title =. Proceedings of the 32nd International Conference on Neural Information Processing Systems , pages =. 2018 , publisher =

    2018

  15. [15]

    Proceedings of the 36th International Conference on Neural Information Processing Systems , articleno =

    Yu, Chao and Velu, Akash and Vinitsky, Eugene and Gao, Jiaxuan and Wang, Yu and Bayen, Alexandre and Wu, Yi , title =. Proceedings of the 36th International Conference on Neural Information Processing Systems , articleno =. 2022 , isbn =

    2022

  16. [16]

    and Drew, Daniel S

    Lambert, Nathan O. and Drew, Daniel S. and Yaconelli, Joseph and Levine, Sergey and Calandra, Roberto and Pister, Kristofer S. J. , journal=. Low-Level Control of a Quadrotor With Deep Model-Based Reinforcement Learning , year=

  17. [17]

    and Yuan, Zhaocong and Zhou, Siqi and Panerati, Jacopo and Schoellig, Angela P

    Brunke, Lukas and Greeff, Melissa and Hall, Adam W. and Yuan, Zhaocong and Zhou, Siqi and Panerati, Jacopo and Schoellig, Angela P. Safe Learning in Robotics: From Learning-Based Control to Safe Reinforcement Learning. Annual Review of Control, Robotics, and Autonomous Systems. 2022. doi:https://doi.org/10.1146/annurev-control-042920-020211

    work page doi:10.1146/annurev-control-042920-020211 2022

  18. [18]

    Proceedings of the 6th Annual Learning for Dynamics & Control Conference , pages =

    Hoffmann, Jasper and Clausen, Diego Fernandez and Brosseit, Julien and Bernhard, Julian and Esterle, Klemens and Werling, Moritz and Karg, Michael and B\". Proceedings of the 6th Annual Learning for Dynamics & Control Conference , pages =. 2024 , editor =

    2024

  19. [19]

    Learning When to Trust the Expert for Guided Exploration in

    Felix Schulz and Jasper Hoffmann and Yuan Zhang and Joschka Boedecker , booktitle=. Learning When to Trust the Expert for Guided Exploration in. 2024 , url=

    2024

  20. [20]

    A Painless Deterministic Policy Gradient Method for Learning-based MPC , year=

    Anand, Akhil S and Reinhardt, Dirk and Sawant, Shambhuraj and Gravdahl, Jan Tommy and Gros, Sebastien , booktitle=. A Painless Deterministic Policy Gradient Method for Learning-based MPC , year=

  21. [21]

    Imitation Learning from Nonlinear MPC via the Exact Q-Loss and its Gauss-Newton Approximation , year=

    Ghezzi, Andrea and Hoffman, Jasper and Frey, Jonathan and Boedecker, Joschka and Diehl, Moritz , booktitle=. Imitation Learning from Nonlinear MPC via the Exact Q-Loss and its Gauss-Newton Approximation , year=

  22. [22]

    Policy Search for Model Predictive Control With Application to Agile Drone Flight , year=

    Song, Yunlong and Scaramuzza, Davide , journal=. Policy Search for Model Predictive Control With Application to Agile Drone Flight , year=

  23. [23]

    Shankar Sastry and Claire Tomlin , keywords =

    Anil Aswani and Humberto Gonzalez and S. Shankar Sastry and Claire Tomlin , keywords =. Provably safe and robust learning-based model predictive control , journal =. 2013 , issn =. doi:https://doi.org/10.1016/j.automatica.2013.02.003 , url =

    work page doi:10.1016/j.automatica.2013.02.003 2013

  24. [24]

    Predictive Control with Learning-Based Terminal Costs Using Approximate Value Iteration , journal =

    Francisco Moreno-Mora and Lukas Beckenbach and Stefan Streif , keywords =. Predictive Control with Learning-Based Terminal Costs Using Approximate Value Iteration , journal =. 2023 , note =. doi:https://doi.org/10.1016/j.ifacol.2023.10.1320 , url =

    work page doi:10.1016/j.ifacol.2023.10.1320 2023

  25. [25]

    and Diehl, Moritz , journal=

    Reiter, Rudolf and Ghezzi, Andrea and Baumgärtner, Katrin and Hoffmann, Jasper and McAllister, Robert D. and Diehl, Moritz , journal=. AC4MPC: Actor-Critic Reinforcement Learning for Guiding Model Predictive Control , year=

  26. [26]

    DiffTune-MPC: Closed-Loop Learning for Model Predictive Control , year=

    Tao, Ran and Cheng, Sheng and Wang, Xiaofeng and Wang, Shenlong and Hovakimyan, Naira , journal=. DiffTune-MPC: Closed-Loop Learning for Model Predictive Control , year=

  27. [27]

    Annual Conference on Learning for Dynamics and Control , author =

    Safe Reinforcement Learning with Chance-constrained Model Predictive Control: , url =. Annual Conference on Learning for Dynamics and Control , author =

  28. [28]

    Proceedings of Robotics: Science and Systems , YEAR =

    Alex Oshin AND Hassan Almubarak AND Evangelos Theodorou , TITLE =. Proceedings of Robotics: Science and Systems , YEAR =

  29. [29]

    Proceedings of the 31st International Conference on Neural Information Processing Systems , pages =

    Lowe, Ryan and Wu, Yi and Tamar, Aviv and Harb, Jean and Abbeel, Pieter and Mordatch, Igor , title =. Proceedings of the 31st International Conference on Neural Information Processing Systems , pages =. 2017 , isbn =

    2017

  30. [30]

    Mallick and F

    S. Mallick and F. Airaldi and A. Dabiri and B. Multi-agent reinforcement learning via distributed. Automatica , volume=. 2024 , doi=

    2024

  31. [31]

    2024 , eprint=

    An Introduction to Centralized Training for Decentralized Execution in Cooperative Multi-Agent Reinforcement Learning , author=. 2024 , eprint=

    2024

  32. [32]

    Multi-agent deep reinforcement learning: A survey.Artificial Intelligence Review, 55(2):895–943, 2022

    Gronauer, Sven and Diepold, Klaus , title=. Artificial Intelligence Review , year=. doi:10.1007/s10462-021-09996-w , url=

    work page doi:10.1007/s10462-021-09996-w

  33. [33]

    and Amato, Christopher , title =

    Oliehoek, Frans A. and Amato, Christopher , title =. 2016 , isbn =

    2016

  34. [34]

    2024 , eprint=

    Fully Decentralized Cooperative Multi-Agent Reinforcement Learning: A Survey , author=. 2024 , eprint=

    2024

  35. [35]

    Santos and Diogo S

    Pedro P. Santos and Diogo S. Carvalho and Miguel Vasco and Alberto Sardinha and Pedro A. Santos and Ana Paiva and Francisco S. Melo , keywords =. Centralized training with hybrid execution in multi-agent reinforcement learning via predictive observation imputation , journal =. 2025 , issn =. doi:https://doi.org/10.1016/j.artint.2025.104404 , url =

    work page doi:10.1016/j.artint.2025.104404 2025

  36. [36]

    Proceedings of the 30th International Conference on Neural Information Processing Systems , pages =

    Sukhbaatar, Sainbayar and Szlam, Arthur and Fergus, Rob , title =. Proceedings of the 30th International Conference on Neural Information Processing Systems , pages =. 2016 , isbn =

    2016

  37. [37]

    and Assael, Yannis M

    Foerster, Jakob N. and Assael, Yannis M. and de Freitas, Nando and Whiteson, Shimon , title =. Proceedings of the 30th International Conference on Neural Information Processing Systems , pages =. 2016 , isbn =

    2016

  38. [38]

    and Egorov, Maxim and Kochenderfer, Mykel

    Gupta, Jayesh K. and Egorov, Maxim and Kochenderfer, Mykel. Cooperative Multi-agent Control Using Deep Reinforcement Learning. Autonomous Agents and Multiagent Systems. 2017

    2017

  39. [39]

    2025 , eprint=

    Differentiable Nonlinear Model Predictive Control , author=. 2025 , eprint=

    2025

  40. [40]

    2022 , eprint=

    Differentiable Optimal Control via Differential Dynamic Programming , author=. 2022 , eprint=

    2022

  41. [41]

    2025 , eprint=

    Differentiable Model Predictive Control on the GPU , author=. 2025 , eprint=

    2025

  42. [42]

    2021 , eprint=

    Pontryagin Differentiable Programming: An End-to-End Learning and Control Framework , author=. 2021 , eprint=

    2021

  43. [43]

    Leveraging Proximal Optimization for Differentiating Optimal Control Solvers , year=

    Bounou, Oumayma and Ponce, Jean and Carpentier, Justin , booktitle=. Leveraging Proximal Optimization for Differentiating Optimal Control Solvers , year=

  44. [44]

    Mathematical Programming Computation , Year =

    acados -- a modular open-source framework for fast embedded optimal control , Author =. Mathematical Programming Computation , Year =

  45. [45]

    Fast integrators with sensitivity propagation for use in

    Frey, Jonathan and De Schutter, Jochem and Diehl, Moritz , Booktitle = ECC, Year =. Fast integrators with sensitivity propagation for use in

  46. [46]

    doi:10.5281/zenodo.17244101 , url =

    Leonard Fichtner and dirkpr and JasperHoffmann and Filippo Airaldi and Jonathan Frey and Josip Kir Hromatko and Katrin Baumgaertner and Mazen Amria and RudolfReiter and Shambhuraj Sawant , title =. doi:10.5281/zenodo.17244101 , url =

    work page doi:10.5281/zenodo.17244101

  47. [47]

    and Jensen, Spencer W

    Llanes, Christian and Williams, Kyle A. and Jensen, Spencer W. and Coogan, Samuel , title =. To appear in Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) , year =

  48. [48]

    2018 , eprint=

    High-Dimensional Continuous Control Using Generalized Advantage Estimation , author=. 2018 , eprint=

    2018

  49. [49]

    skrl: Modular and Flexible Library for Reinforcement Learning , journal =

    Antonio Serrano. skrl: Modular and Flexible Library for Reinforcement Learning , journal =. 2023 , volume =

    2023

  50. [50]

    2009 , eprint=

    Multi-Agent Model Predictive Control: A Survey , author=. 2009 , eprint=

    2009

  51. [51]

    and Hua, Yufei and Goudar, Abhishek and Zhou, SiQi and Schoellig, Angela P

    Schuck, Martin and Rath, Marcel P. and Hua, Yufei and Goudar, Abhishek and Zhou, SiQi and Schoellig, Angela P. , title =. 2026 , note =

    2026

  52. [52]

    Preiss* and Wolfgang H\"onig* and Gaurav S

    James A. Preiss* and Wolfgang H\"onig* and Gaurav S. Sukhatme and Nora Ayanian , title =. 2017 , url =. doi:10.1109/ICRA.2017.7989376 , note =

    work page doi:10.1109/icra.2017.7989376 2017

  53. [53]

    CrazySim: A Software-in-the-Loop Simulator for the Crazyflie Nano Quadrotor , year=

    Llanes, Christian and Kakish, Zahi and Williams, Kyle and Coogan, Samuel , booktitle=. CrazySim: A Software-in-the-Loop Simulator for the Crazyflie Nano Quadrotor , year=

  54. [54]

    N. A. Shneydor , title =

  55. [55]

    Paul Zarchan , title =