pith. machine review for the scientific record. sign in

arxiv: 2604.20862 · v1 · submitted 2026-03-20 · 💻 cs.AI · cs.MA

Recognition: no theorem link

Architecture of an AI-Based Automated Course of Action Generation System for Military Operations

Ji-il Park , Inwook Shim , Chong Hui Kim
Authors on Pith no claims yet

Pith reviewed 2026-05-15 08:28 UTC · model grok-4.3

classification 💻 cs.AI cs.MA
keywords course of action planningAI automationmilitary operationssystem architecturedecision support systemsdefense technology
0
0 comments X

The pith

This paper proposes an architecture for an AI-automated course of action planning system by matching public military doctrines with applicable AI technologies at each stage.

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

The paper seeks to introduce publicly available military doctrines and match them with suitable AI technologies for every step in the course of action planning process. It does so because traditional human-based planning struggles with the speed, range, and scale of modern military operations. If successful, the proposed architecture would enable the creation of an automated system that generates courses of action more rapidly. This matters for defense organizations looking to adapt to future warfare requirements without full reliance on classified details.

Core claim

By introducing relevant doctrines within the scope of publicly available information and presenting applicable AI technologies for each stage of the CoA planning process, an architecture for an automated CoA planning system can be proposed.

What carries the argument

The proposed system architecture that combines doctrines and AI technologies for automated CoA generation.

If this is right

  • Traditional manned CoA planning becomes increasingly challenging as operational areas expand.
  • AI-based systems are necessary for future warfare to match increasing maneuver speeds and weapon ranges.
  • Applicable AI technologies can be identified and applied to specific stages of the planning process.

Where Pith is reading between the lines

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

  • Integration with real-time surveillance data could enhance the system's responsiveness beyond the basic architecture.
  • Validation through unclassified simulations might reveal gaps in using only public doctrines.
  • Similar modular AI architectures could be adapted for civilian emergency planning scenarios.

Load-bearing premise

Publicly available doctrines and generic AI technologies suffice to build a functional, secure automated system without needing classified constraints or additional validation.

What would settle it

Building and testing the proposed architecture in an open simulation environment using only publicly available doctrines, then checking if it produces viable courses of action compared to standard manual methods.

Figures

Figures reproduced from arXiv: 2604.20862 by Chong Hui Kim, Inwook Shim, Ji-il Park.

Figure 2
Figure 2. Figure 2: Intelligence Preparation of the Battlefield (IPB) Process and Major [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Concept of differential implementation of echelon-specific wargaming [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Mission Analysis via LLM-Based Operational Plan Interpretation and [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Generation of multiple candidate courses of action using artificial intelligence. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: AI-Based Wargaming Simulation for Evaluating Multiple Courses of Action Using Quantitative Metrics. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Battlespace Evaluation for Defining the Area of Operations (AO) and [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Battlespace Analysis Based on Terrain Analysis [32]. [PITH_FULL_IMAGE:figures/full_fig_p007_9.png] view at source ↗
Figure 7
Figure 7. Figure 7: Representative AI Techniques Applicable to Enemy Infiltration [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 10
Figure 10. Figure 10: Battlespace Analysis Using Terrain and Weather Information. [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: AI-Enabled Enemy Capability Assessment in IPB Step 3 Integrating Doctrinal Patterns and Real-Time Enemy Observations. [PITH_FULL_IMAGE:figures/full_fig_p009_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Development of Feasible Enemy CoAs and Comparative Analysis [PITH_FULL_IMAGE:figures/full_fig_p009_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: AI-Enabled Multi-Modal IPB and Wargaming Architecture for Updating Enemy and Friendly Courses of Action. [PITH_FULL_IMAGE:figures/full_fig_p010_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: AI-Driven IPB Workflow Across Steps 1–4 from Battlespace Analysis to Enemy CoA Estimation. [PITH_FULL_IMAGE:figures/full_fig_p010_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: AI-Driven IPB-Based Decision-Making Framework for Course of Action Development and Execution. [PITH_FULL_IMAGE:figures/full_fig_p011_15.png] view at source ↗
read the original abstract

The automation system for Course of Action (CoA) planning is an essential element in future warfare. As maneuver speeds increase, surveillance ranges extend, and weapon ranges grow, the operational area expands, making traditional manned-based CoA planning increasingly challenging. Consequently, the development of an AI-based automated CoA planning system is becoming increasingly necessary. Accordingly, several countries and defense organizations are actively developing AI-based CoA planning systems. However, due to security restrictions and limited public disclosure, the technical maturity of such systems remains difficult to assess. Furthermore, as these systems are military-related, their details are not publicly disclosed, making it difficult to accurately assess the current level of development. In response to this, this study aims to introduce relevant doctrines within the scope of publicly available information and present applicable AI technologies for each stage of the CoA planning process. Ultimately, it proposes an architecture for the development of an automated CoA planning system.

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 / 3 minor

Summary. The manuscript reviews publicly available military doctrines on Course of Action (CoA) planning, maps generic AI technologies (such as machine learning and optimization methods) to each stage of the CoA process, and proposes a high-level architecture for an AI-based automated CoA generation system intended to address challenges from expanded operational tempos in future warfare.

Significance. If the mapping and architecture hold as a coherent conceptual framework, the work could serve as a useful public-domain reference point for researchers exploring AI integration into military decision support, by organizing open doctrines and indicating potential technology insertion points without claiming deployability.

major comments (2)
  1. [Proposed Architecture] The architecture proposal (final section) presents component blocks and data flows at a conceptual level only, without specifying interfaces, data schemas, or latency requirements between modules; this omission is load-bearing because the central claim is that such an architecture can be assembled from public doctrines and generic AI techniques.
  2. [Conclusion and Future Work] No evaluation criteria, metrics, or validation approach (e.g., simulation benchmarks against human planners) are defined for the proposed system; this weakens the claim that the architecture addresses the necessity driven by increasing maneuver speeds, as feasibility remains untestable within the manuscript.
minor comments (3)
  1. [Proposed Architecture] Figure 1 (architecture diagram) uses generic block labels without legends or example data flows, reducing clarity for readers unfamiliar with military planning terminology.
  2. [AI Technologies for CoA Stages] Several AI technique descriptions (e.g., reinforcement learning for CoA evaluation) cite only broad surveys rather than specific military-relevant implementations; adding 2-3 targeted references would strengthen the mapping.
  3. [Abstract] The abstract states the goal as proposing an architecture but does not explicitly note the absence of implementation or empirical results, which could set clearer reader expectations.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the scope and limitations of our conceptual architecture proposal. We address each major comment below and have revised the manuscript accordingly to improve transparency without altering its public-domain, high-level focus.

read point-by-point responses

  1. Referee: [Proposed Architecture] The architecture proposal (final section) presents component blocks and data flows at a conceptual level only, without specifying interfaces, data schemas, or latency requirements between modules; this omission is load-bearing because the central claim is that such an architecture can be assembled from public doctrines and generic AI techniques.

    Authors: We agree the architecture remains at a conceptual level. Detailed interfaces, data schemas, and latency requirements cannot be specified from publicly available doctrines alone, as these would require classified implementation details or proprietary system specifications. We have revised the final section to explicitly state the conceptual scope, add example high-level data flow descriptions drawn from standard AI planning pipelines (e.g., generic input/output formats for optimization and ML modules), and clarify that the proposal identifies insertion points rather than providing a deployable blueprint. This revision reinforces the manuscript's claim while acknowledging its boundaries. revision: partial

  2. Referee: [Conclusion and Future Work] No evaluation criteria, metrics, or validation approach (e.g., simulation benchmarks against human planners) are defined for the proposed system; this weakens the claim that the architecture addresses the necessity driven by increasing maneuver speeds, as feasibility remains untestable within the manuscript.

    Authors: We concur that the absence of defined evaluation criteria limits the ability to test feasibility claims. As the manuscript presents a conceptual architecture without an implemented system, empirical validation was outside its original scope. We have expanded the Conclusion and Future Work section to propose evaluation criteria, including metrics such as CoA generation time reduction, doctrinal compliance scores, and comparative quality against human planners in open simulation environments (e.g., using wargaming benchmarks). This addition outlines a path for future validation of the architecture's relevance to increased operational tempos. revision: yes

Circularity Check

0 steps flagged

No significant circularity in conceptual architecture proposal

full rationale

The paper reviews publicly available military doctrines and maps generic AI techniques to CoA planning stages to outline a high-level architecture. No equations, quantitative derivations, fitted parameters, or predictions exist. The central claim is a conceptual mapping from external public sources, which remains independent and does not reduce to self-definition, self-citation chains, or renaming of its own inputs. This is a standard non-circular review-and-proposal structure.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The proposal depends on the assumption that public military doctrines provide sufficient structure for AI integration and that standard AI methods can be applied without domain-specific barriers.

axioms (1)
  • domain assumption Publicly available doctrines accurately represent the stages of CoA planning that can be automated.
    The abstract states the intent to introduce relevant doctrines within publicly available information.

pith-pipeline@v0.9.0 · 5460 in / 1049 out tokens · 39577 ms · 2026-05-15T08:28:17.861058+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

32 extracted references · 32 canonical work pages

  1. [1]

    Department of Defense,Joint Publication 5-0: Joint Planning, Washington, DC, USA, Dec

    U.S. Department of Defense,Joint Publication 5-0: Joint Planning, Washington, DC, USA, Dec. 2020

    work page 2020

  2. [2]

    NATO Allied Command Transformation,Artificial Intelligence in Multi- Domain Operations, NATO ACT Report, Norfolk, V A, USA, 2021

    work page 2021

  3. [3]

    Scharre,Army of None: Autonomous Weapons and the Future of War, W

    P. Scharre,Army of None: Autonomous Weapons and the Future of War, W. W. Norton & Company, New York, NY , USA, 2018

    work page 2018

  4. [4]

    Exploration of wargaming and AI applications in military decision-making,

    H.-X. Li, “Exploration of wargaming and AI applications in military decision-making,” inProc. 2025 International Conference on Military Technologies (ICMT), May 2025, doi: 10.1109/ICMT65201.2025.11061360

    work page doi:10.1109/icmt65201.2025.11061360 2025

  5. [5]

    Army Futures Command,Operationalizing AI for Decision Domi- nance: AI Strategy Implementation Plan, AFC Technical Report No

    U.S. Army Futures Command,Operationalizing AI for Decision Domi- nance: AI Strategy Implementation Plan, AFC Technical Report No. 22- 103, Austin, TX, USA, 2022

    work page 2022

  6. [6]

    Army Futures Command,Future Study Plan 2019: Operational- izing Artificial Intelligence for Multi-Domain Operations, Futures and Concepts Center, Fort Eustis, V A, USA, Aug

    U.S. Army Futures Command,Future Study Plan 2019: Operational- izing Artificial Intelligence for Multi-Domain Operations, Futures and Concepts Center, Fort Eustis, V A, USA, Aug. 2019

    work page 2019

  7. [7]

    Available: https://www.mnd.go.kr/mbshome/ mbs/mndEN/download/Defense Innovation 4.0 brochure EN.pdf 10 Fig

    Ministry of National Defense, Republic of Korea,Defense Innovation 4.0 Brochure, [Online]. Available: https://www.mnd.go.kr/mbshome/ mbs/mndEN/download/Defense Innovation 4.0 brochure EN.pdf 10 Fig. 13. AI-Enabled Multi-Modal IPB and Wargaming Architecture for Updating Enemy and Friendly Courses of Action. Fig. 14. AI-Driven IPB Workflow Across Steps 1–4 ...

    work page

  8. [8]

    AI-enabled wargaming in the military decision making process,

    P. J. Schwartz, D. V . O’Neill, M. E. Bentz, A. Brown, B. S. Doyle, O. C. Liepa, R. Lawrence, and R. D. Hull, “AI-enabled wargaming in the military decision making process,” inArtificial Intelligence and Machine Learning for Multi-Domain Operations Applications II, vol. 11413, pp. 118–134, SPIE, Apr. 2020

    work page 2020

  9. [9]

    Course of Action Display and Evaluation Tool (CADET) enhancements,

    L. Groud and A. Kott, “Course of Action Display and Evaluation Tool (CADET) enhancements,” inProc. Winter Simulation Conference, Orlando, FL, USA, Dec. 2000, pp. 123–130

    work page 2000

  10. [10]

    Knowledge requirements and man- agement in expert decision support systems for (military) situation assessment,

    M. Ben-Bassat and A. Freedy, “Knowledge requirements and man- agement in expert decision support systems for (military) situation assessment,”IEEE Transactions on Systems, Man, and Cybernetics, vol. 12, no. 4, pp. 479–490, 1982

    work page 1982

  11. [11]

    Case-based decision support system: Architecture for simu- lating military command and control,

    S. H. Liao, “Case-based decision support system: Architecture for simu- lating military command and control,”European Journal of Operational Research, vol. 123, no. 3, pp. 558–567, 2000

    work page 2000

  12. [12]

    A knowledge-based method for the validation of military simulation,

    F. Min, P. Ma, and M. Yang, “A knowledge-based method for the validation of military simulation,” inProc. 2007 Winter Simulation Conf., pp. 1395–1402, Dec. 2007

    work page 2007

  13. [13]

    MABSIM: A multi agent based simulation model of military unit combat,

    I. Cil and M. Mala, “MABSIM: A multi agent based simulation model of military unit combat,” inProc. 2009 Second International Conf. on the Applications of Digital Information and Web Technologies, pp. 731– 736, Aug. 2009

    work page 2009

  14. [14]

    Simulating small unit military operations with agent- based models of complex adaptive systems,

    V . Middleton, “Simulating small unit military operations with agent- based models of complex adaptive systems,” inProc. 2010 Winter 11 Fig. 15. AI-Driven IPB-Based Decision-Making Framework for Course of Action Development and Execution. Simulation Conf., pp. 119–134, Dec. 2010

    work page 2010

  15. [15]

    Resilient end-to-end connectivity for software defined unmanned aerial vehicular networks,

    G. Secinti, P. B. Darian, B. Canberk, and K. R. Chowdhury, “Resilient end-to-end connectivity for software defined unmanned aerial vehicular networks,” inProc. 2017 IEEE 28th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), pp. 1–5, Oct. 2017

    work page 2017

  16. [16]

    An end-to-end signal strength model for underwater optical communications,

    M. Doniec, M. Angermann, and D. Rus, “An end-to-end signal strength model for underwater optical communications,”IEEE Journal of Oceanic Engineering, vol. 38, no. 4, pp. 743–757, 2013

    work page 2013

  17. [17]

    End-to-end network modeling and simulation of integrated terrestrial, airborne and space environments,

    L. Baranyai, E. G. Cuevas, S. Davidow, C. Demaree, and P. DiCaprio, “End-to-end network modeling and simulation of integrated terrestrial, airborne and space environments,” inProc. 2005 IEEE Aerospace Conf., pp. 1348–1353, Mar. 2005

    work page 2005

  18. [18]

    Multi-security domain man- agement integration architecture for end-to-end service management in military networks,

    K. D. Tuchs, T. Halmai, and M.vanSelm, “Multi-security domain man- agement integration architecture for end-to-end service management in military networks,” inProc. 2011-MILCOM 2011 Military Communica- tions Conf., pp. 1375–1380, Nov. 2011

    work page 2011

  19. [19]

    F. E. Morgan, B. Boudreaux, A. J. Lohn, M. Ashby, C. Curriden, K. Klima, and D. Grossman,Military applications of artificial intel- ligence, RAND Corporation, Santa Monica, 2020

    work page 2020

  20. [20]

    Appli- cation of artificial intelligence in military: from projects view,

    Y . Zhang, Z. Dai, L. Zhang, Z. Wang, L. Chen, and Y . Zhou, “Appli- cation of artificial intelligence in military: from projects view,” inProc. 2020 6th International Conf. on Big Data and Information Analytics (BigDIA), pp. 113–116, Dec. 2020

    work page 2020

  21. [21]

    Artificial intelligence in the military: An overview of the capabilities, applications, and challenges,

    A. B. Rashid, A. K.Kausik, A.AlHassanSunny, and M. H. Bappy, “Artificial intelligence in the military: An overview of the capabilities, applications, and challenges,”International Journal of Intelligent Sys- tems, vol. 2023, no. 1, p. 8676366, 2023

    work page 2023

  22. [22]

    Investigation on works and military applications of artificial intelligence,

    W. Wang, H. Liu, W. Lin, Y . Chen, and J. A. Yang, “Investigation on works and military applications of artificial intelligence,”IEEE Access, vol. 8, pp. 131614–131625, 2020

    work page 2020

  23. [23]

    AI-enabled wargaming in the military de- cision making process,

    P. J. Schwartz, D. V . O’Neill, M. E. Bentz, A. Brown, B. S. Doyle, Q. C. Liepa, and R. D. Hull, “AI-enabled wargaming in the military de- cision making process,” inArtificial Intelligence and Machine Learning for Multi-Domain Operations Applications II, vol. 11413, pp. 118–134, Apr. 2020

    work page 2020

  24. [24]

    CoA-GPT: Generative pre-trained transformers for accelerated course of action development in military operations,

    V . G. Goecks and N. Waytowich, “CoA-GPT: Generative pre-trained transformers for accelerated course of action development in military operations,” inProc. 2024 International Conference on Military Com- munication and Information Systems (ICMCIS), IEEE, pp. 1–10, Apr. 2024

    work page 2024

  25. [25]

    AI Arms and Influence: Frontier Models Exhibit So- phisticated Reasoning in Simulated Nuclear Crises,

    K. Payne, “AI Arms and Influence: Frontier Models Exhibit So- phisticated Reasoning in Simulated Nuclear Crises,”arXiv preprint arXiv:2602.14740, 2026

    work page arXiv 2026

  26. [26]

    Anthropic’s AI tool Claude central to U.S. campaign in Iran, amid a bitter feud,

    T. Copp, E. Dwoskin, and I. Duncan, “Anthropic’s AI tool Claude central to U.S. campaign in Iran, amid a bitter feud,”The Washington Post, Mar. 4, 2026

    work page 2026

  27. [27]

    The future of military applications of artificial intelligence: A role for confidence-building measures?

    M. C. Horowitz, L. Kahn, and C. Mahoney, “The future of military applications of artificial intelligence: A role for confidence-building measures?”Orbis, vol. 64, no. 4, pp. 528–543, 2020

    work page 2020

  28. [28]

    Artificial intelligence within the military domain and cyber warfare,

    B. Hallaq, T. Somer, A. M. Osula, K. Ngo, and T. Mitchener-Nissen, “Artificial intelligence within the military domain and cyber warfare,” inProc. Eur. Conf. Inf. Warf. Secur. (ECCWS), pp. 153–157, Jun. 2017

    work page 2017

  29. [29]

    Yoon,3D Combined Flight Path Algorithm for UAV Considering Turning and Elevation, M.S

    S. Yoon,3D Combined Flight Path Algorithm for UAV Considering Turning and Elevation, M.S. thesis, Dept. of Mechanical Engineering, TU Korea, Siheung, South Korea, 2019

    work page 2019

  30. [30]

    Army,ADP 5-0: The Operations Process, Washington, DC, USA, Jul

    U.S. Army,ADP 5-0: The Operations Process, Washington, DC, USA, Jul. 2019

    work page 2019

  31. [31]

    Department of the Army,ATP 2-01.3: Intelligence Preparation of the Battlefield, Headquarters, Department of the Army, Washington, DC, 12 USA, Mar

    U.S. Department of the Army,ATP 2-01.3: Intelligence Preparation of the Battlefield, Headquarters, Department of the Army, Washington, DC, 12 USA, Mar. 2019

    work page 2019

  32. [32]

    Department of the Army,FM 34-130: Intelligence Preparation of the Battlefield, Headquarters, Department of the Army, Washington, DC, USA, Jul

    U.S. Department of the Army,FM 34-130: Intelligence Preparation of the Battlefield, Headquarters, Department of the Army, Washington, DC, USA, Jul. 1994. Ji-il Parkreceived the B.S. degree in mechanical engineering from Korea Military Academy, Seoul, South Korea, in 2005, and the M.E. and Ph.D. degrees in mechanical engineering from KAIST, Daejeon, South ...

    work page 1994