Мультиагентна система підтримки випробувань озброєння та військової техніки

Vadym Slyusar

doi:10.34169/2414-0651.2026.1(49).65-75

Multi-agent system for supporting the testing of weapons and military equipment

Authors

Vadym Slyusar Central Scientific Research Institute of Armament and Military Equipment of Armed Forces of Ukraine https://orcid.org/0000-0002-2912-3149

DOI:

https://doi.org/10.34169/2414-0651.2026.1(49).65-75

Keywords:

artificial intelligence, multi-agent systems, large language models

Abstract

The article proposes a multi-agent system for supporting the testing of weapons and military equipment, which consists of a set of specialized intelligent agents with clearly defined functions. The basic ones are the agents for technical documentation analysis, test result formalization, tactical scenario generation, and the logistics support agent. Separate attention is paid to the role of the orchestrator, the user interaction agent, and the cybersecurity agent. Such a combination creates an end-to-end Machine-in-the-Loop cycle in which the planning, execution, and analysis of tests take place as a controlled engineering process with transparent traceability and the possibility of rapid repetition. The article also considers the practical feasibility of the proposed architecture thanks to modern language models. On the one hand, the infrastructure of the n8n automation platform and of industrial-class platforms such as Amazon Bedrock AgentCore already provide a managed environment for deploying agents. On the other hand, local Ukrainian-language multimodal models MamayLM-Gemma-3 and Lapa already support working with text and images, large context windows, and are capable of processing technical materials in Ukrainian. This makes it possible to perform analysis of documentation and visual data without moving sensitive information outside the controlled environment. The article concludes that the multi-agent system for testing weapons and military equipment can transform testing from a slow sequential mode into a fast iterative cycle with digital tracing of requirements, automated scenario generation, continuous data collection and analysis, semi-automated reporting, and controlled logistics. This creates the preconditions for a significant reduction of the time required to introduce new samples of weapons and military equipment down to the level of weeks or even days without compromises in safety, reproducibility, and compliance with regulatory requirements.

Downloads

Download data is not yet available.

Author Biography

Vadym Slyusar, Central Scientific Research Institute of Armament and Military Equipment of Armed Forces of Ukraine

Doctor of Technical Sciences, Professor

Group of Chief Research Scientists for Scientific Research Management of Central Scientific Research Institute of Armament and Military Equipment of Armed Forces of Ukraine

References

Kondratenko, Y., Kondratenko, G., Shevchenko, A., Slyusar, V., Zhukov, Y. & Vakulenko, M. (2023). Towards Implementing the Strategy of Artificial Intelligence Development: Ukraine Peculiarities. CEUR Workshop Proc. Vol. 3513. Pp. 106—117. Available at:

https://ceur-ws.org/Vol-3513/paper 09.pdf.

Kondratenko, Y., Shevchenko, A., Zhukov, Y., Klymenko, M., Slyusar, V., Kondratenko, G. & Striuk, O. (2023). Analysis of the Priorities and Perspectives in Artificial Intelligence Implementation. 13th Intern. IEEE Conf. «Dependable Systems, Services and Technologies» (DESSERT’2023). Greece. Athens. 8 p. October 13–15. https://doi.org/10.1109/DESSERT61349.2023.10416432. DOI: https://doi.org/10.1109/DESSERT61349.2023.10416432

Kondratenko, Y.P., Slyusar, V.I., Solesvik, M.B., Kondratenko, N.Y. & Gomolka, Z. (2024). Interrelation and inter-influence of artificial intelligence and higher education systems. Research Tendencies and Prospect Domains for AI Development and Implementation. Pp. 31—58. DOI: https://doi.org/10.1201/9788770046947-2

Slyusar, V.I., Kondratenko, Y.P., Shevchenko, A.I. & Yeroshenko, T.V. (2024). Some Aspects of Artificial Intelligence Development Strategy for Mobile Technologies. J. of Mobile Multimedia. Vol. 20_3. Pp. 525—554. Available at: 10.13052/jmm1550-4646.2031. DOI: https://doi.org/10.13052/jmm1550-4646.2031

Patwardhan, T. et al. GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks. OpenAI. Oct. 2025. Available at: DOI: https://doi.org/10.70777/si.v2i4.17197

https://cdn.openai.com/pdf/d5eb7428-c4e9-4a33-bd86-86dd4bcf12ce/GDPval.pdf. [Accessed: Oct. 25, 2025].

Slyusar, V. (2025). Distributed Multi-agent Systems Based on the Mixture of Experts Architecture in the Context of 6G Wireless Technologies. In: Dovgyi, S., et al. (eds) Applied Innovations in Information and Communication Technology. ICAIIT 2024. Lecture Notes in Networks and Systems. Vol. 1338. Springer. Pp. 81—110. https://doi.org/10.1007/978-3-031-89296-7_6. DOI: https://doi.org/10.1007/978-3-031-89296-7_6

Zhang, D., Li, Z., Wang, P., Zhang, X., Zhou, Y. & Qiu, X. (2024). SpeechAgents: Human-Communication Simulation with Multi-Modal Multi-Agent Systems. arXiv. Available at:

https://arxiv.org/abs/2401.03945.

Baier, T., Baez Santamaria, S. & Vossen, P. (2022). A modular architecture for creating multimodal agents. arXiv. Available at:

https://arxiv.org/abs/2206.00636.

Слюсар В.І. Результати тестування локальної мовної моделі gpt-oss-20b. XXV Наук.-техн. конф. «Випробування і сертифікація озброєння та військової техніки». 25 вересня 2025 р. Черкаси: ДНДІ випробувань і сертифікації ОВТ.

Bhardwaj, D., Beniwal, A., Chaudhari, S., Kalyan, A., Rajpurohit, T., Narasimhan, K.R., Deshpande, A., & Murahari, V. (2025). Agent context protocols enhance collective inference. arXiv. at:

https://arxiv.org/abs/2505.14569.

Habler, I., Huang, K., Narajala, V. S., & Kulkarni, P. (2025). Building a secure agentic AI application leveraging A2A protocol. arXiv. at:

https://arxiv.org/abs/2504.16902.

Jeong, C. (2025). A study on the MCP × A2A framework for enhancing interoperability of LLM-based autonomous agents. arXiv. at:

https://arxiv.org/abs/2506.01804.

Liu, J., Yu, K., Chen, K., Li, K., Qian, Y., Guo, X., Song, H., & Li, Y. (2025). ACPs: Agent collaboration protocols for the Internet of Agents. arXiv. at: DOI: https://doi.org/10.1109/IC-NIDC67200.2025.11390349

https://arxiv.org/abs/2505.13523.

Martineau, K. (2025, May 28). The simplest protocol for AI agents to work together. IBM Research. at: https://research.ibm.com/blog/agent-communication-protocol-ai.

Слюсар, В. (2024). Локальні великі мовні моделі для обробки конфіденційної інформації . Озброєння та військова техніка, 44(4), 79–91.

https://doi.org/10.34169/2414-0651.2024.4(44).79-91 DOI: https://doi.org/10.34169/2414-0651.2024.4(44).79-91

Слюсар, В., & Гусаковський, І. (2025). Практичні аспекти розгортання великих мовних моделей в локальних мережах. Озброєння та військова техніка, 45(1), 84–96.

https://doi.org/10.34169/2414-0651.2025.1(45).84-96 DOI: https://doi.org/10.34169/2414-0651.2025.1(45).84-96

Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H. Lewis, M. Yih, W.-t. Rocktäschel, T. et al. (2020). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Advances in Neural Information Processing Systems. Vol. 33. Pp. 9459—9474. Available at:

https://arxiv.org/abs/2005.11401.

INSAIT-Institute/MamayLM-Gemma-3-4B-IT-v1.0. Available at: https://huggingface.co/mradermacher/MamayLM-Gemma-3-4B-IT-v1.0-GGUF.

INSAIT-Institute/MamayLM-Gemma-3-12B-IT-v1.0. Available at: https://huggingface.co/mradermacher/MamayLM-Gemma-3-12B-IT-v1.0-GGUF.

Lapa LLM. Available: https://huggingface.co/lapa-llm. Accessed: Oct. 25, 2025.

Agarwal, S. et al. (2025). GPT-OSS-120B & GPT-OSS-20B model card. arXiv. Available at: https://doi.org/10.48550/arXiv.2508.10925.

Нос І.А. Особливості формування загальних вимог та технічних умов ОВТ при використанні інформаційних технологій підтримки випробувань в умовах правового режиму воєнного часу. XXV Наук.-техн. конф. «Випробування і сертифікація озброєння та військової техніки». 25 вересня 2025 р. Черкаси: ДНДІ випробувань і сертифікації ОВТ.

Parthasarathy, V. B., Zafar, A., Khan, A. & Shahid, A. The Ultimate Guide to Fine-Tuning LLMs from Basics to Breakthroughs: An Exhaustive Review of Technologies, Research, Best Practices, Applied Research Challenges and Opportunities. arXiv. Available at:

https://arxiv.org/abs/2408.13296.

OpenAI, GPT-5 System Card, Aug. 7, 2025. Available at: https://cdn.openai.com/gpt-5-system-card.pdf. Accessed: Oct. 25, 2025.

Google DeepMind, Gemini 2.5 Deep Think Model Card, Aug. 1, 2025. Available at: https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Deep-Think-Model-Card.pdf. Accessed: Oct. 25, 2025.

Anthropic, Claude Sonnet 4.5 System Card, Oct. 10, 2025. Available at:

https://www.anthropic.com/claude-sonnet-4-5-system-card. Accessed: Oct. 25, 2025.

n8n – Workflow Automation, GitHub, n8n-io (verified organization). 2025. Available at:

https://github.com/n8n-io. Accessed: Oct. 25, 2025.

Гусаковський І.П., Слюсар В.І., Чепков І.Б. Трансформація робочих процесів: практичне застосування Microsoft Power Automate та Google Gemini для автоматизації аналітики та звітності. Міжнар. наук.-практ. семінар «Проблематика, тенденції та перспективи розвитку воєнної науки та освіти в умовах сучасних глобальних викликів та конфліктів». 26 листопада 2025 р. Київ: Центральний наук.-досл. інст. Збройних Сил України.

Amazon Web Services. What is Amazon Bedrock AgentCore? Jul. 16, 2025. Available at: https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/what-is-bedrock-agentcore.html. . Accessed: Oct. 25, 2025.

OpenEnv: Agentic Execution Environments, Hugging Face organization page. Oct. 22, 2025. Available at: https://huggingface.co/openenv. Accessed: Oct. 25, 2025.