Metraipolis › archive › dc_d84cc6ec36ef
LLM-Guided Reinforcement Learning for Adaptive NPC Behavior in Multi-Agent Combat Games — arXiv:2609.02931
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.02931
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T01:02:57.883Z
- SHA-256 of the archived text
abddc2498455abe17c5fa736dad0995ee59022ddf0650a7a3fa52b19275fb311- Stable URL of this page
https://metraipolis.com/documento/dc_d84cc6ec36ef
How to cite
LLM-Guided Reinforcement Learning for Adaptive NPC Behavior in Multi-Agent Combat Games — arXiv:2609.02931 — arXiv (en), https://arxiv.org/abs/2609.02931 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d84cc6ec36ef (Metraipolis archive dc_d84cc6ec36ef, sha256 abddc2498455abe17c5fa736dad0995ee59022ddf0650a7a3fa52b19275fb311)
BibTeX
@misc{dcd84cc6ec36ef,
title = {LLM-Guided Reinforcement Learning for Adaptive NPC Behavior in Multi-Agent Combat Games — arXiv:2609.02931},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.02931},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d84cc6ec36ef; sha256 abddc2498455abe17c5fa736dad0995ee59022ddf0650a7a3fa52b19275fb311},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.02931. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Scripted and rule-based non-player characters (NPCs) in combat video games often exhibit predictable behaviors that experienced players can exploit, while reinforcement learning (RL) agents typically retain a fixed policy after training and cannot readily adapt their strategy to different opponents. We investigate a runtime strategy-selection framework in which a large language model (LLM) guides a trained RL policy without modifying its underlying behavior. To demonstrate this, we train five NPC agents with a shared PPO policy in Unity and compare a baseline configuration, in which the policy acts independently, with an LLM-augmented configuration in which a locally hosted Mistral 7B model, accessed through Ollama, reads the live game state every five seconds and assigns one of four tactical tags. We evaluate both configurations against three scripted opponent types across 600 episodes and analyze outcomes using the Mann-Whitney U test. Against a Balanced opponent that changes tactics during an episode, the LLM-augmented agents more than doubled their win rate from 11% to 24% and produced significantly longer episodes. Against an Evasive opponent, the augmented agents achieved a higher win rate and faster kills, although their shorter episode duration did not satisfy the strict hypothesis definition. Against an Aggressive opponent, the LLM's near-constant preference for encirclement was counterproductive. Analysis of 2,430 strategy selections showed that Surround was selected in 83.8% of cases regardless of opponent type, indicating limited zero-shot strategic differentiation at this model scale. These results demonstrate both the potential and limitations of LLM-guided runtime strategy selection for adaptive multi-agent game AI. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-04 in cs.MA). Categorie arXiv: Multiagent Systems (cs.MA) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.02931 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T00:22:10.221Z dalla pagina https://arxiv.org/catchup/cs.MA/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).