Building Autonomous AI Agents for FPS Battlefields — From High-Level Reasoning to Real-Time Execution

Monday, August 24, 2026 12:15 PM to 1:15 PM · 1 hr. (Europe/Berlin)
Stage 2 - Confex Level 1
Talk
~All~AIAI in Gameplay

Information

This presentation will systematically break down the process of building an FPS battlefield AI agent. We will explore how to combine the situational understanding of Large Language Models with the decision-making optimization of Reinforcement Learning, creating a "cognitive architecture" capable of both autonomous tactical planning and understanding player commands for seamless collaboration. The talk will cover the core "dual-brain" system design, share implementation methods for high-precision skill execution and natural human-agent interaction, and use live combat demos to demonstrate its significant impact on the future of gaming experiences.
Topics
~All~AIAI in Gameplay
Track
AI in Gameplay
Session Type
Talk
Target Audiences
Game AI engineers and technical directors who want to learn about next-generation agent architectures for FPS gameplay. Gameplay designers, level designers, and producers interested in how AI can enhance tactical depth, squad coordination, and immersion. Researchers and practitioners in machine learning, reinforcement learning, and LLMs looking for real-world applications of hybrid AI systems in interactive games. Teams developing online shooters and cooperative experiences who are exploring AI teammates, AI companions, AI commanders, or more advanced NPC behaviors. Industry professionals interested in the future of game interaction including intelligent NPCs, human-AI collaboration, dynamic gameplay, and AI-driven player experiences.
Experience Level
Intermediate
Key Take Aways
Gain a structured understanding of how an FPS battlefield agent is built from perception and context understanding to decision-making and execution. Learn the design logic behind a “dual-brain” architecture combining LLMs and Reinforcement Learning, where LLMs handle high-level semantic understanding and tactical reasoning, while RL optimizes real-time decisions and action control. Recognize the broader impact of this technology on future game experiences, especially in AI companions, player training, cooperative gameplay, and new forms of interactive design.
Available to Ticket Type(s)
  • gcdev Student Pass
    • gcdev Content Pass
      • gcdev Trade Pass
        • gcdev Press Pass
          • gcdev Speaker Pass
            • gcdev VIP Pass

              Join the event!

              See all the content and easy-to-use features by logging in or registering!