mlsanityData to 7 Oct 2026

Fly, Fail, Fix: Iterative Game Repair with Reinforcement Learning and Large Multimodal Models

arXiv 2507.12666 ↗First posted 16 Jul 2025cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Game design hinges on understanding how static rules and content translate into dynamic player behavior - something modern generative systems that inspect only a game's code or assets struggle to capture. We present an automated design iteration framework that closes this gap by pairing a reinforcement learning (RL) agent, which playtests the game, with a large multimodal model (LMM), which revises the game based on what the agent does. In each loop the RL player completes several episodes, producing (i) numerical play metrics and/or (ii) a compact image strip summarising recent video frames. The LMM designer receives a gameplay goal and the current game configuration, analyses the play traces, and edits the configuration to steer future behaviour toward the goal. We demonstrate results that LMMs can reason over behavioral traces supplied by RL agents to iteratively refine game mechanics, pointing toward practical, scalable tools for AI-assisted game design.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 19 of its 28 references that are in the map. First posted on arXiv 16 Jul 2025.

Builds on

19 of 28 references.

References · 19

  1. 2019Dota 2 with Large Scale Deep Reinforcement Learningcited by 1,084
  2. 2021Deep Reinforcement Learning at the Edge of the Statistical Precipicecited by 781
  3. 2023Eureka: Human-Level Reward Design via Coding Large Language Modelscited by 542
  4. 2023Reward Design with Language Modelscited by 257
  5. 2023Holodeck: Language Guided Generation of 3D Embodied AI Environmentscited by 257
  6. 2023RoboGen: Towards Unleashing Infinite Data for Automated Robot Learning via Generative Simulationcited by 206
  7. 2021Open-Ended Learning Leads to Generally Capable Agentscited by 177
  8. 2020Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutionscited by 103
  9. 2020PCGRL: Procedural Content Generation via Reinforcement Learningcited by 90
  10. 2024DrEureka: Language Model Guided Sim-To-Real Transfercited by 78
  11. 2023Level Generation Through Large Language Modelscited by 68
  12. 2023MarioGPT: Open-Ended Text2Level Generation through Large Language Modelscited by 63
7 more
  1. 2023Gen2Sim: Scaling up Robot Learning in Simulation with Generative Modelscited by 62
  2. 2024EnvGen: Generating and Adapting Environments via LLMs for Training Embodied Agentscited by 41
  3. 2021Adversarial Reinforcement Learning for Procedural Content Generationcited by 36
  4. 2023MAESTRO: Open-Ended Environment Design for Multi-Agent Reinforcement Learningcited by 30
  5. 2024The Ink Splotch Effect: A Case Study on ChatGPT as a Co-Creative Game Designercited by 12
  6. 2019Automatic Playtesting for Game Parameter Tuning via Active Learningcited by 9
  7. 2024FactorSim: Generative Simulation via Factorized Representationcited by 9

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Autoverse: An Evolvable Game Language for Learning Robust Embodied Agents
  2. 2025Learning Game-Playing Agents with Generative Code Optimization
  3. 2024Read to Play (R2-Play): Decision Transformer with Multimodal Game Instruction
  4. 2026Synthesizing Reactive Character Behaviors for Continuous Games via Programmatic Policy Search
  5. 2026GameDirector: Decoupling Gameplay Logic from Rendering for Player-Configurable Game World Models
  6. 2026RLHarness: Co-evolving Procedural Skills with Reinforcement Learning for Long-horizon Multimodal Reasoning
  7. 2019Automated Video Game Testing Using Synthetic and Human-Like Agents
  8. 2026CAST: Game Solvers as Turn-Level Teachers for LLM Agents