arXiv:2510.23216v3 Announce Type: replace-cross 
Abstract: While several high profile video games have served as testbeds for Deep Reinforcement Learning (DRL), this technique has rarely been employed by the game industry for crafting authentic AI behaviors. Previous research focuses on training super-human agents with large models, which is impractical for game studios with limited resources aiming for human-like agents. This paper proposes a sample-efficient DRL method tailored for training and fine-tuning agents in industrial settings such as the video game industry. Our method improves sample efficiency of value-based DRL by leveraging pre-collected data and increasing network plasticity. We evaluate our method training a goalkeeper agent in EA SPORTS FC 25, one of the best-selling football simulations today. Our agent outperforms the game's built-in AI by 10% in ball saving rate. Ablation studies show that our method trains agents 50% faster compared to standard DRL methods. Finally, qualitative evaluation from domain experts indicates that our approach creates more human-like gameplay compared to hand-crafted agents. As a testament to the impact of the approach, the method has been adopted for use in the most recent release of the series.

تقدم دراسة جديدة طريقة فعالة في التعلم المعزز قد تحدث ثورة في الذكاء الاصطناعي في ألعاب الفيديو من خلال تمكين إنشاء حراس مرمى يشبهون البشر. تتناول هذه الطريقة التحديات التي تواجهها استوديوهات الألعاب ذات الموارد المحدودة، مما يسمح لها بتطوير سلوكيات ذكاء اصطناعي أصيلة دون الحاجة إلى نماذج تدريب واسعة. هذه الخطوة مهمة لأنها تفتح آفاقًا جديدة لتجارب لعب أكثر واقعية وجاذبية، مما يسهل على المطورين تنفيذ ذكاء اصطناعي متطور في ألعابهم.

Un nuevo estudio presenta un método de aprendizaje por refuerzo eficiente en muestras que podría revolucionar la IA en los videojuegos al permitir la creación de porteros similares a humanos. Este enfoque aborda los desafíos que enfrentan los estudios de juegos con recursos limitados, permitiéndoles desarrollar comportamientos de IA auténticos sin necesidad de modelos de entrenamiento extensos. Este avance es significativo ya que abre nuevas posibilidades para experiencias de juego más realistas y atractivas, facilitando la implementación de IA sofisticadas en sus juegos.

Une nouvelle étude présente une méthode d'apprentissage par renforcement échantillon efficace qui pourrait révolutionner l'IA dans les jeux vidéo en permettant la création de gardiens de but ressemblant à des humains. Cette approche répond aux défis rencontrés par les studios de jeux avec des ressources limitées, leur permettant de développer des comportements d'IA authentiques sans avoir besoin de modèles d'entraînement étendus. Cette avancée est significative car elle ouvre de nouvelles possibilités pour des expériences de jeu plus réalistes et engageantes, facilitant ainsi l'implémentation d'IA sophistiquées dans leurs jeux.

A new study introduces a sample-efficient reinforcement learning method that could revolutionize AI in video games by enabling the creation of human-like goalkeepers. This approach addresses the challenges faced by game studios with limited resources, allowing them to develop authentic AI behaviors without the need for extensive training models. This advancement is significant as it opens up new possibilities for more realistic and engaging gameplay experiences, making it easier for developers to implement sophisticated AI in their games.

Human-Like Goalkeeping in a Realistic Football Simulation: a Sample-Efficient Reinforcement Learning Approach

One More Thing in AI – Your Shortcut to AI Mastery

Human-Like Goalkeeping in a Realistic Football Simulation: a Sample-Efficient Reinforcement Learning Approach

Was this article worth reading? Share it

One More Thing in AI

LucidQuery AI

Humanize AI

Chattermate

AIvilization

Synthx

Ready to build your own newsroom?