ResearcharXiv cs.AI

Coachable agents for interactive gameplay

#reinforcement learning#interactive gameplay#ai agents#game design#real-time control

English

The paper presents a framework for coaching AI agents to exhibit specific styles in interactive gameplay using reinforcement learning techniques. It demonstrates the application of this framework in various domains, including AAA video games and a humanoid test environment, allowing users to control agent behavior in real-time while maintaining task performance.

中文

本文提出了一种框架，用于指导AI代理在互动游戏中展示特定风格，采用强化学习技术。它展示了该框架在多个领域的应用，包括AAA视频游戏和人形测试环境，使用户能够实时控制代理行为，同时保持任务性能。