arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.15907cs.LGcs.SYeess.SY
收藏

基于博弈论的边境防御强化学习:基于解析解的提前终止

Game-Theory-Assisted Reinforcement Learning for Border Defense: Early Termination based on Analytical Solutions

School of Aeronautics and Astronautics, Purdue University(航空航天学院,普渡大学);Army Research Directorate, DEVCOM Army Research Laboratory, APG(陆军研究局,国防高级研究计划局陆军研究实验室,APG);Munitions Directorate, Air Force Research Laboratory(弹药局,空军研究实验室);Department of Mechanical Engineering, George Mason University(机械工程系,乔治·马歇尔大学)

Goutam Das, Michael Dorothy, Kyle Volle, Daigo Shishika

AI总结:

本文提出一种结合博弈论与强化学习的方法,通过Apollonius圆计算均衡,实现RL回合提前终止,提升效率和奖励。

摘要:

博弈论为对抗性对抗提供了金标准,提供强最优性保证。然而,当假设如完全信息被违反时,这些保证往往变得脆弱。相比之下,强化学习(RL)具有适应性,但在大型复杂领域中样本效率低下。本文介绍了一种混合方法,利用博弈论见解来提高RL训练效率。我们研究了一个具有有限感知范围的边境防御游戏,其中防御性能取决于搜索和追击策略,使得经典微分博弈解决方案不适用。我们的方法利用Apollonius圆(AC)在检测后阶段计算均衡,使RL回合能够提前终止而不学习追击动态。这使RL能够专注于学习搜索策略,同时在检测后保证最优延续。在单个和多个防御者设置中,这种提前终止方法产生了10-20%更高的奖励、更快的收敛速度和更高效的搜索轨迹。广泛的实验验证了这些发现并展示了我们方法的整体有效性。

英文摘要:

Game theory provides the gold standard for analyzing adversarial engagements, offering strong optimality guarantees. However, these guarantees often become brittle when assumptions such as perfect information are violated. Reinforcement learning (RL), by contrast, is adaptive but can be sample-inefficient in large, complex domains. This paper introduces a hybrid approach that leverages game-theoretic insights to improve RL training efficiency. We study a border defense game with limited perceptual range, where defender performance depends on both search and pursuit strategies, making classical differential game solutions inapplicable. Our method employs the Apollonius Circle (AC) to compute equilibrium in the post-detection phase, enabling early termination of RL episodes without learning pursuit dynamics. This allows RL to concentrate on learning search strategies while guaranteeing optimal continuation after detection. Across single- and multi-defender settings, this early termination method yields 10-20% higher rewards, faster convergence, and more efficient search trajectories. Extensive experiments validate these findings and demonstrate the overall effectiveness of our approach.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑