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 2210.09378cs.ROcs.AIcs.MAcs.SYeess.SY
收藏

基于图神经网络的多智能体导航学习控制容许性模型

Learning Control Admissibility Models with Graph Neural Networks for Multi-Agent Navigation

UCSD(加州大学圣地亚哥分校)

Chenning Yu, Hongzhan Yu, Sicun Gao

摘要:

连续域中的深度强化学习专注于学习控制策略,这些策略将状态映射到动作上的分布,理想地集中在每个步骤中的最优选择上。在多智能体导航问题中,最优行动在很大程度上依赖于智能体的密度。它们的交互模式相对于这样的密度呈指数级增长,使得基于学习的方法很难推广。我们建议将学习目标从预测最佳行动切换到预测容许行动集,我们称之为控制容许模型(CAM),这样它们就可以很容易地组成并用于任意数量代理的在线推理。我们使用图神经网络设计CAM,并开发在标准无模型设置中优化CAM的训练方法,其额外好处是消除了通常需要平衡避免碰撞和达到目标要求的奖励工程的需要。我们评估所提出的方法在多智能体导航环境。我们表明,CAM模型可以在只有几个代理的环境中进行训练,并且可以很容易地在具有数百个代理的密集环境中进行部署,从而实现比最先进的方法更好的性能。

英文摘要:

Deep reinforcement learning in continuous domains focuses on learning control policies that map states to distributions over actions that ideally concentrate on the optimal choices in each step. In multi-agent navigation problems, the optimal actions depend heavily on the agents' density. Their interaction patterns grow exponentially with respect to such density, making it hard for learning-based methods to generalize. We propose to switch the learning objectives from predicting the optimal actions to predicting sets of admissible actions, which we call control admissibility models (CAMs), such that they can be easily composed and used for online inference for an arbitrary number of agents. We design CAMs using graph neural networks and develop training methods that optimize the CAMs in the standard model-free setting, with the additional benefit of eliminating the need for reward engineering typically required to balance collision avoidance and goal-reaching requirements. We evaluate the proposed approach in multi-agent navigation environments. We show that the CAM models can be trained in environments with only a few agents and be easily composed for deployment in dense environments with hundreds of agents, achieving better performance than state-of-the-art methods.

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。

↑