GuardPIBT：反事实门控神经引导用于超大规模三维多智能体路径规划
GuardPIBT: Counterfactually Gated Neural Guidance for Ultra-Large-Scale 3D Multi-Agent Path Finding
- Institute of Cyber-Systems and Control, College of Control Science and Engineering, Zhejiang University（浙江大学控制科学与工程学院网络系统与控制研究所）
- Differential Robotics Technology Company（微分机器人技术公司）
- Huzhou Institute, Zhejiang University（浙江大学湖州研究院）
- School of Automation, Hangzhou Dianzi University（杭州电子科技大学自动化学院）
- School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳）理工学院）
- International Digital Economy Academy（国际数字经济学院）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
GuardPIBT通过反事实门控神经引导增强PIBT，在超大规模三维多智能体路径规划中实现可靠完成，支持多达10万智能体且无图违规。
AI中文摘要：
在密集交通条件下，大规模三维多智能体路径规划变得越来越困难。带回溯的优先级继承（PIBT）方法具有良好的可扩展性，但其单步目标导向排序在密集交互和大规模拥堵情况下可能变得不足。我们提出GuardPIBT，它增强而非取代PIBT执行器：神经预测仅提出对PIBT原生候选顺序的残差重排序，而最终动作仍由PIBT决定。首先，局部图注意力建模附近交互，而全局源-目标传输特征提供用于候选重排序的群体级协调上下文。其次，反事实组门控过滤掉其闭环效应可能降低协调性的重排序。第三，对于超大规模群体，群体自适应分组保留决策粒度，异步缓存推理摊销神经计算，选择性修复解决长尾智能体。PIBT在整个过程中保留有效性检查、优先级继承和回溯。在多达100,000个智能体的实验中，展示了在2D和3D环境中可靠的完成率，包括所有三次100,000智能体仓库运行，且审计的图违规为零。项目网站可在{\color{magenta}\texttt{this https URL}}获取。
英文摘要：
Large-scale 3D multi-agent path finding becomes increasingly difficult under dense traffic. Priority Inheritance with Backtracking (PIBT) scales well, but its one-step goal-directed ordering may become insufficient under dense interactions and large-scale congestion. We present GuardPIBT, which augments rather than replaces the PIBT executor: neural predictions only propose residual reorderings of PIBT's native candidates, while final actions remain determined by PIBT. First, local graph attention models nearby interactions, while global source--goal transport features provide population-level coordination context for candidate reordering. Second, a counterfactual group gate filters reorderings whose closed-loop effects may degrade coordination. Third, for ultra-large populations, population-adaptive grouping preserves decision granularity, asynchronous cached inference amortizes neural computation, and selective repair resolves long-tail agents. PIBT retains validity checking, priority inheritance, and backtracking throughout. Experiments with up to 100,000 agents demonstrate reliable completion across 2D and 3D environments, including all three 100,000-agent warehouse runs with zero audited graph violations. The project website is available at {\color{magenta}\texttt{https://guardpibt.github.io/GuardPIBT/}}.
