arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节（Veterans Day），12 日无更新
arXiv 2603.04750cs.AIcs.CL
收藏

HiMAP-Travel: 基于长时域约束旅行的分层多智能体规划

HiMAP-Travel: Hierarchical Multi-Agent Planning for Long-Horizon Constrained Travel

Department of XXX, University of YYY, Location, Country（YYY大学XXX系）；School of ZZZ, Institute of WWW, Location, Country（WWW研究所ZZZ学院）；Singapore Management University（新加坡管理大学）

The Viet Bui, Wenjun Li, Yong Liu

备注：
33 pages, v1
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

HiMAP-Travel通过分层多智能体框架提升长时域约束旅行规划效率，实现更高通过率和更低延迟。

摘要：

顺序LLM代理在处理具有预算和多样性要求等硬约束的长时域规划时表现不佳。随着规划进程和上下文的增长，这些代理会偏离全局约束。我们提出了HiMAP-Travel，一种分层多代理框架，将规划分为战略协调和并行日级执行。协调器在多天范围内分配资源，而日执行器在并行中独立规划。三个关键机制使这一实现成为可能：一个事务监控器强制在并行代理之间执行预算和唯一性约束，一个谈判协议允许代理拒绝不可行的子目标并触发重新规划，以及一个通过角色条件训练的单一策略，为所有代理提供动力。在TravelPlanner上，HiMAP-Travel与Qwen3-8B实现52.78%的验证通过率和52.65%的测试最终通过率（FPR）。在与相同模型、训练和工具的受控比较中，它比顺序DeepTravel基线高出+8.67个百分点。它还超过了ATLAS高出+17.65个百分点，超过MTP高出+10.0个百分点。在FlexTravelBench多轮场景中，它在2轮时达到44.34%的FPR，在3轮时达到37.42%的FPR，同时通过并行化将延迟减少2.5倍。

英文摘要：

Sequential LLM agents fail on long-horizon planning with hard constraints like budgets and diversity requirements. As planning progresses and context grows, these agents drift from global constraints. We propose HiMAP-Travel, a hierarchical multi-agent framework that splits planning into strategic coordination and parallel day-level execution. A Coordinator allocates resources across days, while Day Executors plan independently in parallel. Three key mechanisms enable this: a transactional monitor enforcing budget and uniqueness constraints across parallel agents, a bargaining protocol allowing agents to reject infeasible sub-goals and trigger re-planning, and a single policy trained with GRPO that powers all agents through role conditioning. On TravelPlanner, HiMAP-Travel with Qwen3-8B achieves 52.78% validation and 52.65% test Final Pass Rate (FPR). In a controlled comparison with identical model, training, and tools, it outperforms the sequential DeepTravel baseline by +8.67~pp. It also surpasses ATLAS by +17.65~pp and MTP by +10.0~pp. On FlexTravelBench multi-turn scenarios, it achieves 44.34% (2-turn) and 37.42% (3-turn) FPR while reducing latency 2.5x through parallelization.

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。

↑