arXiv cs.MA· Haishan Zeng, Mengna Wang, Peng Li·· 12 小时前AI 评分32
EmboTeam：面向异构多机器人协作的非对称状态接地框架
EmboTeam: Asymmetric State Grounding for Heterogeneous Multi-Robot Collaboration
AI 导读
EmboTeam 提出一种神经符号框架，让 LLM 生成结构化状态期望，经 PDDL 符号搜索验证后接地为团队级并行行为树，实现异构多机器人长程协作。团队还构建了涵盖 8 种室内布局、42 个异构多机器人任务的 MACE-THOR 基准。相比 LaMMA-P，任务成功率从 12% 提升至 55%，目标条件召回率从 32% 提升至 75%。
正文 · 原文
Abstract:Heterogeneous multi-robot collaboration over long horizons requires both global task planning and robust coordination during execution. Existing plan- ners using large language models (LLMs) can generate task decompositions or symbolic plans, but often execute them as rigid action sequences, making them fragile to local state mismatches and synchronization failures. We propose Em- boTeam, a neuro-symbolic framework for asymmetric state grounding. An LLM generates structured state expectations, which are verified and coordinated through symbolic search using Planning Domain Definition Language (PDDL) and then grounded into a team-level parallel behavior tree. Robot-specific subtrees per- form precondition checking, skill execution, effect validation, local replanning, and blackboard-based synchronization. We further introduce MACE-THOR, con- taining 42 heterogeneous multi-robot tasks across 8 indoor layouts. EmboTeam improves task success rate from 12% to 55% and goal condition recall from 32% to 75% over LaMMA-P, demonstrating the effectiveness of synchronized reactive grounding for collaboration over long horizons.
| Comments: | Accepted to the 33rd International Conference on Multimedia Modeling (MMM 2027) |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2601.11063 [cs.RO] |
| (or arXiv:2601.11063v3 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2601.11063 arXiv-issued DOI via DataCite |
Submission history
From: Haishan Zeng [view email]
[v1]
Fri, 16 Jan 2026 07:59:50 UTC (33,397 KB)
[v2]
Thu, 5 Mar 2026 13:18:20 UTC (1,788 KB)
[v3]
Thu, 5 Nov 2026 07:34:20 UTC (1,171 KB)
来源：arXiv cs.MA · arxiv.org