EmboTeam: Asymmetric State Grounding for Heterogeneous Multi-Robot Collaboration
EmboTeam: Asymmetric State Grounding for Heterogeneous Multi-Robot Collaboration
This paper presents EmboTeam, a neuro-symbolic framework designed for asymmetric state grounding in heterogeneous multi-robot collaboration. The system uses an LLM to generate structured state expectations, which are then verified via symbolic search using PDDL and grounded into a team-level parallel behavior tree. Robot-specific subtrees handle precondition checking, skill execution, effect validation, local replanning, and blackboard-based synchronization. The authors evaluate the framework on MACE-THOR, a dataset containing 42 heterogeneous multi-robot tasks across 8 indoor layouts. Results show that EmboTeam improves task success rate from 12% to 55% and goal condition recall from 32% to 75% compared to LaMMA-P, demonstrating the effectiveness of synchronized reactive grounding for long-horizon collaboration.
The authors introduce EmboTeam, a neuro-symbolic framework that combines LLM-generated state expectations with PDDL-based symbolic search and parallel behavior trees to improve task success rates from 12% to 55% on the MACE-THOR benchmark of 42 heterogeneous multi-robot tasks across 8 indoor layouts.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported success rate
55%
Reported trials
View original evidence
oard-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.
Open source S4Reported recall
75%
Reported trials
View original evidence
oard-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.
Open source S4Reported success rate
12%
Reported trials
View original evidence
oard-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.
Open source S4Reported recall
32%
Reported trials
View original evidence
oard-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.
Open source S4
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
oard-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.
Open source S4
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 blackb
Open source S3
Source：arXiv Robotics — research abstracts · arxiv.org