Closed-Loop Vision-Language Planning for Multi-Agent Coordination

Zhiyuan Li, Wenshuai Zhao, Joni Pajarinen · 2025

Read Closed-Loop Vision-Language Planning for Multi-Agent Coordination by Zhiyuan Li, Wenshuai Zhao, Joni Pajarinen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-10148.txt.

Description

Zhiyuan Li, Wenshuai Zhao, and Joni Pajarinen present a 2025 paper on teams of AI agents that plan together using vision and language. Standard multi-agent learning is slow to train, hard to read, and poor at new tasks. Language models plan well, but past uses leaned on text alone and missed the shifting, partly hidden nature of team tasks.

Their system builds and refines plain, readable code strategies, stored in a skill library seeded by expert demos. A structured message chain passes facts about entities between agents, so the team shares one picture. On a hard battle benchmark, the system beat top rivals by a wide margin in one match type.

Catalog record

CREATORZhiyuan Li, Wenshuai Zhao, Joni Pajarinen
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE94 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-10148

Browse the full library · About the Tower of Babel