VIKI-R: Coordinating Embodied Multi-Agent Cooperation via Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kang, Li, Song, Xiufeng, Zhou, Heng, Qin, Yiran, Yang, Jie, Liu, Xiaohong, Torr, Philip, Bai, Lei, Yin, Zhenfei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918299174961152
author Kang, Li
Song, Xiufeng
Zhou, Heng
Qin, Yiran
Yang, Jie
Liu, Xiaohong
Torr, Philip
Bai, Lei
Yin, Zhenfei
author_facet Kang, Li
Song, Xiufeng
Zhou, Heng
Qin, Yiran
Yang, Jie
Liu, Xiaohong
Torr, Philip
Bai, Lei
Yin, Zhenfei
contents Coordinating multiple embodied agents in dynamic environments remains a core challenge in artificial intelligence, requiring both perception-driven reasoning and scalable cooperation strategies. While recent works have leveraged large language models (LLMs) for multi-agent planning, a few have begun to explore vision-language models (VLMs) for visual reasoning. However, these VLM-based approaches remain limited in their support for diverse embodiment types. In this work, we introduce VIKI-Bench, the first hierarchical benchmark tailored for embodied multi-agent cooperation, featuring three structured levels: agent activation, task planning, and trajectory perception. VIKI-Bench includes diverse robot embodiments, multi-view visual observations, and structured supervision signals to evaluate reasoning grounded in visual inputs. To demonstrate the utility of VIKI-Bench, we propose VIKI-R, a two-stage framework that fine-tunes a pretrained vision-language model (VLM) using Chain-of-Thought annotated demonstrations, followed by reinforcement learning under multi-level reward signals. Our extensive experiments show that VIKI-R significantly outperforms baselines method across all task levels. Furthermore, we show that reinforcement learning enables the emergence of compositional cooperation patterns among heterogeneous agents. Together, VIKI-Bench and VIKI-R offer a unified testbed and method for advancing multi-agent, visual-driven cooperation in embodied AI systems.
format Preprint
id arxiv_https___arxiv_org_abs_2506_09049
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VIKI-R: Coordinating Embodied Multi-Agent Cooperation via Reinforcement Learning
Kang, Li
Song, Xiufeng
Zhou, Heng
Qin, Yiran
Yang, Jie
Liu, Xiaohong
Torr, Philip
Bai, Lei
Yin, Zhenfei
Artificial Intelligence
Computer Vision and Pattern Recognition
Robotics
Coordinating multiple embodied agents in dynamic environments remains a core challenge in artificial intelligence, requiring both perception-driven reasoning and scalable cooperation strategies. While recent works have leveraged large language models (LLMs) for multi-agent planning, a few have begun to explore vision-language models (VLMs) for visual reasoning. However, these VLM-based approaches remain limited in their support for diverse embodiment types. In this work, we introduce VIKI-Bench, the first hierarchical benchmark tailored for embodied multi-agent cooperation, featuring three structured levels: agent activation, task planning, and trajectory perception. VIKI-Bench includes diverse robot embodiments, multi-view visual observations, and structured supervision signals to evaluate reasoning grounded in visual inputs. To demonstrate the utility of VIKI-Bench, we propose VIKI-R, a two-stage framework that fine-tunes a pretrained vision-language model (VLM) using Chain-of-Thought annotated demonstrations, followed by reinforcement learning under multi-level reward signals. Our extensive experiments show that VIKI-R significantly outperforms baselines method across all task levels. Furthermore, we show that reinforcement learning enables the emergence of compositional cooperation patterns among heterogeneous agents. Together, VIKI-Bench and VIKI-R offer a unified testbed and method for advancing multi-agent, visual-driven cooperation in embodied AI systems.
title VIKI-R: Coordinating Embodied Multi-Agent Cooperation via Reinforcement Learning
topic Artificial Intelligence
Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2506.09049