arXiv cs.AI· Changxuan Fan, Zhongwei Xie, Boyi Zhang, Shaoli Hu, Chang Xu, Xiyu Ren, Yufei Li, Qiao Xiao, Huihao Jing, Kawai Chung, Wenbin Hu, Hongyu Luo, Lingyun Xie, Yisen Gao, Chunyang Li, Haoyu Huang, Wuganjing Song, Yuxuan Liu, Chiu Yeung Tang, Fangneng Zhan, Xi Yang·· 4 小时前AI 评分42
RoboTeam:实时多机械臂视觉多智能体控制基准
RoboTeam: Benchmarking Vision-Based Multi-Agent Control of Many Robot Arms in Real Time
AI 导读
RoboTeam 是一个模拟基准,让智能体仅凭摄像头图像实时控制 4 至 16 台机械臂,完成 36 项需协同的长任务。在最灵活的自适应系统中,gpt-6.1-sol 和 opus-5.5 分别完成 28 和 27 项任务,其余四个模型最多完成 3 项;但典型任务耗时 3.5 小时、消耗 2.18 亿输入 token,且三分之一步骤至少失败一次需重试。
正文 · 原文
Authors:Changxuan Fan, Zhongwei Xie, Boyi Zhang, Shaoli Hu, Chang Xu, Xiyu Ren, Yufei Li, Qiao Xiao, Huihao Jing, Kawai Chung, Wenbin Hu, Hongyu Luo, Lingyun Xie, Yisen Gao, Chunyang Li, Haoyu Huang, Wuganjing Song, Yuxuan Liu, Chiu Yeung Tang, Fangneng Zhan, Xi Yang
Abstract:Multimodal agents can now control robot arms by calling tools and writing code. Existing benchmarks fall short in three ways: most test one or two arms on short tasks, many pause the world while the agent thinks, and many give it ground-truth object positions. We present RoboTeam, a simulated benchmark where agents control 4 to 16 robot arms in real time from camera images alone. Its 36 tasks are long jobs that the arms must do together, such as passing timbers or building a picture from hundreds of cubes. We start from the most flexible agent system, the adaptive system, where one agent decides when to fork delegates. With it, gpt-6.1-sol and opus-5.5 complete 28 and 27 tasks, and four other models at most 3. But at a high price: a typical task takes gpt-6.1-sol 3.5 hours and 218 million input tokens, and a third of its steps fail at least once and must be retried. We then compare it with three fixed systems: single-agent, hierarchical, and decentralized. All four complete a similar number of tasks but differ in time and tokens. The hierarchical and decentralized systems are faster but use up to 1.9 times the tokens, and the single agent uses fewer tokens but takes 1.6 times as long, so the adaptive system lies between them. Even so, no model or system is yet fast, cheap, and reliable at once. RoboTeam measures these limits where earlier benchmarks cannot: long multi-arm jobs, camera images only, and a world that never pauses.
| Comments: | 17 pages, 15 figures, 10 tables, including appendix |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.22138 [cs.RO] |
| (or arXiv:2611.22138v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2611.22138 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Changxuan Fan [view email]
[v1]
Tue, 17 Nov 2026 03:24:06 UTC (4,291 KB)
来源:arXiv cs.AI · arxiv.org