Summary of Combo: Compositional World Models For Embodied Multi-agent Cooperation, by Hongxin Zhang et al.
COMBO: Compositional World Models for Embodied Multi-Agent Cooperation
by Hongxin Zhang, Zeyuan Wang, Qiushi Lyu, Zheyuan Zhang, Sunli Chen, Tianmin Shu, Behzad Dariush, Kwonjoon Lee, Yilun Du, Chuang Gan
First submitted to arxiv on: 16 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper investigates embodied multi-agent cooperation, where decentralized agents must cooperate given only egocentric views of the world. To address partial observability, generative models are trained to estimate the overall world state from partial egocentric observations. A compositional world model is proposed for multi-agent cooperation by factorizing joint actions and generating videos conditioned on the world state. The approach integrates Vision Language Models to infer other agents’ actions and facilitates online cooperative planning via tree search. Three benchmarks with 2-4 agents are used to evaluate the methods, showing effective cooperation across various tasks and agent numbers. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper explores how robots can work together without a shared view of the world. It’s like trying to plan a game of soccer when you’re just watching your own teammate from your own perspective! To solve this problem, the researchers developed a special kind of model that can imagine what the whole world looks like based on what each robot sees. This allows them to make plans and work together more efficiently. The results show that their approach works well for different numbers of robots and tasks. |
