Collective Behavior Clone with Visual Attention via Neural Interaction Graph Prediction
Kai Li, Zhao Ma, Liang Li, Shiyu Zhao · 2025
Read Collective Behavior Clone with Visual Attention via Neural Interaction Graph Prediction by Kai Li, Zhao Ma, Liang Li, Shiyu Zhao (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-2503-06869.txt.
Description
Kai Li, Zhao Ma, Liang Li, and Shiyu Zhao proposed in 2025 a framework called collective behavioral cloning for learning how a robot swarm interacts and acts. From trajectory data alone, it is hard to know which neighbors each robot reacts to. Their method first learns that hidden interaction graph.
A graph variational autoencoder infers the local interaction graph from swarm paths. Behavioral cloning then learns the control policy from the graph and the paths together. The system was deployed on a real decentralized robot swarm that uses vision. A visual attention network, trained on the learned graph, picks neighbors online. Tests showed better prediction of both the graph and swarm actions than prior methods.
Catalog record
| CREATOR | Kai Li, Zhao Ma, Liang Li, Shiyu Zhao |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-06869