Action Chunk Scheduling for Batched Robot Policy Serving
Serving one robot policy to a fleet from a remote GPU, posed as scheduling. Naive batching fails when robots consume action chunks at different rates; their scheduler reports up to 18% higher real-world throughput.把从远程 GPU 为多台机器人提供策略推理建模为调度问题。当各机器人消耗动作块的速度不同时，简单批处理失效；所提调度算法在真机上报告吞吐量最高提升 18%。
BibTeX
@misc{bansal2026action,
title = {{Action Chunk Scheduling for Batched Robot Policy Serving}},
author = {Rohan Bansal and David He and Nadun Ranawaka Arachchige and Zhenyang Chen and Soobum Kim and Kexin Rong and Danfei Xu},
year = {2026},
eprint = {2608.00337},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2608.00337},
}