FreqPolicy: Frequency Autoregressive Visuomotor Policy with Continuous Tokens

Yiming Zhong, Yumeng Liu, Chuyang Xiao, Zemin Yang, Youzhuo Wang, Yufei Zhu, Ye Shi, Yujing Sun, Xinge Zhu, Yuexin Ma · 2025

Read FreqPolicy: Frequency Autoregressive Visuomotor Policy with Continuous Tokens by Yiming Zhong, Yumeng Liu, Chuyang Xiao, Zemin Yang, Youzhuo Wang, Yufei Zhu, Ye Shi, Yujing Sun, Xinge Zhu, Yuexin Ma (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-2506-01583.txt.

Description

Yiming Zhong and co-authors teach robot arms by splitting moves into wave bands. The work dates to 2025. A good arm policy must act with care while it runs fast. Past forms of move coding and net design held scores back.

Low wave bands hold broad moves, while high bands hold fine detail. Tasks of distinct weight need distinct care in each band. The new method builds the bands step by step, from broad to fine. It also codes moves as smooth flows, not sharp steps. Tests on bench tasks in flat and full space beat past methods in score and speed.

Catalog record

CREATORYiming Zhong, Yumeng Liu, Chuyang Xiao, Zemin Yang, Youzhuo Wang, Yufei Zhu, Ye Shi, Yujing Sun, Xinge Zhu, Yuexin Ma
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-01583

Browse the full library · About the Tower of Babel