← Back to Model Beat
Research·Jul 28·all news from July 28, 2026

MoMo: Dial Motion Mode in Robot Manipulation with Spatiotemporal Action Tokenization

Researchers at Apple have introduced MoMo, a method for training robots to adjust their movement patterns based on the specific requirements of different tasks and environments. By using spatiotemporal action tokenization, the system allows robots to learn flexible behaviors that can be applied across various manipulation scenarios. This approach aims to improve how machines adapt their physical execution to better handle diverse objects and interaction settings.

Covered by 2 sources · 9 articles

  • AApple Machine Learning BlogJul 30
  • AarXiv CS.AIAjay Sridhar, Jensen Gao, Jonathan Yang, Jean Mercat, Suneel Belkhale, Dorsa SadighJul 31
  • AarXiv CS.AIJia LuoJul 30
  • AarXiv CS.AIXiangcheng Zhang, Yilun DuJul 31
  • AarXiv CS.AIAnthony Liang, Pavel Czempin, Matthew M. Hong, Yutai Zhou, Jingzhen Wang, Erdem Biyik, Stephen TuJul 31
  • AarXiv CS.AISami Azirar, Enrico Pallotta, Jan Nogga, J\"urgen Gall, Sven Behnke, Hermann BlumJul 30
  • AarXiv CS.AITim R. Winter, Leonard Kl\"upfel, Ashok M. Sundaram, Werner Friedl, Maximo A. Roa, Freek Stulp, Jo\~ao Silv\'erioJul 29
  • AarXiv CS.AIZuojin Tang, Feifan Luo, Haoyun Liu, Botai Yuan, Dekang Qi, Ronghan Chen, Yandan Yang, Tong Lin, Xinyuan Chang, Mu Xu, Bin Liu, De Ma, Zhiheng MaJul 30
  • AarXiv CS.AIMengqi Zhang, Sahil Khose, Simar Kareer, Yuchen Song, Unnat Jain, Judy HoffmanJul 28

Related stories

ResearchAdvancing responsible AI across EuropeJul 29 · 24 sourcesResearchAI Investment Boom Faces Reality Check From Markets and RegulatorsJul 29 · 6 sourcesResearchAccelerating scientific discovery with ChatGPT for Academic ResearchersJul 29ResearchMemory Efficient Audio Synthesis with Decoupled Temporal Depth Diffusion TransformersJul 28 · 2 sources