arXiv Robotics — research abstracts· Bingao Chen, Lihan Zha, C. Karen Liu, Guy Tevet·· 11 hours agoEditorial score65
FoveaVLA: Foveation Actions Replace the Wrist Camera in VLAs
FoveaVLA: Foveation Actions Replace the Wrist Camera in VLAs
Summary
This paper presents FoveaVLA, a Vision-Language-Action model that replaces the wrist camera with learned foveation actions to improve visual perception. By predicting interaction regions for observation, FoveaVLA achieves higher success rates in real-world tasks compared to traditional wrist-camera setups.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
Source：arXiv Robotics — research abstracts · arxiv.org