2026 IEEE International Conference on Robotics and Automation (ICRA) / 高效注意力

RetoVLA: Reusing Register Tokens for Spatial Reasoning in Vision-Language-Action Models

作者：Jiyeon Koo, Taewan Cho, Hyunjoon Kang, Eunseom Pyo, Tae Gyun Oh, Taeryang Kim, Andrew Jaeyong Choi · 单位：School of Computing, Gachon University · 会议/期刊：2026 IEEE International Conference on Robotics and Automation (ICRA) · 方向：高效注意力 · 发布日期：2025-09-25

视觉语言动作高效注意力高效推理高效架构模型压缩
新标签打开 AlphaXiv新标签打开 Video
RetoVLA: Reusing Register Tokens for Spatial Reasoning in Vision-Language-Action Models figure
AlphaXiv 中文概览（可滚动查看）