From CheatCode to Cross-Modal Attention: Building MgACT for Contact-Rich Manipulation
中文摘要
MgACT 从 CheatCode 转向跨模态注意力机制，通过多感官融合提升接触密集型机器人的操控能力，解决了视觉信息的局限性。
English Summary
MgACT transitions from CheatCode to cross-modal attention to enhance contact-rich robotic manipulation, overcoming vision-only limitations through multi-sensory integration.
原文节选
The Challenge: When Vision Isn’t Enough Continue reading on Medium »