1 paper · 1 filter
Yewon Han, Jisu Jeon, Suhyun Choi +5
Egocentric multimodal assistants see through the user's eyes and guide what the user does next, so their outputs directly shape physical actions, including those that affect bystan…