Nan-Jie Yao

This framework surveys manipulation, navigation, locomotion, autonomous driving, and general embodied learning, tracing technical progressions, clarifying capability requirements, and examining datasets, benchmarks, and evaluation protocols.

EvolveNav achieves a 10.1% boost in success rates for zero-shot navigation by enabling agents to learn and adapt in real-time from their experiences.

Streaming 3D reconstruction over thousands of frames is now possible with constant memory and linear time, thanks to a novel attention mechanism that learns to propagate geometric evidence across multiple timescales.

Achieve more realistic 3D human avatars from single images by jointly optimizing geometry and texture, even when training data is scarce.