Dayou Li

PhD student @ Texas A&M | Robotics

[Dayou Li @EdisonLeeROC]

@GetXBot

[Dayou Li @EdisonLeeROC]

📢 EgoHOI: Egocentric World Model for Photorealistic Hand-Object Interaction Synthesis

EgoHOI integrates physics-informed embeddings to regularize latent dynamics for realistic first-person hand-object interaction rollouts.

Code & Paper(open source): https://t.co/oc9mzRrMq3

[Zhiwen(Aaron) Fan @zhiwen_fan_]

#EgoTL is an egocentric dataset with human think-aloud annotations, step-by-step goals, and metric spatial grounding.

It captures what people plan, say, and do before acting, enabling more faithful study of planning, spatial reasoning, and world modeling.

Current VLMs and world models still struggle with long-horizon egocentric reasoning, but training on EgoTL leads to clear gains.

We release the dataset at https://t.co/yI2gpDu6Gw