AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

EgoMotion: Hierarchical Vision-Language Learning and Diffusion for Egocentric Motion Generation

Ruibing Hou, Mingyue Zhou, Yuwei Gui, Mingshuang Luo, Bingpeng Ma, Hong Chang, Shiguang Shan, Xilin Chen

TPAMI 2026 · Vision-language and multimodal › Robot action and navigation · approach: Embodied multimodal LLM agents · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace VQ-VAE with RVQ-VAE (ablation)autoregressive generator · Nymeria Ego-VL motion generationFID: 0.051 -> 0.028Improved
replace VQ-VAE with RVQ-VAE (ablation)autoregressive generator · Nymeria Ego-VL motion generationJerk: 0.0041 -> 0.0023Improved
replace VQ-VAE with RVQ-VAE (ablation)masked generator · Nymeria Ego-VL motion generationR@Top1: 62.7 -> 69.1Improved
use diffusion instead of autoregressive generator (ablation)EgoMotion · Nymeria Ego-VL motion generationFID: 0.028 -> 0.015Improved
use diffusion instead of masked generator (ablation)EgoMotion · Nymeria Ego-VL motion generationFID: 0.031 -> 0.015Improved
move diffusion from raw space to VAE latent space (ablation)EgoMotion · Nymeria Ego-VL motion generationFoot Sliding: 1.56 -> 1.05Improved
move diffusion from raw space to VAE latent space (ablation)EgoMotion · Nymeria Ego-VL motion generationInference time (ms): 295 -> 226Improved
move diffusion from raw space to VAE latent space (ablation)EgoMotion · Nymeria Ego-VL motion generationFID: 0.015 -> 0.018Hurt
two-stage training instead of freezing the VLM (ablation)EgoMotion · Nymeria Ego-VL motion generationFID: 0.030 -> 0.015Improved
two-stage training instead of joint tuning (ablation)EgoMotion · Nymeria Ego-VL motion generationFID: 0.027 -> 0.015Improved
two-stage training instead of joint tuning (ablation)EgoMotion · Nymeria Ego-VL motion generationTraining time (h): 202.6 -> 112.4Improved
two-stage training instead of joint tuning (ablation)EgoMotion · Nymeria Ego-VL motion generationR@Top1: 71.9 -> 69.8Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy