AI Paper Atlas

Research map / Video / Video generation

Geometry-aware 4D Video Generation for Robot Manipulation

Zeyi Liu, Shuang Li, Eric Cousineau, Siyuan Feng, Benjamin Burchfiel, Shuran Song

ICLR 2026 · Video › Video generation · approach: World-model video generation · 21 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
remove multi-view cross-attention in the U-Net decoder (ablation)OURS · four 4D generation settingscross-view mIoU: 0.70 -> 0.41; 0.69 -> 0.44; 0.64 -> 0.26; 0.56 -> 0.32Hurt
remove multi-view cross-attention in the U-Net decoder (ablation)OURS · four 4D generation settingsAbsRel-m: 0.11 -> 0.31; 0.07 -> 0.33; 0.07 -> 0.49; 0.20 -> 0.34Hurt
remove multi-view cross-attention in the U-Net decoder (ablation)OURS · four 4D generation settingsFVD-n: 411.20 -> 497.43; 377.68 -> 451.54; 490.88 -> 597.05; 384.26 -> 694.23Hurt
add multi-view cross-attention over RGB features to SVD (ablation)SVD · PutSpatulaOnTable, PlaceAppleFromBowlIntoBin, real worldFVD-n: 370.92 -> 536.02 (task 2); 659.52 -> 812.94 (task 3); 605.10 -> 1151.05 (real world)Hurt
add multi-view cross-attention over RGB features to SVD (ablation)SVD · StoreCerealBoxUnderShelfFVD-n: 977.06 -> 941.73Improved
use generated RGB-D 4D video with pose tracking instead of baseline policiesOURS · three simulation manipulation taskstask success rate: 0.12 / 0.12 / 0.25 -> 0.64 averageImproved
use global point clouds as input (DP3 vs DP)diffusion policy · PutSpatulaOnTable, PlaceAppleFromBowlIntoBintask success rate: 0.27 -> 0.27 (task 2); 0.00 -> 0.00 (task 3)No clear effect
use point clouds as input (DP3 vs DP)diffusion policy · StoreCerealBoxUnderShelftask success rate: 0.10 -> 0.23Improved
add a third input view at inference without retraining (ablation)OURS · StoreCerealBoxUnderShelfFVD / mIoU: FVD 515.53 -> 566.86 -> 662.83; mIoU 0.62 -> 0.54Hurt
use OURS instead of SVD in inference costOURS · inference latencyinference time: 13.4 s -> 30.0 sHurt
use 4D Gaussian baseline instead of OURS4D Gaussian · novel-view multi-view generationcross-view mIoU: 0.70 -> 0.39 (task 1); 0.56 -> 0.00 (real)Hurt

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