AI Paper Atlas

Research map / Video / Video generation

ORV: 4D Occupancy-centric Robot Video Generation

Xiuyu Yang, Bohan Li, Shaocong Xu, Nan Wang, Chongjie Ye, Zhaoxi Chen, Minghan Qin, Yikang Ding, Zheng Zhu, Xin Jin, Hang Zhao, Hao Zhao

CVPR 2026 · Video › Video generation · approach: Trajectory-guided motion control · 16 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
add occupancy-derived visual conditioning to action conditioning (ablation)ORV · BridgeData V2FVD: 17.682 -> 16.525Improved
use Action Expert AdaLN instead of Text Expert for action injection (ablation)ORV · BridgeData V2FVD: 23.586 -> 17.682Improved
add chunk-level temporal action encoding (ablation)ORV · VP2 visual planningsuccess (%): 70.6 -> 74.7Improved
inject occupancy conditioning into input noise rather than ControlNet-style layers (ablation)ORV · BridgeData V2FVD: 20.069 -> 16.525Improved
add pixel-level fine depth conditions instead of coarse (ablation)ORV · BridgeData V2FVD: 18.548 -> 14.321Improved
start from pretrained CogVideoX-2B rather than from scratch (ablation)ORV · BridgeData V2FVD: 84.831 -> 17.682Improved
train on coarse conditions rather than fine (zero-shot cross-granularity) (ablation)ORV · BridgeData V2FVD: 132.685 -> 24.134Improved
add ORV-synthesized data to finetuning dataRoboVLM · SimplerEnv-WidowXaverage success: 29.8% -> 33.9%Improved
add ORV-synthesized data to finetuning dataSpatialVLA · SimplerEnv-WidowXaverage success: 36.2% -> 38.5%Improved
add multiview visual priors (with vs without) (ablation)ORV-MV · BridgeData V2 (3 views)PSNR: 25.77 / 28.25 (view0)Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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