AI Paper Atlas

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

4D Visual Pre-training for Robot Learning

Chengkai Hou, Yanjie Ze, Yankai Fu, Zeyu Gao, Songbo Hu, Yue Yu, Shanghang Zhang, Huazhe Xu

ICCV 2025 · Vision-language and multimodal › Robot action and navigation · approach: Memory and subgoal-based manipulation · 3 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
use history frame instead of current frame as pre-training input (ablation)DP3+FVP · UR5 real task PickSquaresuccess count (out of 20 trials): 15/20 -> 20/20Improved
use history frame instead of current frame as pre-training input (ablation)DP3+FVP · UR5 real task ToastBreadsuccess count (out of 20 trials): 13/20 -> 16/20Improved
fine-tune visual encoder instead of freezing it (ablation)DP3+FVP · UR5 real task PlaceBottlesuccess count (out of 20 trials): 9/20 -> 20/20Improved
fine-tune visual encoder instead of freezing it (ablation)DP3+FVP · UR5 real task ToastBreadsuccess count (out of 20 trials): 7/20 -> 16/20Improved
use 1 history frame instead of 4 (ablation)DP3+FVP · UR5 real task PlaceBottlesuccess count (out of 20 trials): 14/20 -> 20/20Improved
use 1 history frame instead of 4 (ablation)DP3+FVP · UR5 real task PickSquaresuccess count (out of 20 trials): 15/20 -> 20/20Improved
pre-train DP3 encoder with FVP instead of 2D R3M encoderDP3 · Real tasks avg (UR5, 6 tasks, Diffusion Policy)success count (out of 20 trials): 12.5/20 -> 16.4/20Improved
pre-train DP3 encoder with FVP instead of MVPDP3 · UR5 real task FlipCupsuccess count (out of 20 trials): 17/20 -> 16/20Hurt
add 3D point cloud input to RDT-1BRDT-1B · RDT-1B generalization tasks avgsuccess count (out of 20 trials): 6/20 -> 8.25/20Improved
add FVP pre-training of the 3D encoder to RDT-1B with point cloudRDT-1B · RDT-1B generalization tasks avgsuccess count (out of 20 trials): 8.25/20 -> 10/20Improved
add FVP pre-training to RDT-1B with point cloudRDT-1B · RDT-1B real task PickSquaresuccess count (out of 20 trials): 14/20 -> 18/20Improved
add FVP pre-training to RDT-1B with point cloudRDT-1B · RDT-1B real task PutBoxsuccess count (out of 20 trials): 9/20 -> 9/20No clear effect

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