AI Paper Atlas

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

Cortical Policy: A Dual-Stream View Transformer for Robotic Manipulation

Xuening Zhang, Qi Lv, Xiang Deng, Miao Zhang, Xingbo Liu, Liqiang Nie

ICLR 2026 · Vision-language and multimodal › Robot action and navigation · approach: Memory and subgoal-based manipulation · 1 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 VGGT-supervised cross-view geometric consistency loss (L_cgc) (ablation)single-stream RVT-2 · RLBenchavg success (%): 77.5 -> 80.1Improved
Add L_cgc (ablation)dual-stream model · RLBenchavg success (%): 79.5 -> 81.0Improved
Add dynamic-view stream (with L_cgc) (ablation)static-view model · RLBenchavg success (%): 80.1 -> 81.0Improved
Freeze position-aware pretrained gaze model instead of end-to-end training (ablation)dual-stream model · RLBenchavg success (%): 77.6 -> 79.5Improved
Add gaze-model heatmaps to dynamic-view features (ablation)dual-stream model · RLBenchavg success (%): 73.3 -> 79.5Improved
Add L_cgc (ablation)single-stream RVT-2 · COLOSSEUMavg success (%): 60.5 -> 63.8Improved
Add dynamic-view stream (ablation)static-view model · COLOSSEUMavg success (%): 63.8 -> 68.7Improved
Add L_cgc (ablation)dual-stream model · COLOSSEUMavg success (%): 68.7 -> 69.9Improved
Full Cortical Policy vs RVT-2RVT-2 · RLBenchavg success (%): 77.5 -> 81.0Improved
Full Cortical Policy vs RVT-2RVT-2 · COLOSSEUMavg success (%): 60.5 -> 69.9Improved
Full model vs variant E on specific COLOSSEUM perturbations (Light Color, Table Color, RO-Size) (ablation)dual-stream model · COLOSSEUM per-perturbationsuccess (%): Light Color 78.3 -> 69.3; Table Color 60.0 -> 46.7Mixed
Add dynamic-view streamstatic-view policy · real-world dynamic taskssuccess rate: 0% -> 80%Improved

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