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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Add VGGT-supervised cross-view geometric consistency loss (L_cgc) (ablation) | single-stream RVT-2 · RLBench | avg success (%): 77.5 -> 80.1 | Improved |
| Add L_cgc (ablation) | dual-stream model · RLBench | avg success (%): 79.5 -> 81.0 | Improved |
| Add dynamic-view stream (with L_cgc) (ablation) | static-view model · RLBench | avg success (%): 80.1 -> 81.0 | Improved |
| Freeze position-aware pretrained gaze model instead of end-to-end training (ablation) | dual-stream model · RLBench | avg success (%): 77.6 -> 79.5 | Improved |
| Add gaze-model heatmaps to dynamic-view features (ablation) | dual-stream model · RLBench | avg success (%): 73.3 -> 79.5 | Improved |
| Add L_cgc (ablation) | single-stream RVT-2 · COLOSSEUM | avg success (%): 60.5 -> 63.8 | Improved |
| Add dynamic-view stream (ablation) | static-view model · COLOSSEUM | avg success (%): 63.8 -> 68.7 | Improved |
| Add L_cgc (ablation) | dual-stream model · COLOSSEUM | avg success (%): 68.7 -> 69.9 | Improved |
| Full Cortical Policy vs RVT-2 | RVT-2 · RLBench | avg success (%): 77.5 -> 81.0 | Improved |
| Full Cortical Policy vs RVT-2 | RVT-2 · COLOSSEUM | avg success (%): 60.5 -> 69.9 | Improved |
| Full model vs variant E on specific COLOSSEUM perturbations (Light Color, Table Color, RO-Size) (ablation) | dual-stream model · COLOSSEUM per-perturbation | success (%): Light Color 78.3 -> 69.3; Table Color 60.0 -> 46.7 | Mixed |
| Add dynamic-view stream | static-view policy · real-world dynamic tasks | success rate: 0% -> 80% | Improved |
Closest papers by meaning
- SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation (ICML 2025 · similarity 0.88)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.87)
- RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete (CVPR 2025 · similarity 0.87)
- CofactVLA: Deconfounding Vision-Language-Action Models via Counterfactual Intervention (NeurIPS 2026 · similarity 0.87)
- Chain-of-Action: Trajectory Autoregressive Modeling for Robotic Manipulation (NeurIPS 2025 · similarity 0.87)
- DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation (NeurIPS 2026 · similarity 0.87)
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.
