Research map / Vision-language and multimodal / Robot action and navigation
GeoPredict: Leveraging Predictive Kinematics and 3D Gaussian Geometry for Precise VLA Manipulation
Jingjing Qian, Boyao Han, Chen Shi, Lei Xiao, Long Yang, Shaoshuai Shi, Li Jiang
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 history Track Encoder (ablation) | pi0 · RoboCasa Human-50 | success rate (%): 42.3 -> 44.8 | Improved |
| Add Future Track Query with L_track (ablation) | GeoPredict · RoboCasa Human-50 | success rate (%): 44.8 -> 47.2 | Improved |
| Add depth supervision on initial Gaussians (ablation) | GeoPredict · RoboCasa Human-50 | success rate (%): 47.2 -> 49.4 | Improved |
| Add track-guided Gaussian refinement (ablation) | GeoPredict · RoboCasa Human-50 | success rate (%): 50.5 -> 52.4 | Improved |
| Render color in addition to depth (ablation) | GeoPredict · RoboCasa Human-50 | success rate (%): 49.4 -> 49.2 | No clear effect |
| Double global Gaussian primitives (N_G 4 to 8) (ablation) | GeoPredict · RoboCasa Human-50 | success rate (%) and training time: 49.4 -> 51.4; 12.0 -> 19.1 h/epoch | Mixed |
| Use track-guided refinement with N'_G=64 instead of global N_G=8 (ablation) | GeoPredict · RoboCasa Human-50 | success rate (%): 51.4 -> 52.4 | Improved |
| Add GeoPredict predictive modules to pi0 | pi0 · RoboCasa Human-50 | success rate (%): 42.3 -> 52.4 | Improved |
| Add GeoPredict predictive modules to pi0 | pi0 · LIBERO-Long | success rate (%): 87.6 -> 94.0 | Improved |
| Add GeoPredict predictive modules to pi0 | pi0 · LIBERO-Goal | success rate (%): 94.2 -> 95.7 | No clear effect |
| Add GeoPredict predictive modules to pi0 | pi0 · real-world geometry generalization | success rate (%): 50.0 -> 95.0 | Improved |
| Add GeoPredict predictive modules to pi0 | pi0 · real-world visual robustness | success rate (%): 35.0 -> 90.0 | Improved |
Closest papers by meaning
- Localizing, Structuring, and Rendering: Bridging 3D and 2D Vision-Language-Action Models for Robotic Manipulation (CVPR 2026 · similarity 0.90)
- OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints (CVPR 2025 · similarity 0.90)
- BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language Models (NeurIPS 2025 · similarity 0.89)
- DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 2025 · similarity 0.89)
- Beyond World-Frame Action Heads: Motion-Centric Action Frames for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
- Beyond Data Scaling: Representation-Centric Pre-training for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
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.
