Research map / 3D vision / 3D generation
Zero-Shot Robotic Manipulation via 3D Gaussian Splatting-Enhanced Multimodal Retrieval-Augmented Generation
Zilong Xie, Jingyu Gong, Xin Tan, Zhizhong Zhang, Yuan Xie
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 multimodal retrieval (text + CosSim + IMD) over vision-only input (ablation) | MLLM zero-shot manipulation · PartNet-Mobility Test Unseen | ASR (%): 19.01 -> 42.08 | Improved |
| add IMD to CosSim retrieval (ablation) | RobMRAG retrieval · PartNet-Mobility Test Unseen | ASR (%): 34.62 -> 42.08 | Improved |
| add 3D-Aware Pose Refinement (3DGS) (ablation) | RobMRAG zero-shot · PartNet-Mobility Test Unseen | ASR (%): 42.08 -> 43.53 | Improved |
| add 3D-Aware Pose Refinement (3DGS) (ablation) | RobMRAG zero-shot · PartNet-Mobility Test Seen | ASR (%): 47.34 -> 49.25 | Improved |
| upgrade backbone from Qwen2-VL-7B to Qwen2.5-VL-7B (ablation) | RobMRAG fine-tuned (All) · PartNet-Mobility Test Unseen | ASR (%): 57.54 -> 60.63 | Improved |
| increase retrieval top-n beyond 5 (ablation) | RobMRAG · PartNet-Mobility Test Seen | ASR (%): degrades 0.61-1.48% from peak 62.15% | Hurt |
| RobMRAG zero-shot vs RAM zero-shot | zero-shot manipulation · PartNet-Mobility Test Unseen | ASR (%): 35.77 -> 43.53 | Improved |
| RobMRAG fine-tuned vs ManipLLM | fine-tuned manipulation · PartNet-Mobility Test Unseen (All) | ASR (%): 51.00 -> 57.54 | Improved |
| RobMRAG on tiny-interaction-region objects (Scissors) | RobMRAG · PartNet-Mobility Test Unseen (All) | ASR (%): 17.84 | Hurt |
Closest papers by meaning
- ManipLLM: Embodied Multimodal Large Language Model for Object-Centric Robotic Manipulation (CVPR 2024 · similarity 0.89)
- OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints (CVPR 2025 · similarity 0.88)
- Object-Centric Prompt-Driven Vision-Language-Action Model for Robotic Manipulation (CVPR 2025 · similarity 0.88)
- Exploring the Limits of Vision-Language-Action Manipulation in Cross-task Generalization (NeurIPS 2025 · similarity 0.87)
- GravMAD: Grounded Spatial Value Maps Guided Action Diffusion for Generalized 3D Manipulation (ICLR 2025 · similarity 0.87)
- SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and Object Manipulation (NeurIPS 2025 · similarity 0.86)
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.
