Research map / Vision-language and multimodal / Robot action and navigation
Programmatically Grounded, Compositionally Generalizable Robotic Manipulation
Renhao Wang, Jiayuan Mao, Joy Hsu, Hang Zhao, Jiajun Wu, Yang Gao
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 |
|---|---|---|---|
| use program-based modular grounding and action modules instead of CLIPort's monolithic language-conditioned policy | robotic manipulation · packing-color-box, 100 demonstrations | success score: 23.0 -> 72.0 | Improved |
| use program-based modular grounding and action modules instead of CLIPort's monolithic language-conditioned policy | robotic manipulation · packing-location-box, 100 demonstrations | success score: 11.0 -> 62.0 | Improved |
| use program-based modular grounding and action modules instead of CLIPort's monolithic language-conditioned policy | robotic manipulation · separating-location-piles, 100 demonstrations | success score: 48.1 -> 69.2 | Improved |
| use program-based modular grounding and action modules instead of CLIPort's monolithic language-conditioned policy | robotic manipulation · pushing-shapes, 100 demonstrations | success score: 2.7 -> 36.6 | Improved |
| use program-based modular grounding and action modules instead of CLIPort's monolithic language-conditioned policy | robotic manipulation · packing-prepositions | success score: 28.0 -> 45.0 | Improved |
| use program-based modular grounding and action modules instead of CLIPort's monolithic language-conditioned policy | robotic manipulation · packing-nested-prepositions | success score: 17.0 -> 31.0 | Improved |
| replace MaskCLIP grounding maps with ground-truth segmentation maps (ablation) | ProgramPort (single) · packing-shapes, 100 demonstrations | success score: 52.0 -> 97.0 | Improved |
| replace MaskCLIP grounding maps with ground-truth segmentation maps (ablation) | ProgramPort (single) · packing-google-objects-seq, 100 demonstrations | success score: 78.4 -> 95.8 | Improved |
| replace MaskCLIP grounding maps with ground-truth segmentation maps (ablation) | ProgramPort (single) · assembling-kits, 100 demonstrations | success score: 40.6 -> 92.0 | Improved |
| replace MaskCLIP grounding maps with ground-truth segmentation maps (ablation) | ProgramPort (single) · separating-small-piles, 100 demonstrations | success score: 76.8 -> 80.6 | Improved |
| replace MaskCLIP grounding maps with ground-truth segmentation maps (ablation) | ProgramPort (single) · put-blocks-in-bowls, 100 demonstrations | success score: 38.2 -> 97.2 | Improved |
| train on 100 rather than 1000 demonstrations for packing-shapes | ProgramPort versus CLIPort · packing-shapes | success score: +25.0 points | Improved |
Closest papers by meaning
- OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints (CVPR 2025 · similarity 0.91)
- Exploring the Limits of Vision-Language-Action Manipulation in Cross-task Generalization (NeurIPS 2025 · similarity 0.91)
- Object-Centric Prompt-Driven Vision-Language-Action Model for Robotic Manipulation (CVPR 2025 · similarity 0.91)
- Unleashing Large-Scale Video Generative Pre-training for Visual Robot Manipulation (ICLR 2024 · similarity 0.90)
- Vision-Language Foundation Models as Effective Robot Imitators (ICLR 2024 · similarity 0.90)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.90)
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.
