Research map / Vision-language and multimodal / Robot action and navigation
Beyond Dense Futures: World Models as Structured Planners for Robotic Manipulation
Minghao Jin, Mozheng Liao, Mingfei Han, Zhihui Li, Xiaojun Chang
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 |
|---|---|---|---|
| increase history context from 2 to 4 frames (ablation) | StructVLA · SimplerEnv-WidowX | avg success rate: 70.8% -> 75.0% | Improved |
| increase history context from 2 to 4 frames (ablation) | UniVLA · SimplerEnv-WidowX | avg success rate: 69.8% -> 66.7% | Hurt |
| predict sparse structured frames instead of dense futures (StructVLA vs UniVLA) | UniVLA world model · SimplerEnv-WidowX | avg success rate: 69.8% -> 75.0% | Improved |
| predict structured frames | UniVLA world model · LIBERO-Long | success rate: 89.6% -> 92.2% | Improved |
| predict structured frames | UniVLA world model · LIBERO | avg success rate: 93.7% -> 94.8% avg; Spatial 96.4% -> 95.4% | Mixed |
| predict structured frames | UniVLA · Real-world tidy-up | successes out of 10: 4/10 -> 8/10 | Improved |
| predict structured frames | UniVLA · Real-world OOD pick-and-place | success rate: 30.0% -> 80.0% | Improved |
| predict structured frames | UniVLA · Real-world human intervention | success rate: 40.0% -> 80.0% | Improved |
| predict structured frames | UniVLA · SimplerEnv-WidowX training efficiency | steps to peak: 12k -> 10k steps; 64.8% -> 66.7% | Improved |
| predict structured frames | UniVLA · SimplerEnv-WidowX (Put Eggplant) | success rate: 95.8% -> 91.7% | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DreamVLA — DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 2025) · better on 2 of 3 shared results
- VideoVLA — VideoVLA: Video Generators Can Be Generalizable Robot Manipulators (NeurIPS 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 2025 · similarity 0.92)
- FLIP: Flow-Centric Generative Planning as General-Purpose Manipulation World Model (ICLR 2025 · similarity 0.90)
- VideoVLA: Video Generators Can Be Generalizable Robot Manipulators (NeurIPS 2025 · similarity 0.90)
- Beyond World-Frame Action Heads: Motion-Centric Action Frames for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- VLA-OS: Structuring and Dissecting Planning Representations and Paradigms in Vision-Language-Action Models (NeurIPS 2025 · similarity 0.90)
- RoboCerebra: A Large-scale Benchmark for Long-horizon Robotic Manipulation Evaluation (NeurIPS 2025 · 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.
