Research map / Vision-language and multimodal / Robot action and navigation
World Guidance: World Modeling in Condition Space for Action Generation
Yue Su, Sijin Chen, Haixin Shi, Mingyu Liu, Zhengshen Zhang, Ningyuan Huang, Weiheng Zhong, Zhengbang Zhu, Yuxiao Liu, Xihui Liu
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 SigLIP to DINOv2 future encoders (ablation) | World Guidance · SIMPLER WidowX Success Avg | success rate: 49.0% -> 63.5% | Improved |
| add Wan VAE to DINOv2 future encoders (ablation) | World Guidance · SIMPLER Google Robot Overall Avg | success rate: 69.5% -> 70.9% | Improved |
| swap SigLIP for Wan VAE in the condition encoder (ablation) | World Guidance · SIMPLER WidowX Success Avg | success rate: 63.5% -> 58.4% | Hurt |
| add Q-Former Future Encoder compression in both stages (ablation) | World Guidance (dino-vae) · SIMPLER Google Robot Overall Avg | success rate: 66.7% -> 70.9% | Improved |
| add Q-Former Future Encoder compression in both stages (ablation) | World Guidance (dino-vae) · SIMPLER WidowX Success Avg | success rate: 57.3% -> 58.4% | Improved |
| add condition-prediction co-training in stage two (ablation) | World Guidance · Real-world Fold the Towel (ID) | success rate: 30% -> 60% | Improved |
| add condition-prediction co-training in stage two (ablation) | World Guidance · Real-world Pick and Place (ID) | success rate: 45% -> 60% | Improved |
| add future-condition guidance without co-training vs vanilla VLA (ablation) | World Guidance w/o cotrain · Real-world Pick and Place (ID) | success rate: 45% -> 45% | No clear effect |
| add unannotated human videos for condition supervision only (ablation) | World Guidance · Real-world Pick and Place / Fold the Towel (ID) | success rate: 60% -> 70% (P&P), 60% -> 50% (Fold) | Mixed |
| add annotated human videos in stage one plus unannotated in stage two (ablation) | World Guidance · Real-world Fold the Towel (ID) | success rate: 60% -> 65% | Improved |
| add 120 UMI trajectories in stage two fine-tuning | World Guidance · Real-world Pick and Place | success rate: 60% -> 85% | Improved |
| add UMI data in stage two fine-tuning | World Guidance · Real-world Fold the Towel | success rate: 60% -> 80% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Moto — Moto: Latent Motion Token as the Bridging Language for Learning Robot Manipulation from Videos (ICCV 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
- World–Value–Action Model: Implicit Planning for Vision–Language–Action Systems (NeurIPS 2026 · similarity 0.90)
- 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 0.88)
- From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction (NeurIPS 2025 · similarity 0.88)
- World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetry (NeurIPS 2026 · similarity 0.88)
- DriveDreamer-Policy: A Geometry-Grounded World-Action Model for Unified Generation and Planning (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.
