Research map / Video / Video generation
Active Intelligence in Video Avatars via Closed-loop World Modeling
Xuanhua He, Tianyu Yang, Ke Cao, Ruiqi Wu, Cheng Meng, Yong Zhang, Zhuoliang Kang, Xiaoming Wei, Qifeng Chen
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 belief state tracking (ablation) | ORCA · L-IVA Workshop | TSR: 0.67 -> 0.77 | Improved |
| add Reflect stage (closed-loop verification) (ablation) | ORCA · L-IVA Workshop | BWS (%): -20.0% -> 26.7% | Improved |
| add Reflect stage (ablation) | ORCA · L-IVA Workshop | TSR: 0.72 -> 0.77 | Improved |
| add System 1 action grounding (ablation) | ORCA · L-IVA Workshop | TSR: 0.74 -> 0.77 | Improved |
| add System 1 action grounding (ablation) | ORCA · L-IVA Workshop | BWS (%): -6.72% -> 26.7% | Improved |
| add belief state tracking (ablation) | ORCA · L-IVA Workshop | BWS (%): 0.00% -> 26.7% | Mixed |
| closed-loop OTAR vs open-loop planning | ORCA vs Open-Loop Planner · L-IVA average | Task Success Rate (%): 62.3 -> 71.0 | Improved |
| closed-loop OTAR vs open-loop planning | ORCA vs Open-Loop Planner · L-IVA Garden | Task Success Rate (%): 46.2 -> 81.5 | Improved |
| closed-loop OTAR vs open-loop planning | ORCA vs Open-Loop Planner · L-IVA Livestream | Task Success Rate (%): 72.3 -> 58.4 | Hurt |
| add reflection and dual-system over VAGEN-style CoT | ORCA · L-IVA average | Task Success Rate (%): 61.2 -> 71.0 | Improved |
| ORCA versus baselines on aesthetics | ORCA · L-IVA average | Aesthetics: 0.59 vs 0.58 | No clear effect |
Closest papers by meaning
- WorldSimBench: Towards Video Generation Models as World Simulators (ICML 2025 · similarity 0.85)
- ThinkAct: Vision-Language-Action Reasoning via Reinforced Visual Latent Planning (NeurIPS 2025 · similarity 0.85)
- WorldReasonBench: Human-Aligned Stress Testing of Video Generators as Future World-State Predictors (NeurIPS 2026 · similarity 0.85)
- PhysVLM-AVR: Active Visual Reasoning for Multimodal Large Language Models in Physical Environments (NeurIPS 2025 · similarity 0.85)
- 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models (CVPR 2026 · similarity 0.84)
- DriveDreamer-Policy: A Geometry-Grounded World-Action Model for Unified Generation and Planning (NeurIPS 2026 · similarity 0.84)
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.
