Research map / Vision-language and multimodal / Robot action and navigation
EvoWorld: A World-Model-Centric Framework for Continuous Self-Evolution of Modular Embodied Skills
boshi zhang, Sen Cui, Baohua Yin, Youyi Kou, Junyu Wu, Zuo Pu, TAO XUE, Zhikang Chen, Shanshan Wei, Min Zhang, Miao Liu, Changshui Zhang and 1 more
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 EvoW routing over the monolithic base policy (ablation) | GR00T-N1.6 · Skill families S1-S5 average | success rate: 0.256 -> 0.468 | Improved |
| add EvoW routing over the monolithic base policy (ablation) | PI-0.5 · Skill families S1-S5 average | success rate: 0.232 -> 0.412 | Improved |
| add EvoW routing over the monolithic base policy (ablation) | PI-0 · Skill families S1-S5 average | success rate: 0.284 -> 0.460 | Improved |
| add EvoW routing over the monolithic base policy (ablation) | PI-0-fast · Skill families S1-S5 average | success rate: 0.264 -> 0.452 | Improved |
| add EvoW routing and diagnosis-driven updates on a real Franka arm | GR00T-N1.6 · Real-robot Franka (5 tasks) | success rate (%): 29% -> 70% | Improved |
| add routing, averaged over backbones (ablation) | EvoW · Skill families S1-S5 average | success rate: 0.259 -> 0.448 | Improved |
| use EvoW task generation versus RoboGen | EvoW task set · Task diversity | Self-BLEU (lower better): 0.284 -> 0.281 | Improved |
| use EvoW task generation versus RoboGen | EvoW task set · Task diversity | SentenceBERT similarity (lower better): 0.165 -> 0.312 | Hurt |
| use EvoW task generation versus RoboGen | EvoW task set · Task diversity | ViT similarity (lower better): 0.193 -> 0.337 | Hurt |
Closest papers by meaning
- EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models (NeurIPS 2025 · similarity 0.88)
- Not All Features Are Created Equal: A Mechanistic Study of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.87)
- EVLP: Learning Unified Embodied Vision-Language Planner with Reinforced Supervised Fine-Tuning (ICLR 2026 · similarity 0.87)
- GenRL: Multimodal-foundation world models for generalization in embodied agents (NeurIPS 2024 · similarity 0.87)
- DriveDreamer-Policy: A Geometry-Grounded World-Action Model for Unified Generation and Planning (NeurIPS 2026 · similarity 0.87)
- Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models (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.
