Research map / Vision-language and multimodal / Robot action and navigation
EvoVLA: Self-Evolving Vision-Language-Action Model
Zeting Liu, ZIDA YANG, Zeyu Zhang, Hao Tang
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 counterfactual hard negatives to SAR (ablation) | OpenVLA-OFT · Discoverse-L avg | success rate (%): 59.0 -> 61.8 | Improved |
| add counterfactual hard negatives to SAR (ablation) | OpenVLA-OFT · Discoverse-L avg | hallucination rate (%): 38.5 -> 31.2 | Improved |
| add temporal smoothing of stage scores (ablation) | OpenVLA-OFT+hard negatives · Discoverse-L avg | success rate (%): 61.8 -> 63.7 | Improved |
| add temporal smoothing of stage scores (ablation) | OpenVLA-OFT+hard negatives · Discoverse-L avg | hallucination rate (%): 31.2 -> 23.4 | Improved |
| add Long-Horizon Memory (ablation) | EvoVLA partial · Discoverse-L avg | success rate (%): 63.7 -> 66.1 | Improved |
| add Long-Horizon Memory (ablation) | EvoVLA partial · Discoverse-L avg | hallucination rate (%): 23.4 -> 19.5 | Improved |
| add pose-based object exploration (ablation) | EvoVLA partial · Discoverse-L avg | success rate (%): 66.1 -> 69.2 | Improved |
| add pose-based object exploration (ablation) | EvoVLA partial · Discoverse-L avg | hallucination rate (%): 19.5 -> 14.8 | Improved |
| EvoVLA modules vs OpenVLA-OFT backbone | EvoVLA · Discoverse-L (sim) | success rate (%): 59.0 -> 69.2 | Improved |
| EvoVLA vs OpenVLA-OFT on real robot | EvoVLA · Real-world 4 tasks avg | success rate (%): 43.6 -> 54.6 | Improved |
| intrinsic weight rho 0.6 vs 0.3 (ablation) | EvoVLA · Discoverse-L avg | success rate (%): 67.1 -> 69.2 | Improved |
| intrinsic weight rho 0.6 vs 0.9 (ablation) | EvoVLA · Discoverse-L avg | success rate (%): 66.8 -> 69.2 | Improved |
Closest papers by meaning
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.88)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.87)
- VLA-Corrector: Lightweight Detect-and-Correct Inference for Adaptive Action Horizon (NeurIPS 2026 · similarity 0.87)
- CofactVLA: Deconfounding Vision-Language-Action Models via Counterfactual Intervention (NeurIPS 2026 · similarity 0.87)
- VLA-REPLICA: A Low-Cost, Reproducible Benchmark for Real-World Evaluation of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.87)
- VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching (NeurIPS 2025 · 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.
